THE ROLE OF TURBULENCE IN BROADCAST SPAWNING AND LARVAL
SETTLEMENT IN FRESHWATER DREISSENID MUSSELS
A Thesis
Presented to
The Faculty of Graduate Studies
of
The University of Guelph
by
NOEL PETER QUINN
In partial fulfilment of requirements
for the degree of
Doctor of Philosophy
December, 2009
© Noel P. Quinn, 2009 Library and Archives Bibliotheque et 1*1 Canada Archives Canada Published Heritage Direction du Branch Patrimoine de ('edition
395 Wellington Street 395, rue Wellington Ottawa ON K1A 0N4 OttawaONK1A0N4 Canada Canada
Your file Voire reference ISBN: 978-0-494-58276-3 Our file Notre reference ISBN: 978-0-494-58276-3
NOTICE: AVIS:
The author has granted a non L'auteur a accorde une licence non exclusive exclusive license allowing Library and permettant a la Bibliotheque et Archives Archives Canada to reproduce, Canada de reproduire, publier, archiver, publish, archive, preserve, conserve, sauvegarder, conserver, transmettre au public communicate to the public by par telecommunication ou par I'lnternet, preter, telecommunication or on the Internet, distribuer et vendre des theses partout dans le loan, distribute and sell theses monde, a des fins commerciales ou autres, sur worldwide, for commercial or non support microforme, papier, electronique et/ou commercial purposes, in microform, autres formats. paper, electronic and/or any other formats.
The author retains copyright L'auteur conserve la propriete du droit d'auteur ownership and moral rights in this et des droits moraux qui protege cette these. Ni thesis. Neither the thesis nor la these ni des extraits substantiels de celle-ci substantial extracts from it may be ne doivent etre imprimes ou autrement printed or otherwise reproduced reproduits sans son autorisation. without the author's permission.
In compliance with the Canadian Conformement a la loi canadienne sur la Privacy Act some supporting forms protection de la vie privee, quelques may have been removed from this formulaires secondaires ont ete enleves de thesis. cette these.
While these forms may be included Bien que ces formulaires aient inclus dans in the document page count, their la pagination, il n'y aura aucun contenu removal does not represent any loss manquant. of content from the thesis.
1*1 Canada ABSTRACT
THE ROLE OF TURBULENCE IN BROADCAST SPAWNING AND LARVAL SETTLEMENT IN FRESHWATER DREISSENID MUSSELS
Noel Peter Quinn Advisor: University of Guelph, 2009 Professor J. D. Ackerman
The role of turbulence has been shown theoretically to influence external fertilization and larval settlement/transport in benthic invertebrates. This is especially true for turbulence generated by bottom roughness in the near-bed region of lakes. This thesis examined the role of bottom roughness created by the presence of freshwater mussels (Dreissena polymorpha and D. bugensis) by addressing three objectives: (1) how fertilization success is influenced by water velocity and near-bed turbulence generated by bottom roughness in a laboratory flow chamber and in the field (Evans Point, Lake Erie); (2) how larval transport and settlement is influenced by this near-bed turbulence in laboratory flow chamber and field (Lake Erie) experiments; and (3) how gamete and larval transport are influenced by the ratio of roughness spacing (X) to roughness height (k) using computational fluid dynamic (CFD) modeling. Results indicated that dreissenid mussels are sperm limited, but the extent to which sperm dilution affects them is lower than what has been reported for other broadcast spawners. The nature of near-bed turbulence was determined through digital particle imaging velocimetry (PIV) measurements in the laboratory and acoustic Doppler velocimetry (ADV) measurements in the field. Laboratory results indicated the importance of bottom roughness on the flow regime, notably skimming flow was observed in the high density mussel configuration and wake interference was observed in the mussel patch configuration. Fertilization success and larval settlement in the field was highest under the mussel patch configuration, which was positively associated with turbulent ejections.
Results from the CFD modeling, which incorporated a released scalar as a proxy for gamete and larvae, accurately predicted the flow regimes classified using the ratio of X/k (e.g., < 8 for skimming flow and ~8 for wake interference flow), but this varied with the geometry of the modeled roughness elements. These results indicate that the spatial configuration of bottom roughness, including mussels, determines the flow regime (i.e., skimming vs. wake interference flow), which in turn affects fertilization success and larval transport/settlement in benthic species. ACKNOWLEDGEMENTS
I would like to thank Kelly McNichols, Brendan Hunt, Sarah Glover, Peter
Blouw, Rob Schindler, and Neil Menezes for all the assistance in the laboratory and field.
I would also like to thank Greg Nishihara and Patrick Ragaz for their assistance with the flume and,PIV setup.
Throughout my PhD, I have had invaluable feedback from my PhD committee, so
I would like to thank Dr. David Barton, Dr. Ray Kostaschuk, and Dr. KevinJVlcCann for helping me over these last few years. I would also like to thank Dr. Pete Jumars for acting as my External Reviewer, and providing some invaluable insight into my thesis.
My advisor, Dr. Joe Ackerman, has made me a better researcher, writer, and scientist, and without his guidance and friendship I would not be where I am today.
I would to thank my parents for all their support over the last few years, and finally I would like to thank my wife, Tara, for being there for me when I needed someone to bounce ideas off of, when I needed guidance, or even when I needed help with fieldwork or in the laboratory. I could not have done this without her.
This research was supported in part by funding from the University of Guelph and the Natural Sciences and Engineering Research Council of Canada to J. D. A. TABLE OF CONTENTS
General Introduction 1
Chapter 1: Biological and ecological mechanisms for overcoming sperm limitation in a freshwater mussel
Abstract 22
Introduction 23
Methods 25
Results 32
Discussion 37
Chapter 2: The effect of near-bed turbulence on sperm dilution and fertilization success of broadcast spawning bivalves.
Abstract 59
Introduction 60
Methods 62
Results 69
Discussion 76
Chapter 3: Effects of near-bed turbulence on settlement and resuspension of freshwater mussel larvae
Abstract 99
Introduction 100
Methods 102
Results 108
u Discussion .1.14
Chapter 4: The role of bottom roughness parameters on the transport of sperm and larvae of benthic organisms
Abstract 133
Introduction 134
Methods 136
Results 140
Discussion 143
General Conclusions 162
111 LIST OF TABLES
Table II. Factors influencing fertilization success for broadcast spawners (from Okubo et al. 2001). 9
Table 3.1. Significant associations with R values greater than 0.100 for laboratory and field data using linear regression analysis. 124 LIST OF FIGURES
Figure II. Forces acting on a static body. U represents direction of flow. 3
Figure 12. Vertical profile of a benthic boundary layer (from Nowell and Jumars 1984). 6
Figure 13. An example of an acoustic Doppler velocimeter velocity time series (taken 30 cm above the bed at Evans Point, Lake Erie on August 9, 2008) indicating an example of a fluctuation in the stream-wise velocity component, U. 6
Figure 14. Diagram of the u'w' quadrant plane (based on Cellino and Lemmin 2004). 6
Figure 15. Life history stages of Dreissena polymorpa (from Ackerman etal. 1994). : 13
Figure 1.1. (A) Results from the polyspermy study on dreissenid fertilization. Freshly spawned eggs and sperm were mixed and monitored over a 24 h period, with the proportion of successfully developing zygotes/embryos recorded at each time interval. (B) The stages of zygote development at; 0 h - unfertilized egg; 0.5 - fertilization envelope; 1.0 h - 2-cell stage; 2.0 h - 4-cell stage; 3.0 h - 8-cell stage; and at 24 h - trochophore larval stage. 50
Figure 1.2. Fertilization success as a function of sperm concentration for Dreissena polymorpha and Dreissena bugensis. Fertilization success increases relative to the increase in sperm concentration. Plots are means ± SE, N D. polymorpha = 8, N D.bugensts= 4. There was no significant difference between species. 51
Figure 1.3. Fertilization success as a function of egg concentration for Dreissena polymorpha and Dreissena bugensis. There was no significant relationship between fertilization success and egg concentration for either species. 52
Figure 1.4. Fertilization success in Dreissena polymorpha and Dreissena bugensis as a function of time post spawning for five sperm concentrations. Regardless of concentration, there is a decrease in fertilization success as sperm age. Values are means ± oil. IN D .polymorpha ~ O, FN o. bugensis ~ J- ?.J
Figure 1.5. Results from COMSOL modeling for the mussel cluster scenario (A) and coral head scenario (B). Flow is left to right, with lines representing velocity streamlines and areas of recirculation are rotating in a clockwise direction. Areas of recirculation are indicated by the presence of vortices, examples of which can be seen behind the individual mussel and reef elements. Scalar (1 mol ml"1) which was released from the downstream slope of the first mussel cluster (C) or coral head (D) is retained in the recirculation zone or transported downstream. Note the different spatial scales for the two scenarios. White line in panel D represents an example of a transect used for scalar concentration measurements. 54
v Figure 1.6. The effect of modeled roughness element height (HRei = HR/Z) on the upstream retention vs. downstream transport of scalar to the next element (i.e., downstream of the second roughness element). The dotted line indicates the point at which 50% of released scalar switches from downstream transport to upstream retention, Note that the mussel cluster and coral head element fit the curve as well. Values are means ± SE. 55
Figure 1.7. The influence of bathymetry on modeled fertilization success downstream. The measured fertilization rate from Fig. 1 and the modeled scalar concentration downstream was used to calculate fertilization success. The vertical line indicates the location of the second roughness element. Distance downstream was normalized to the element diameter, where the point of scalar release was equal to 1 (i.e., from the first cluster/coral head). Values are means ± SE. 56
Figure 1.8. A comparison of fertilization success as function of sperm concentration for seven broadcast spawning species. The two dreissenid species, represented by the circles, display much higher fertilization success than the other five species for sperm concentrations less than 103 sperm ml"1. The three species Montipora digitata, Haliotis laevigata, and Platygyra sinensis are affected by polyspermy at high sperm 7 1 concentrations, indicated by a drop in fertilization at 10 sperm ml" . Data from Pennington (1985); Levitan et al. (1991); Oliver and Babcock (1992); and Babcock and Kessing (1999). Values are means ± SE. Differences are significant (%52 = 11.1, P < 0.01). 57 Figure 1.9. A comparison of sperm half-life as function of sperm concentration for nine broadcast spawning species. The sperm half-life is defined as the time it takes to reach 50% of the maximal fertilization rate achieved at any given initial sperm concentration. The two dreissenid species, represented by the circles, display a sperm half-life curve with a low slope indicating a uniform sperm decay rate regardless of sperm concentration, suggesting more potent sperm at lower concentrations. Data from Levitan (1993); Benzie and Dixon (1994); Andre and Lindegarth (1995); Babcock and Kessing (1999); and Baker and Tyler (2001). 58
Fig. 2.1. Examples of turbulence generating conditions: mussel shells attached to the flume bottom in (A) low mussel density (860 mussels m"2) configuration over the entire 50 cm sampling area; (B) high mussel density (1700 mussels m"2) over the entire 50 cm sampling area; (C) high mussel density clusters (1700 mussels m"2) 10 cm in length and placed 10 cm apart from each other; (D) 1.5 cm radius half cylinder (indicated by white arrow) placed upstream of sampling area; and (E) 1 cm2 grid (indicated by white arrow) placed upstream of sampling area. Each white square on flume bottom is 1 cm2. Direction of flow is bottom to top in all cases. 88
Fig. 2.2. Location of Lake Erie field site and individual sampling configurations (1-4). Location of study area is indicated by the black circle in the insert. Image of Lake Erie from the National Geophysical Data Center, National Oceanic and Atmospheric Administration, U.S. Department of Commerce, http://www.ngdc.noaa.gov/. 89
VI Fig. 2.3. Laboratory fertilization success as a function of distance downstream for Dreissena polymorpha at five different flow chamber velocities. Plots are means ± SE, N = 6. 90
Fig. 2.4. The relationship between fertilization success and sperm dilution, determined through a linear regression of the log-transformed sperm concentrations on fertilization success using data from all experiments. N = 108. 91
Fig. 2.5. Laboratory fertilization success and ejection (Q2) and sweep (Q4) events as a function of distance downstream for Dreissena polymorpha at seven different turbulence configurations and two flow chamber velocities. Fertilization plots are means ± SE, N = 3, and quadrant plots are based on 150 PIV frame-averaged means. 92
Fig. 2.6. PIV vector plots indicating direction and magnitude of the overall spatial flow fields for: (A) a mussel cluster; (B) an open patch between two mussel clusters; (C) a half cylinder 5 cm upstream (seen on left hand side of panel); and (D) a wire grid placed at point 0 (indicated by blue vertical bar on left hand side of panel). Y-axis represents height in cm, and x-axis downstream distance in cm. The length of the vector represents higher relative velocity. 93
Fig. 2.7. (A) Dimensionless velocity measurements for five downstream locations based on local velocities (u) measured by particle image velocimetry (PIV) normalized to the free-stream velocity (U). Black symbols are 5 cm s" and white symbols are 1.5 cm s" . Values are means ± SE, N = 5. (B) Measured bed shear stresses at 5 cm s"1 (black) and 1.5 cm s"1 (white) and their corresponding theoretical laminar and turbulent calculated shear stresses (based on eqs. 7.25 and 7.44, respectively, in White 1994). 94
Fig. 2.8. Fertilization success as a function of distance downstream for Dreissena bugensis at four different bottom roughnesses and two different velocities in Lake Erie. Plots are means ± SE, N = 5. 95
Fig. 2.9. Bottom roughness for the four field configurations determined from a roughness profiler along aim transect, and vertical flow profiles from ADV measurements taken at 10, 30, and 50 cm along the sampling transect (data at 20 and 40 cm not illustrated). 96
Fig. 2.10. Fertilization success and ejection (Q2) and sweep (Q4) events as a function of distance downstream for Dreissena bugensis at four different bottom roughnesses and two different velocities in Lake Erie. Fertilization success plots are means ± SE, N = 5, and quadrant plots are based on the means of 120 s of data recorded at 25 Hz. 97
Fig. 2.11. Bed shear stress and Reynolds shear stress as a function of distance downstream for ADV data collected at four different bottom roughnesses and two different velocities in Lake Erie. Symbols represent means of 120 s of data recorded at 25 Hz. 98
Vll Fig. 3.1. Percent of individual Dreissena sp. larvae and model polystyrene spheres that exhibited bedload transport (panels A - D) and suspended (panels E - H) movement along the flow chamber bottom for three mussel bed configurations and the two reference turbulence configurations (e.g., grid and cylinder) at five different velocities ranging from 3.1 cm s"1 to 11 cm s"1. Plots are means ± SE, N = 15 (5 locations x 3 replicates). 127
Fig. 3.2. Percent of individual Dreissena sp. larvae and model polystyrene spheres that exhibited bedload transport at a velocity of 3.1 cm s"1 (panels A - D) and suspended movement at a velocity of 11 cm s"1 (panels E - H) along the flow chamber bottom for three mussel bed configurations and the two reference turbulence configurations (e.g., grid and cylinder) at five different locations downstream. Plots are means ± SE, N = 3. 128
Fig. 3.3. Particle image velocimetry (PIV) velocity flow profiles for U= 11 cm s" over four 13-cm long sections of flow chamber bottom roughness configurations; (A) flat bottom, (B) low mussel density, (C) high mussel density, and (D) high mussel density patches 10 cm apart. Areas of high velocity swimming flow are represented by the dark grey contours and larger vectors. 129
Fig. 3.4. Shear velocities determined from PIV data using the law of the wall, calculated by multiplying the von Karman constant (K = 0.41) by the slope of the velocity versus ln(z) in the logarithmic portion of the boundary layer, for eight different turbulence configurations over five different locations downstream at velocities of 3.1 cm s"1 and 11 cm s" . Values are means ± SE, N = 3. 130
Fig. 3.5. (B) Larval settlement for Dreissena sp. larvae onto 12x11 cm, 0.6 cm thick nylon scouring pads deployed over mussel beds of different configurations at Evan's Point in the eastern basin on Lake Erie, and in the water column above each of the different configurations. Plots are means ± SE, N = 20. (A) The roughness of each configuration is represented by kavg, with the coefficient of variation (%) of each plot in brackets. 131
Fig. 3.6. Quadrant analysis of the turbulence measured over mussel beds of different configurations at Evan's Point in the eastern basin on Lake Erie. Quadrant plots are three separate 120 second time-series means ± SE from three different ADV profiles between 2007 and.2008. Depth of mussel beds = 40 cm 132
Figure 4.1. COMSOL flow streamline plots illustrating the three main flow regime types over 2D transverse square roughness elements: (A) isolated roughness over A/k = 12; (B) skimming flow over A/k =3.3; and (C) wake interference flow over A/k =8.3. Also shown on (A) is the roughness parameters of roughness height (k), water depth (d), the longitudinal distance between roughness elements, or roughness spacing (A), and roughness groove width (/'). Note that the panels represent a portion of the model that illustrates the flow around the element. 151
Vlll Figure 4.2. COMSOL plot of the modeling environment for 2D transverse round roughness elements of A/k = 6.3 and the model free mesh structure, which adds more mesh elements around a complex boundary, indicated by the smaller and smaller mesh elements seen around the round roughness elements. 152
Figure 4.3. COMSOL plot of the released scalar concentration over 2D transverse triangular roughness elements. Scalar was released at a point 1 m in front of the first roughness element and at the same height of the first roughness element to a distance of one roughness height above it. The scalar was released from a 5-cm high flat interior boundary perpendicular to the flow. The A/k ratio is 6.25, corresponding to a predicted flow regime of skimming flow, which was confirmed by the scalar carried downstream with limited fluid entering the spaces between roughness elements. The white dashed line between the first two elements indicates the location of the RT value transect. 153
Figure 4.4. COMSOL plot comparing the number of bottom roughness elements (A; three elements, B; four elements, and C; five elements) downstream on the nature of the flow regime. The parameters shown are k = 70 and k = 6, giving a A/k ratio of 11.7 for all three plots. The predicted flow regime of isolated roughness is illustrated, which matches with the flow regime generated for all plots. 154
Figure 4.5. The flow regimes created from the ratio of roughness spacing (X) to roughness height (k). Isolated roughness flows are found to the left of the A/k ratio = 8, skimming flows are found to the right of that line, whereas wake interference flows are found right around the line A/k = 8. 155
Figure 4.6. COMSOL plots indicating a A/k ratio of 10 for the (A) triangle, (B) square, and (C) round roughness element geometries. Isolated roughness flow was observed for each geometry as indicated by the small flow recirculation regions behind most of the elements that dissipate before the next element downstream. 156
Figure 4.7. COMSOL plots indicating: (A) A/k ratio of 6.25 illustrating wake interference flow over 2D transverse round roughness elements; and (B) A/k ratio of 6.25 over 2D transverse square roughness elements illustrating skimming flow; and (C) A/k ratio of 2.5 over 2D transverse round roughness elements illustrating skimming flow. 157
Figure 4.8. The relationship between velocity and scalar relative transport within three flow regimes over square roughness elements: skimming flow (k/k < 7); wake interference flow (7 < X/k < 9); and isolated roughness flow (k/k > 9). Values are means ± SE, N = at least 5. 158
Figure 4.9. RT contours for the ratios of bottom roughness spacing (k) to roughness height (k) over square roughness elements. The lighter contours indicate higher RT values. The dashed white line represents the region of predicted wake interference flow (i.e., A/k = 8), with skimming flow to the right of the solid line and isolated roughness to the left. The solid white line represents the RT= 16 threshold value observed for
IX skimming flow. The white dotted lines in the bottom left comer indicate the subsection of A/k ratios and contour plots used for the triangle and round roughness geometries. 159
Figure 4.10. The relationship between roughness element geometry and. scalar relative transport for skimming flow (A/k < 7), wake interference flow (7 < A/k < 9) and isolated roughness flow (A/k > 9). Note the different values for the square geometries using the 49 A/k ratios, and the 16 A/k ratio subset. Values are means ± SE, N = at least 10. 160
Figure 4.11. Contours of RT over (A) triangular roughness elements and (B) round roughness elements for ratios of roughness spacing (X) and height (k). The lighter contours indicate higher R T values. The dashed white line represents the region of wake interference flow, with skimming flow to the right of the solid line and isolated roughness to the left. The solid white line represents the RT threshold for observed skimming flow. 161
x GENERAL INTRODUCTION AND LITERATURE REVIEW.
Introduction
External fertilization and a free-swimming larval stage are two key life-history strategies used by many marine and freshwater benthic sessile organisms. These life-history stages connect benthic organisms to the pelagic zone, and once larvae settle on the benthos, they close the loop. Both of these stages are strongly influenced by near-bed hydrodynamics, as sperm and larval swimming speeds are typically much lower than ambient velocities, and thus gametes and larvae are typically modeled as passive particles. Turbulence near the substratum and related physical forces are examples of 'just enough, but not too much' for free-spawning, for sessile organisms like bivalves (Mead and Denny 1995).
Fluid dynamics
Both biological and physical processes play important roles in determining the distribution of settled larvae (Pawlik and Butman 1993). The Reynolds number can be used as a central scaling parameter that makes order of a diverse set of physical variables
(Vogel 1994). The Reynolds number is defined as the ratio of inertial to viscous forces given by
Re = ^ (1) M where U is flow velocity, 1 is characteristic length (usually the water depth), p is density, and u. is dynamic viscosity (Vogel 1994). Re can be used to determine which forces are the most important in motion and allow for the prediction of the flow characteristics. The
Reynolds number for flow through an open channel can be evaluated using the channel hydraulic diameter for the characteristic length. Creeping flow refers to very low
1 Reynolds number, Re«l. Laminar flow would have Re of 1-500, transitional flow from laminar to turbulent would have a Re from 500 to 1000, and fully turbulent flow would have a Re over 1000 (Vogel 1994). Re transitions differ for each given geometry.
Particles or bodies are also subject to numerous hydrodynamic forces associated with turbulent flow (Crimaldi et al. 2002). These forces include lift (Fya), form drag (FDrag), gravitational forces (Foravity), fluid acceleration reaction forces, and forces due to viscous and turbulent shear stresses (Fig. II).
Recruitment and dispersal of benthic organisms is strongly influenced by bottom boundary-layer flow dynamics (Nowell and Jumars 1984). The boundary layer can be defined as the zone near a solid boundary (e.g., the lake bed) where a velocity gradient and fluid shear stress exists (Nowell and Jumars 1984). The zone where no mean velocity gradient exists is called the free stream (Fig. 12). In the benthic boundary layer, the simplest flow regime is laminar, unidirectional flow with no temporal or spatial fluctuations across a flat bed. A turbulent boundary layer, however, is typically divided into at least three regions vertically. The viscous sublayer is a very thin region, usually less than a millimeter thick, immediately next to the boundary where the velocity gradient is linear, viscous shear stress dominates and turbulence is absent (Nowell and Jumars
1984). Flow is essentially laminar in this region, with the thickness termed 8V. The buffer or log layer is the next region of turbulent flow, and represents the transitional region between laminar and fully turbulent flow. Small-scale eddies are created in this region that are subsequently advected into the fully turbulent region above (Nowell and
Jumars 1984). This fully turbulent outer part is the final region of turbulent flow, occupying most of the flow depth up to the free-stream region.
2 Fyft
'Drag
f1 Gravity
Figure II. Forces acting on a static body. U represents direction of flow.
Most boundary-layer flows of interest to aquatic benthic organisms are turbulent because of spatial scales and velocities present in the environment, although laminar flow can be important for very small organisms and at small spatial scales.
Turbulence has classically been defined as the unpredictable, chaotic-like motion of a fluid (Denny and Shibata 1989). A more appropriate definition of turbulence would be the fluctuations in water velocity in space and time. In order to understand turbulence, one must first remember that velocity is an example of a vector, with both magnitude and direction. Each individual or instantaneous measurement of the velocity (u) in the x or the streamwise direction can considered as the sum of two components; the mean velocity
(if) and the fluctuation from the mean velocity (u'). It is u' that represents the magnitude of the turbulent component of the streamwise flow, which can be expressed statistically as the root mean square velocity (£/RMS) and as the turbulence intensity (77= URMS/ W).
When u' is incorporated with the vertical fluctuation component in the z direction (w')
3 using a similar equation to that of u', we can then have an idea of the direction of the turbulence (Fig. 14).
Turbulence can be created in a number of ways, and turbulence near the bottom is caused in part by the no-slip condition, which generate shear stress or friction at the bed
(xbed)- The bed shear stress is defined as the rate at which a fluid's momentum is decreased by viscous interaction with substratum (Denny and Shibata 1989), and can then be used to determine the shear velocity, u*, expressed as:
u* = (Tbedbed/p)^ (2) and by the law of the wall given by
u, f - \ u- —-l Irn (3) K vzoy
Where K is the von Karman constant and zo is the roughness height (Ackerman and
Hoover 2001). Within the water column, turbulence magnitude can be quantified by the
Reynolds shear stress, a measure of the correlation between the instantaneous velocity fluctuations in the streamwise and vertical directions. The Reynolds shear stress can be expressed as:
where p is density of the fluid, u' is the instantaneous velocity deviation from the mean in the streamwise x direction, and w' is the instantaneous velocity deviation from the mean in the vertical z direction. Quadrant analysis is a technique that breaks down each instantaneous velocity fluctuation into one of four regions based on the net direction of the turbulent fluctuation (Fig. 14). Turbulent flow near the bed is typically dominated by quadrant 2 events termed ejections, where u' is negative (a fluctuation in the upstream direction) and w' is positive (a fluctuation away from the bottom) resulting in a fluid motion away from the bed, and quadrant 4 events termed sweeps, where u' is positive and w' is negative resulting in a fluid motion towards the bed (O'Connor and Hondzo 2008;
Fig. 14). Ejections and sweeps are about equal in frequency in the near-bed region, whereas ejections dominate the remaining water column (Cellino and Lemmin 2004).
Experimental evidence has shown that these processes are important in sediment re- suspension and transport. Sweeps hit the bed and spread the resuspended particles horizontally, whereas ejections entrain this sediment at the bed and carrying it up through the water column to close to the surface in a depth-limited boundary layer (Cellino and
Lemmin 2004). Topographic forcing of benthic bedforms and bed roughness (e.g. subsurface dunes or depressions) has been suggested as one of the main factors controlling the creation of these turbulent events, as differences in bed roughness and height can cause variable flow patterns (Best and Kostaschuk 2002). If gametes and larvae behave predominantly as passive particles in such systems, then they would be expected to behave like deposited or suspended sediments. Sediment transport models should, therefore, be applicable to biological processes (with only some minor modifications like accounting for the ability to attach and other behaviours) such as spawning and larval settlement. To the best of my knowledge, the validity of sediment transport models for gamete and larval transport, however, has not been investigated in detail and this gap may represent an oversight.
5 free stream
fully turbulent
Boundary layer thickness
velocity (u)
Figure 12. Vertical profile of a benthic boundary layer (adapted from Nowell and Jumars 1984).
u = u + u w-w + W
Figure 13. An example of an acoustic Doppler velocimeter velocity time series (taken 30 cm above the bed at Evans Point, Lake Erie on August 9, 2008) indicating an example of a fluctuation in the stream-wise velocity component, U.
+w'
Ejection Outward interaction Quadrant 2 Quadrant 1
+u'
Inward interaction Sweep Quadrant 3 Quadrant 4
-w Figure 14. Diagram of the u 'w' quadrant plane (based on Cellino and Lemmin 2004).
6 External fertilization
External fertilization, or broadcast spawning, can be divided into four stages: (1) gamete release, (2) gamete dispersal, (3) gamete recognition, and (4) fertilization (Okubo et al.,
2001). Of these four stages, three are directly related to the flow regime surrounding the parent organisms, those being gamete release, dispersal, and fertilization. External fertilization is problematic in a dynamic environment for two main reasons, the rapid dilution and the limited lifespan of gametes (Levitan 1995). Fertilization success also depends on the characteristics of the spawned gametes, the individual spawning organism, the population as a whole, and environmental factors (Table II). To date, sperm dilution has been considered to be the largest problem for most broadcast spawners, and thus has been the most studied. Allee (1931) and Mortensen (1938) were first to recognize that with the short life span and slow swimming speeds of sperm, the probability of sperm-egg encounters would be extremely low. Since sperm would act as passive particles entrained in flow, the potential for currents to dilute sperm to concentrations that would limit fertilization would be high, and thus reduce the number of fertilized eggs. This situation would hence result in 'sperm limited population dynamics'
(Levitan and Petersen 1995).
Organisms have evolved strategies that increase fertilization success, largely by increasing the probability of gamete encounter by limiting the effects of sperm dilution
(Bishop 1998; Yund and Meidel 2003). Fucoid algae, which broadcast both male and female gametes, spawn in response to a drop in dissolved inorganic carbon levels, which correspond to periods of very low water movement. The low flow velocity and therefore low dilution rates results in > 95% of eggs being fertilized (Serrao et al. 1996; Pearson et
7 al. 1998). The rate of sperm release has also been shown to affect fertilization success in the polychaete worm Galeolaria caespitosa, where slow sperm release rates lead to higher overall fertilization success, less sperm wastage, and less polyspermy (Marshall and Bolton 2007).
Denny and Shibata (1989) modeled external fertilization using data from
Pennington's (1985) experiments on sea urchins. They found that the fraction of fertilized eggs decreased with (1) an increase in mean velocity, (2) the distance between males and females, and (3) the spatial and temporal variation in the eddy diffusivity.
Some of the variation in eddy diffusivity includes turbulent events such as 'sweep' and
'ejection' motions. Ejections are high-speed fluid motions away from the bed that are expected to play a critical role in resuspending gametes into the water column (Abelson and Denny 1997). A recent study has suggested that sperm transport patterns can be site specific and thus dependent on the local flow environment (Yund et al. 2007).
Fertilization success is also dependent on turbulent mixing bringing eggs and sperm together. When sea urchins spawned into still water, little fertilization would occur as eggs and sperm would not mix and simply accumulated on the aboral surfaces of the adults (Pennington 1985). At high flow velocity and high turbulence, however, fertilization success was also reduced due to the rapid dilution of sperm and the decreased gamete encounter. High turbulence can also affect gametes post-fertilization as sea urchin eggs exposed to high shear stresses have resulted in abnormal embryonic development and high mortality rates through the early blastula stage (Mead and Denny
1995; Riffel and Zimmer 2007).
8 Table II. Factors influencing fertilization success for broadcast spawners (from Okubo et al. 2001). Gamete Individual Population Environment 1. Sperm 1. Behaviour 1. Configuration 1. Fluid dynamics Morphology Aggregation Density Turbulence Behaviour Synchrony Population size Mixing Velocity Spawning posture Neighbour distance Currents Longevity Spawning rate Water-column discontinuities
2-Egg 2. Morphology 2. Demography 2. Topography Size Size Sex ratio Roughness Jelly coat Fecundity Age structure Water depth Surface chemistry Size structure Shelter
3. Condition 3. Condition 3. Community dynamics 3. Water quality Age Age Spawning sibling-species Temp Compatibility Energy reserves Predation Salinity pH
Whether freshwater benthic organisms in lakes are also exposed to the aforementioned turbulent features seen in marine systems remains to be determined, as
. does whether they utilize similar behavioural strategies seen in marine benthic organisms.
Nearshore (littoral) freshwater organisms would likely experience similar wave-driven hydrodynamic forces, but sublittoral and offshore organisms in the ocean are not likely to experience seiches and other hydrodynamic conditions unique to lentic systems. On the other hand, river species would be exposed to much higher unidirectional flow velocities than in most marine habitats with the exception of estuarine areas, and lake species could be exposed to turbulence induced by internal waves or unstable stratification. These are just a few examples of freshwater turbulent systems that could potentially create problems for broadcast spawners, but very little is known about how freshwater benthic species are actually influenced by such systems and if or how they are able to overcome such hydrodynamic forces.
9 Benthic invertebrate recruitment
Successful larval settlement is critical for the long-term viability of benthic sessile populations (Eckman 1983; Mullineaux and Butman 1990; Pawlik and Butman 1993;
Abelson and Denny 1997; Hadfield and Koehl 2004). Larval settlement can be separated into two key, sequential stages: (1) the transport of larvae to the substratum through passive or active processes, and (2) the subsequent establishment on or in the substratum through some means of attachment or burrowing. Although most larvae have the ability to swim, it has long been recognized that their own movements would not be adequate to carry them very far in relation to the vast expanse of habitat, and that they would subsequently need to rely largely on water movements to bring them into contact with substrata on which to settle (Crisp 1955).
Successful larval settlement is directly influenced by hydrodynamic forces, and in some taxa it is also influenced indirectly through the behaviour of dissolved settlement cues in flow. The induction of settlement by natural external chemical cues has been demonstrated for many and diverse species, with cnidarians, bivalves, and echinoids as examples (Hadfield and Koehl 2004). These chemical cues may come from different types of substrata (e.g., biofilms), prey species, or established adult populations (Pawlik
1992; Wainman et al. 1996). Such cues are dispersed and dissolved mainly via turbulent mixing and diffusion, and experimental evidence has shown a variety of behaviours in response to very dilute concentrations. Whether similar hydrodynamic influences on larval settlement operate in freshwater species remains to be examined.
Hydrodynamic modeling has been used successfully to study the influence of oceanographic processes on dispersal and recruitment of fish larvae with passive
10 dispersal, and to accurately predict larval distribution in space or time (Jenkins et al.
1999). However, one of the few experiments that attempted to link larval transport to sediment transport theory was Pawlik and Butman's (1993) study on passive larval mimics in a flume. Their objective was to investigate larval settlement over a range of steady, unidirectional flows and to include shear velocities that would exceed the threshold for suspended-load transport of larval mimics. Another study that applied a sediment transport model approach to benthic larvae was McNair et al. (1997), who applied sediment transport theory to a model proposed by Denny and Shibata (1989).
They focused strictly on one aspect of turbulent particle transport, however, specifically the average sinking time of a particle at a particular elevation in the water column.
Whether gamete and larval transport can be examined using sediment transport theory, will require more information than just sinking velocities to generate an accurate model.
The most common, and most studied, form of larval behaviour that influences larval distribution and dispersal is vertical migration; even very low swimming speeds can produce unexpected patterns of distribution (Jenkins et al. 1999; Shanks and Brink
2005). Modified hydrodynamic sediment transport models that account for larval behaviour can be used to examine the influence of such behaviours on larval distribution by comparing predicted results to observed distributions, both with and without larval behavioural components. A convergence of these two results would indicate a successful application of passive sediment transport theory to benthic ecological principles.
Dreissenid biology
The dreissenid mussels can be classified as follows (Turgeon et al. 1998):
11 Phylum: Mollusca Class: Bivalvia Linnaeus 1758 Subclass: Heterodonta Neumayr 1884 Order: Veneroida H. and A. Adams 1856 Superfamily Dreissenoidea Gray 1840 Family: Dreissenidae Gray 1840 Genus: Dreissena Beneden 1835 Species: Dreissena bugensis Andrusov 1897 - quagga mussel Species: Dreissenapolymorpha Pallas 1771 - zebra mussel
A key characteristic of all Dressenoidea is an apical shell septum to which the anterior adductor and anterior byssal retractor muscles are attached. This forward extension of the umbones and the drastic ventral anterior flattening have been thought to be associated with the need for firm attachment of the anterior adductor muscle (Morton 1993).
Dreissena polymorpha (zebra mussel) and Dreissena bugensis (quagga mussel) are introduced exotic species that appeared in the Great Lakes in the mid 1980's, brought in by ship ballast water (Hebert et al. 1989). Researchers have found that over the last few years, the once dominant zebra mussel has been displaced by the quagga mussel (Diggins et al. 2004). Dreissenid mussels utilize external fertilization and have a series of planktonic larval stages that are very similar to those of marine mussels (Fig. 15), and yet are unique among the native freshwater species that they are displacing (Ackerman et al.
1994).
Rationale
The role of bottom roughness-induced turbulence in external fertilization and larval recruitment has been an intensely studied topic in marine benthic ecology, but the majority of this work has been theoretical. The empirical application of this approach to freshwater organisms is a natural progression, and the examination of freshwater bivalves
12 that have life-history strategies that are similar to marine species should allow direct comparison. Bivalves are important benthic organisms and are a numercially dominant species in many environments. The fact that the study species to be examined, Dreissena polymorpha and Dreissena bugensis, are introduced species of considerable interest makes this study all the more worthwhile, as it will provide researchers a new insight into the mechanisms controlling their dispersal along with a predictive model to forecast their continuing spread throughout the Great Lakes. With the dreissenid mussels, not much is
tos-sc* Veki'r. Qtiisocanch
Figure 15. Life history stages of Dreissena polymorpa (from Ackerman et al. 1994).
known regarding external fertilization and larval settlement. Most of the experimental work on broadcast spawning has involved sea urchins, with very little work on mussels and nothing on dreissenids. The mechanisms controlling dreissenid spawning have been studied and established, but little is known once those gametes leave the mussel. Larval
13 settlement has been studied in driessenids to a larger degree, but most of this research concerns substrate selection, with physical forces largely ignored.
Hypothesis
The hypothesis of this thesis is that bottom roughness created by the presence of mussel populations will generate turbulence that affects the fertilization success and larval settlement patterns of freshwater dreissenid mussels. Specifically, spatial configurations of mussel populations (e.g., patches), i.e., bottom roughness elements, that are characterized by high 'sweep' frequencies will experience less sperm limitation, higher fertilization success, and higher larval settlement frequencies. Conversely, spatial configurations of mussel populations (e.g., high density contiguous populations) that are characterized by high 'ejection' frequencies are predicted to be sperm-limited and have low fertilization success, and to have lower larval settlement rates but higher larval dispersal rates. Sweep and ejection frequencies will be controlled by the spatial configuration of bottom roughness elements, and will help determine the nature of the flow regime (i.e., skimming flow, wake interference, or isolated roughness flow) immediately above the bed.
Objectives
This thesis will investigate the nature of bottom roughness and turbulent flow in determining two key aspects of benthic freshwater mussel ecology: external fertilization and larval settlement. The thesis has three main foci, addressed through four chapters:
14 1. To examine the influence of bottom roughness created by mussel populations in
the creation of near-bed turbulent events and their subsequent influence on gamete
dispersal and fertilization success;
2. To examine the effects of these near-bed turbulent events on larval dispersal and
settlement.
3. To use computational fluid dynamic modeling to examine the relationship
between bottom roughness configuration (i.e., form, height, and spacing), flow,
and the extent of gamete and larval transport.
Chapter 1 examines the role of biology in dreissenid broadcast spawning, through an examination of sperm dilution and age on fertilization success and modeling the role of roughness on sperm dilution. It addresses aspects of the first and third objectives.
Fertilization experiments under static fluid conditions were carried out to provide baseline data for the fertilization success and to examine whether polyspermy occurs, a condition which affects many broadcast spawning organisms (Marshall and Bolton 2007).
The fertilization results determined from the static tests were incorporated into a computational fluid dynamic model, using the computation fluid dynamic modelling program COMSOL, to examine the role of relative bottom roughness element height (k) on sperm dilution and sperm transport/retention.
Chapter 2 examines the role of velocity, bottom roughness, and turbulence on broadcast spawning fertilization success via sperm dilution, related to the first objective.
The effects of flow velocity, downstream distance from the spawning source, and turbulence on fertilization success were examined in a flow chamber in the laboratory and on a lake bed in the field (Evans Point, Lake Erie). Three types of turbulence were
15 created in the laboratory through; (a) three spatial configurations of mussel shells attached to the bottom of the flow chamber; (b) a mesh placed across the width of the flow chamber; and (c) a half-cylinder placed on the bottom of the flow chamber. Field experiments used a similar approach through the examination of a flat bottom and three mussel configurations under different velocities and downstream distances.
Chapter 3 uses a similar set of roughness and turbulence comparisons with respect to larval transport and settlement, thereby addressing the second objective. Larval bedload and suspended transport was examined in a flow chamber in which Dreissena sp. larval and physical models of larvae (polystyrene beads) were placed under different mussel population and turbulence-generating configurations. Larval settlement was
examined in the field, using nylon scouring pad collectors placed in the field over five
72-h periods.
Chapter 4 is a modeling study using COMSOL to address the third objective. In this case under two-dimensional roughness, the ratio of bed roughness separation (X) to bed roughness height (k) has been predicted to be a primary indicator for the nature of
flow above the bed (Schindler and Ackerman 2009). The hydrodynamics and scalar transport within skimming flow, isolated roughness flow and wake interference flow was
examined using a range of biologically-relevant configurations characterized by non- dimensional numbers. These ratios were then examined with respect to sperm and larval transport using a scalar release diffusion-convection model, to determine the applicability
of the model and scalar transport in describing sperm/larval transport.
16 References
Ackerman JD, Sim B, Nichols SJ, and Claudi R. 1994. A review of the early life history
of zebra mussels (Dreissena polymorpha): comparisons with marine bivalves.
Canadian Journal of Zoology 72: 1169-1179.
Ackerman JD, and Hoover TM. 2001. Measurement of local bed shear stress using a
Preston-static tube. Limnology and Oceanography 46: 2080-2087.
Allee WC. 1931. Animal Aggregations: A Study in General Sociology. University of
Chicago Press, Chicago.
Best J and Kostaschuk R. 2002. An experimental study of turbulent flow over a low-
angle dune. Journal of Geophysical Research 107: 3135-3153.
Bishop JDD. 1998. Fertilization in the sea: are the hazards of broadcast spawning avoided
when free-spawned sperm fertilize retained eggs? Proceedings from the Royal
Society of London B 265: 725-731.
Cellino M and Lemmin U. 2004. Influence of coherent flow structures on the dynamics
of suspended sediment transport in open-channel flow. Journal of Hydraulic
Engineering 130: 1077-1088.
Crimaldi JP, Thompson JK, Rosman JH, Lowe RJ, and Koseff JR. 2002. Hydrodyanmics
of larval settlement: the influence of turbulent stress events at potential recruitment
sites. Limnology and Oceanography 47: 1137-1151.
Crisp DJ. 1955. The behaviour of barnacle cyprids in relation to water movement over a
surface. Journal of Experimental Biology 32: 569-590.
Denny MW and Shibata MF. 1989. Consequences of surf-zone turbulence for settlement
and external fertilization. The American Naturalist 134: 859-889.
17 Diggins TP, Weimer M, Stewart KM, Baier RE, Meyer AE, Forsberg RF, and Goehle
MA. 2004. Epiphytic refugium: are two species of invading freshwater bivalves
partitioning spatial resources? Biological Invasions 6: 83-88.
Eckman JE. 1983. Hydrodynamic processes affecting benthic recruitment. Limnology
and Oceanography 28: 241-257.
Hadfield MG and Koehl MAR. 2004. Rapid behavioral responses of an invertebrate larva
to dissolved settlement cue. Biological Bulletin 207: 28-43.
Hebert PD, Muncaster BW, and Mackie GL. 1989. Ecological and genetic studies on
Dreissenapolymorpha (Pallas): a new mollusc in the Great Lakes. Canadian
Journal of Fisheries and Aquatic Science 46: 1587-1591.
Jenkins GP, Black KP, and Keough MJ. 1999. The role of passive transport and the
influence of vertical migration on the pre-settlement distribution of a temperate,
demersal fish: numnerical model predictions compared with field sampling.
Marine Ecology Progress Series 184: 259-271.
Levitan DR. 1995. The ecology of fertilization in free-spawning invertebrates. In Ecology
of marine invertebrate larvae. Edited by McEdward L. CRC Press, Boca Raton, FL.
pp. 123-156.
Levitan DR and Petersen C. 1995. Sperm limitation in the sea. Trends in Ecology and
Evolution 10: 228-231.
Marshall, DJ and Bolton TF. 2007. Sperm release strategies in marine broadcast
spawners: the costs of releasing sperm quickly. Journal of Experimental Biology
210: 3720-3727.
McNair JN, Newbold JD, and Hart DD. 1997. Turbulent transport of suspended particles
18 and dispersing benthic organisms: how long to hit bottom? Journal of Theoretical
Biology 188: 29-52.
Mead KS and Denny MW. 1995. The effects of hydrodynamic shear stress on
fertilization and early development of the purple sea urchin Strongylocentrotus
purpuratus. Biological Bulletin 188: 46-56.
Mortensen T. 1938. Contributions to the study of the development and larval form of
echinoids. Kong. Danske Vidensk. Selsk. Nat. Math. Afd. 9 4: 1.
Morton, B. 1993. The Anatomy Of Dreissena Polymorpha And The Evolution And
Success Of The Heteromyarian Form In The Dreissenoidea. In T.F. Nalepa And
D.W. Schloesser Eds., Zebra Mussels: Biology Impacts And Control, Pp. 185-216.
Lewis/CRC Press, Inc., Boca Raton Florida.
Mullineaux LS and Butman CA. 1990. Recruitment of encrusting benthic invertebrates in
boundary-layer flows: a deep-water experiment on Cross Seamount. Limnology and
Oceanography 35: 409-423.
Nowell ARM and Jumars PA. 1984. Flow environments of aquatic benthos. Annual
Review of Ecology and Systematics 15: 303-328.
O'Connor BL and Hondzo M. 2008. Dissolved oxygen transfer to sediments by sweep
and eject motions in aquatic environments. Limnology and Oceanography 53: 566-
578.
Okubo A, Ackerman JD, and Swaney DP. 2001. Passive diffusion in ecosystems. In
Okubo and Levin, Diffusion and ecological problems: modern perspectives.
Springer-Verlag, New York. pp. 31-106.
Pawlik JR. 1992. Chemical ecology of the settlement of benthic marine invertebrates.
19 Oceanography and Marine Biology Annual Reviews 30: 273-335.
Pawlik JR and Butman CA. 1993. Settlement of a marine tube worm as a function of
current velocity: interacting effects of hydrodynamics and behavior. Limnology and
Oceanography 38: 1730-1740.
Pearson GA, Serrao EA, and Brawley SH. 1998. Control of gamete release in Fucoid
algae: sensing hydrodynamic conditions via carbon acquisition. Ecology 79: 1725-
1739.
Pennington JT. 1985. The ecology of fertilization of echnoid eggs: the consequences of
sperm dilution, adult aggregation, and synchronous spawning. Biological Bulletin
169: 417-430.
Riffell JA and Zimmer RK. 2007. Sex and flow: the consequences of fluid shear for
sperm-egg interactions. Journal of Experimental Biology 210: 3644-3660.
Schindler RJ and Ackerman JD. 2009. The environmental hydraulics of turbulent
boundary layers. In: D.T. Mihailovic and C. Gualtieri (eds.) Advances in
Environmental Fluid Mechanics. World Scientific, London. 40 pp.
Shanks AI and Brink L. 2005. Upwelling, downwelling, and cross-shelf transport of
bivalve larvae: test of a hypothesis. Marine Ecology Progress Series 302: 1-12.
Serrao EA, Pearson GA, Kautsky L, and Brawley SH. 1996. Successful external
fertilization in turbulent environments. Proc. Nat. Acad. Sci. USA 93: 5286-5290.
Thrush SF, Hewitt JE, Pridmore RD, and Cummings VJ. 1996. Adult/juvenile
interactions of infaunal bivalves: contrasting outcomes in different habitats. Marine
Ecology Progress Series 132: 83-92.
Turgeon, DD., Quinn Jr. JF, Bogan AE, Coan EV, Hochberg FG, Lyons WG, Mikkelsen
20 PM, Neves RJ, Roper GF, Rosenberg G, Roth B, Scheltema A, Thompson FG,
Vecchione M, and Williams JD. 1998. Common and scientific names of aquatic
invertebrates from the United States and Canada: mollusks, 2nd ed. Special
Publications 26. American Fisheries Society, Bethesda, Maryland. 526 p.
Vogel S. 1994. Life in Moving Fluids: the physical biology of flow, 2nd Ed. Princeton
Unversity Press, Princeton NJ. 467 p.
Wainman BC, Hincks SS, Kaushik NK, and Mackie GL 1996. Biofilm and substrate
preference in the dreissenid larvae of Lake Erie. Canadian Journal of Fisheries and
Aquatic Sciences 53: 134-140.
Yund PO and Meidel SK. 2003. Sea urchin spawning in benthic boundary layers: are
eggs fertilized before advecting away from females? Limnology and Oceanography
48: 795-801.
Yund, PO, Murdock K and Johnson SL. 2007. Spatial distribution of ascidian sperm:
two-dimensional patterns and short vs. time-integrated assays. Mar. Ecol. Prog.
Ser. 341: 103-109. Abelson A and Denny M. 1997. Settlement of marine organisms
in flow. Annual Review of Ecology and Systematics 28: 317-339.
21 CHAPTER 1: Biological and ecological mechanisms for overcoming sperm limitation in a freshwater mussel.
Abstract
Many broadcast-spawning benthic invertebrates are subject to sperm limitation yet achieve high population densities, an example being the dreissenid mussel invasion of the
Laurentian Great Lakes. The question remains whether biological or ecological/physical mechanisms (e.g., roughness on lakebeds created by mussel clusters) controls sperm limitation. Gamete dilution/longevity experiments were performed to determine whether dreissenid mussels are subject to sperm limitation, and computational fluid dynamic modeling was used to determine the potential influence of bottom roughness on sperm dilution in natural environments. Results indicated that the dreissenid mussels may be sperm limited, but the extent to which sperm dilution affects them is less than what has been reported for other broadcast spawners. Importantly, mussel clusters influenced external fertilization by retaining sperm in downstream eddies yet allowing some downstream transport from one cluster to another. In addition to high sperm potency at low sperm concentrations, mussel clusters may help explain the success of the dreissenid mussels as invasive species and the importance of ecological mechanisms for overcoming sperm limitation in other broadcast spawners.
22 Introduction
The dreissenid mussel invasion of the Great Lakes has garnered much research (e.g.,
Hebert et al. 1989; Ackerman et al. 2001; Johnson et al. 2006). Individual biological traits (e.g., byssal adhesion, epibenthic habitat) played a major role in this process (e.g.,
Ackerman et al. 1994; Maclsaac et al. 1999; Marsden and Lansky 2000), yet ecological as well as cumulative or antagonistic effects are likely as important (Devin and Beisel
2007). Surprisingly little effort has been devoted to the ecological role of broadcast spawning (i.e., propagule pressure) in fresh water, beyond the recognition that the process is similar to that of marine benthic invertebrates (e.g., broadcast spawning of eggs and sperm into the water column with external fertilization through random encounter; e.g.,
Ackerman et al. 1994). The frequency of broadcast spawning in marine organisms, however, has led to the development of a sperm limitation paradigm for benthic invertebrates (Levitan 1993). In this case, because sperm act as passive particles in flow, they have the potential to become diluted to concentrations that limit fertilization, and thus reduce fertilization success. It has been argued that this 'sperm-limited dynamic' should be common for broadcast-spawning invertebrates (Levitan 1993).
Certain biological and behavioural mechanisms have apparently evolved to limit sperm limitation. For example, the compound ascidian, Diplosoma listerianum, retains eggs and concentrates dilute sperm from the water column. Green sea urchins,
Strongylocentrotus droebachiensis, release eggs in a viscous matrix that adheres to their aboral surfaces (i.e., retains eggs). Variations in synchronous spawning of males and females of the red sea urchin, Strongylocentrotus franciscanus; also have been shown to help reduce sperm limitation (Bishop 1998; Yund and Meidel 2003; Levitan 2005). The
23 rate of sperm release has also been shown to affect fertilization success in the polychaete worm Galeolaria caespitosa, where slow sperm release rates lead to higher overall fertilization success, less sperm wastage, and less polyspermy (Marshall and Bolton
2007). Environmentally mediated behavioral/physiological responses have also evolved to reduce sperm limitation by matching spawning events to physically favorable times or events. For example, the fucoid seaweed Fucus distichus release gametes when dissolved inorganic carbon levels decrease near their surfaces during low tide and hence low current flow (Pearson et al. 1998; Pearson and Serrao 2006). The intertidal anemone
Oulactis mucosa has also been shown to spawn at low tide to minimize gamete dilution
(Marshall et al. 2004). The sea urchin Evechinus chloroticus spawns in shallow water just below a halocline, which limits the upward dispersal of the gametes, maintaining high gamete concentration and encounters, and thus fertilization (Lamare and Stewart
1998).
Fluid shear and small-scale turbulence have been shown to affect fertilization success whereby low shear and turbulence may promote sperm-egg encounter, but high shear and turbulence promote sperm dilution and thus limit fertilization (Mead and
Denny 1995; Riffell and Zimmer 2007). The hydrodynamics of the environment has been shown, both theoretically and experimentally, to play a key role in fertilization success (Denny and Shibata 1989; Levitan and Young 1995). For example, Denny and
Shibata (1989) demonstrated through mathematical modeling that fertilization rates should decline as the distance from a spawning male increased - because of the diffusion of sperm in turbulent flow. Levitan and Young (1995) extended this model and found that high population density increased fertilization success in small populations, but the
24 effect of high density was negligible in large populations. These models provide insight into broadcast spawning, yet they do not address key environmental parameters such as the influence of bathymetry and bed roughness on near-bed hydrodynamics. For example, in an analogous system the movement or retention of waterlogged seeds from tidal marsh plants has been shown to depend largely on hydrodynamics, with the spatial variation in seed deposition being related strongly to landscape elements that trap seeds
(Chang et al. 2008). In addition, a recent study has suggested that sperm transport patterns can be site specific and thus dependent on the local flow environment (Yund et al. 2007). It would be appropriate, therefore, to determine whether site-specific differences including structural features such as near-bed roughness or canopies created by benthic organisms affect sperm transport. Dreissenid mussels present an excellent model system in which to address these issues given that their aggregating behavior leads to mussel clusters (druse) on lakebeds. It remains to be determined whether these clusters affect hydrodynamics, sperm limitation and hence fertilization success. The objectives of this study are, therefore, to examine how sperm dilution and longevity affects the external fertilization of freshwater, broadcast spawning dreissenid mussels and to determine whether ecological/physical mechanisms created by their mussel clusters influences sperm dilution.
Methods
Study Species
Zebra {Dreissena polymorpha (Pallas, 1771)) and quagga mussels {Dreissena bugensis
Andrusov, 1897) are introduced species that appeared in the Laurentian Great Lakes in
25 the mid to late 1980s, brought in by ship ballast water (Hebert et al. 1989). They utilize external fertilization, and have a series of planktonic larval stages that are similar to those of marine mussels, and yet differ from the native freshwater species, which they are displacing (Ackerman et al. 1994). The conservation of life history characteristics more typical of marine species, along with their dominance within the Great Lakes, makes them ideal species for this study.
Spawning and maintenance
Dreissenid mussels (identified using Nichols and Black 1994) were collected from locations in Lake Ontario (Hamilton Harbour) and Lake Erie (Evans Point and Selkirk
Provincial Park), and were maintained in aerated 10 L aquarium in the Hagen Aqualab,
University of Guelph. The tanks were kept in a climate-controlled room at 13°C with a
12-h light:dark cycle. The mussels were fed 30 ml of a prepared mixture of nanoalgae and shellfish diet (Reed Mariculture Inc., CA) twice a week. Aquarium water was replaced with fresh well water weekly. Male and female mussels were induced to spawn using methods of Ram et al. (1993); they were placed in a bath of 10"3 M serotonin solution for 30 min, and then transferred to a beaker with 20 ml of aquarium water. Time until spawning ranged from 15 min to 4 h, with females typically spawning later than males. Once mussels had spawned, gametes were immediately removed from the beaker to determine initial gamete concentrations via four replicate hemocytometer counts, the completion of which defined the start of an experiment (i.e., t = 0). All experiments were run at 21-25°C to maximize fertilization (Ram et al. 1993).
26 Gamete dilution
The effects of sperm dilution on fertilization success of D. polymorpha and D. bugensis were determined using methods modified from Pennington (1985). Two milliliters of eggs were pipetted into 9 beakers each containing 50 ml filtered (200 urn mesh) fresh water (well water). Fresh sperm were quickly run through serial dilutions, to provide 8 sperm concentrations ranging from 107to 1 sperm ml"1. A 1-ml aliquot of each sperm dilution was added to each of the egg solutions in the eight beakers. The ninth beaker, which acted as an egg control to ensure there were no background sperm in the egg mixtures, received no sperm. A sperm control, which was used to check for background eggs in the original sperm solution, received no eggs. Solutions were allowed to incubate for 60 min before eggs were fixed with 2% KC1. Fertilization rates (%) were determined by counting the number of elevated fertilization envelopes on the first 200 eggs encountered under a compound microscope. Four replicates (each consisting of a different male and different female mussel) were conducted for each species, and results were analyzed using two-way ANOVA, following tests of assumptions (normality, homogeneity of variance).
The role of mechanisms to prevent polyspermy has been somewhat controversial in the dreissenid mussels. High sperm concentrations (up to 10 sperm ml") were reported to have not adversely affected embryonic development in D. polymorpha, suggesting some form of polyspermy blocking mechanism (Sprung 1993), whereas another study reported that a significant fraction of eggs at these high concentrations exhibited polyspermy (Misamore et al. 1996). To examine whether polyspermy was present in the dreissenid mussels, an experiment modeled after Styan et al. (2008) was
27 performed at the highest sperm concentration observed. Two milliliters of freshly
7 1 spawned sperm, at a concentration of 10 sperm ml" , were added to 2 ml of freshly spawned unfertilized eggs in four wells/replicates of a 12-well tissue culture plate. Eggs and sperm were mixed, and the developing zygotes/embryos were monitored over a 24-h period until they had passed the blastula stage, an indication of monospermic fertilization
(Styan et al. 2008). At each monitoring interval, 200 zygotes/embryos per replicate were observed, and the proportion of successfully developing zygotes/embryos was recorded, based on zygote/embryos development timelines (e.g., Luetjens and Dorresteijn 1995;
Wright etal. (1996).
The effect of egg dilution on fertilization success was determined over a subset of
27 trials from the sperm dilutions, 17 for D. polymorpha and 10 for D. bugensis. These trials compared the maximal fertilization rate obtained from each sperm sample to the egg concentration spawned by each individual female for that specific trial. The egg concentrations ranged from 220 to 3,600 eggs ml"1 representing the full range of egg concentrations seen over the course of the experiment, which is consistent with egg concentrations seen in other laboratory (Stoeckel et al. 2004) and field studies
(Stoeckmann 2003). Trials were examined using linear regression that compared the maximal fertilization observed in each trial to the egg concentration for each female.
This procedure provides the opportunity to determine whether egg limitation plays any role in fertilization.
A meta-analysis of sperm dilution rates from this and other reports of broadcast spawning invertebrates was undertaken to ascertain patterns among broadcast spawning species. A modified x analysis (Hedges et al. 1999) was used to compare the overall
28 expected mean fertilization rates from all species at a given sperm concentration to the observed species-specific fertilization rates. A significant deviation from the expected rate would indicate a significantly different fertilization rate for that species compared to the other species.
Gamete longevity
The effects of increasing gamete age post spawning on fertilization success was examined using sperm that were run through seven 10-fold dilutions, five of which were used, providing concentrations of 107, 105, 104, 103, and 101 sperm ml"1. One milliliter of freshly spawned eggs was added to six 50-ml beakers (i.e., 5 sperm dilutions and 1 sperm control), and 1 ml of each sperm dilution was added to the egg solution in the first six beakers at time 0. This process was repeated for each sperm dilution at 15, 30, 45, 60, 90 and 120 min, each time freshly spawned eggs were used, and three replicates (different male and female mussels) were used for each mussel species. Mixtures of one egg and one sperm control were used to check for any background levels of sperm or eggs in their respective initial solutions. ANCOVA was used to test for any species or dilution effects, with species and dilution as the categorical predictors and time as the continuous covariate predictor (e.g., Manriquez et al. 2001). Sperm half-life was determined for each sperm concentration by plotting fertilization success vs. time post spawning and identifying the time it took to reach 50% of the maximal fertilization observed
(Manriquez et al. 2001).
29 Computational fluid dynamic modeling
Given that one of the objectives of this study is to determine effects of physical features on the bed on near-bed fluid dynamics and thus sperm dilution, and that detailed field observations are difficult to obtain, computational fluid dynamic modeling was used to model this system. Specifically, COMSOL multiphysics (version 3.4, COMSOL Inc.) computational fluid dynamic modeling program was used to examine how different expressions of bottom roughness influence benthic hydrodynamics including scalar (i.e., sperm) transport and thus external fertilization. In this case, the transport and dilution of released sperm under the particular hydrodynamic conditions around a biological roughness element provide information on sperm limitation.
Relative roughness (HRei = HR/Z; where HR is the height of the roughness element and Z is height or depth of the flow field) was varied between 0.1 and 0.9 to examine the
effect of increasing roughness element height on the flow environment. This span of HRCI
includes a range of biological systems from dreissenid mussel clusters (HRCI = 0.15) representative of natural dreissenid populations found in the Great Lakes (Brady et al.
1995) to much larger coral elements (HRei = 0.75; e.g., coral heads) representative of coral reefs (Glynn 1976; Rogers 1979; Overholtzer-McLeod 2006). Models were run initially under the 3-dimensional (3D) k-s turbulence model that converged for high water velocities (e.g., 20 cm s"1) but did not converge for ecological relevant velocities (e.g., 5 -
10 cm s"1). Therefore, a 2-dimensional (2D) k-s turbulence model that did converge at lower velocities was used. Model solutions were performed using the built-in segregated solver GMRES. Subdomain flow field parameters were set to fresh water and seawater at
20°C (freshwater: density (p) = 1,000 kgm" and dynamic viscosity (u) = 1 x 10" Pas;
30 seawater: p = 1,024 kgm" and jx = 1.08 x 10" Pas). Details for the spatial dimensions correspond to those used in the mussel scenario outlined below. The spatial dimensions for the coral reef are provided separately.
The mussel scenario consisted of a spatial domain 5 m long and 1 m thick above the bottom (e.g., the depth at Evans Point, Lake Erie ranged from 0.5 to 1 m). The bottom boundary was set to the wall logarithmic boundary layer condition, one vertical boundary was set to the inlet condition (10 cm s") and the opposite boundary was set to the outlet condition. All other boundary domains were set to the symmetrical boundary condition (i.e., to restrict the model domain without affecting the hydrodynamic characteristics on either side of the boundary). Three simplified mussel clusters consisting of hemispheres with a 15-cm radius (14.97 ± 2.63 [mean ± standard error] cm,
N = 10 as measured at Evans Point, Lake Erie) were placed on the bottom centerline of the flow field 20 cm apart at 3.5, 4, and 4.5 m downstream to allow adequate flow development. The spatial domain was modeled using a free mesh with 751 mesh elements that was generated with COMSOL (the free mesh approach generates more and smaller mesh elements around boundaries). The full range of HRei (0.1 to 0.9) was examined by varying the vertical element height but keeping the other dimensions (e.g., element width and spacing) constant.
The coral scenario consisted of a spatial domain 25 m long and 4 m thick.
Boundary conditions were identical to those of the mussel scenario. Three simplified coral heads consisting of ellipsoid elements with a height of 3 m and width of 2 m at their base were placed on the bottom centerline of the domain 4 m apart 13, 18, and 23 m downstream of the inlet to replicate natural coral reef conditions and parameters (Glynn
31 1976; Rogers 1979; Overholtzer-McLeod 2006). The spatial domain was modeled using a free mesh with 914 mesh elements that was generated with COMSOL. As indicated above, the velocity (U) for both scenarios was 10 cm s"1 to allow comparison under the fully turbulent flow conditions characteristic for both types of habitats (Ackerman et al.
2001; and Robinson et al. 2007, respectively).
A 2D convection and diffusion, steady-state model was applied to the velocity data from the models described above to examine the transport of a scalar concentration
(e.g., sperm) over the roughness elements. In this case, the scalar was released continuously from the downstream half of the upstream-most roughness element (cluster or coral head) to simulate the release of sperm. A scalar (i.e., sperm) concentration of 1 mol m" was used with a diffusion coefficient of 10" m s" , which has been reported for sperm in other broadcast spawners (Reidel et al. 2005; Inamdar et al. 2007). Each scalar concentration was determined by recording the modeled results at five points along a vertical transect perpendicular to a downstream transect that intersected the top of each roughness element.
Results
Polyspermy
The minimum appropriate stage of development was based on published reports
(Luetjens and Dorresteijn 1995; Wright et al. 1996), which indicate that successful zygotes/embryos would have achieved at least the 2-cell stage 1.0 h post fertilization, the
4-cell stage after 2.0 h, the 8-cell stage after 3.0 h, and the trochophore larval stage after
24 h. Results indicate that > 90 % of the eggs had a fertilization envelope at 0.5 h post
32 fertilization (Fig. 1.1). The proportion of successful development decreased through time but remained reasonably high (i.e., > 80% had achieved at least the 8 cell stage by 3.0 h post fertilization). Specifically, 86.00 ± 0.65% of the zygote/embryos observed had achieved the early trochophore larval stage, indicated by the presence of apical flagellae by 24 h post fertilization.
Gamete dilution
The sperm concentration curve (Fig. 1.2) had a sigmoidal-like shape with limited
1 9 1 fertilization success between 10 and 10 sperm ml" , followed by a large rate of increase in fertilization success between 103 and 104 sperm ml"1, and then a lower rate of increase from 105 to 107 sperm ml"1. These differences in fertilization success related to sperm concentrations were significant (two-way ANOVA, sperm concentration F7:63 = 842.5, P
< 0.0001), but no significant species difference was found (D. polymorpha vs. D. bugensis; two-way ANOVA , species Fi^3 = 2.75, P = 0.104) nor was there significant interaction between the two factors (F7>63 = 0.57, P = 0.775). This relationship between sperm concentration and fertilization provides necessary evidence for a sperm-limiting situation, given that there was no relationship between the naturally-spawned range of egg concentrations and percent fertilization (D. polymorpa: R = 0.06, P =F 0.34; D. bugensis: R2 = 0.067, P = 0.47) (Fig. 1.3). Sufficient conditions for sperm limitation require that sperm dilution occurs in the field.
33 Sperm longevity
Fertilization success declined linearly with the age of sperm for 10 to 10 sperm ml" and nonlinearly for 104 to 107 sperm ml"1 (Fig. 1.4). The decrease in fertilization success between sperm age of 0 and 120 min ranged from 57% in the 107 sperm ml" treatment to
81% in the 101 sperm ml"1 treatment. These were significant differences with respect to sperm dilution but not to species as revealed by ANCOVA (dilution: F4J99 = 581.2, P <
0.0001; species: F1199 = 0.127, P = 0.721). As evident in Fig. 4, the covariate (time) was significant (FU99 = 594.5, P < 0.0001) as was the intercept (FU99 = 4466, P < 0.0001).
No significant species x dilution interaction was found (F4J99 = 0.002, P = 0.999). A three-way repeated measures ANOVA was also performed to examine the fertilization success over time, which was also significant for sperm dilution (F^HO = 111.17, P <
0.00001). No significant difference was found between Dreissena species (F^HO = 0.513,
P = 0.997), nor were there any significant two-way or three way interactions (P values ranged from 0.978 to 0.999).
Potential sperm transport in the field
The overall flow conditions were turbulent based on Reynolds numbers (Re = pUl u"1) of
1.0 x 10 and 3.8 X 10 for the mussel and coral scenarios, respectively, where the characteristic length scale (/) was equal to the depth of the flow (Z) (the hydraulic diameter of the 3D model flow field was also calculated, and gave similar Re values).
The flow conditions would also be turbulent at the scale of the respective roughness elements based on Re = 1.5 X 104 and 2.9 x 105 for the mussel clusters and coral heads, respectively, where / is equal to the height of the roughness element (HR).
34 In both cases, the fluid traveled downstream (from left to right) with some acceleration (convergence of streamlines) immediately upstream of the roughness elements (e.g., mussels and coral heads). There was a recirculation zone and deceleration
(divergence of streamlines) of the flow immediately downstream of the individual roughness elements (Fig. 1.5). In the case of the mussel clusters, the recirculation zone extended above the bottom to less than 10% of the modeled water depth, as indicated by the small vortices immediately behind the individual mussel clusters. In this scenario, 47
± 2.3% (N = 5 in all cases) of the scalar was retained downstream of the first mussel cluster, but 53 ± 2.1% of the released scalar was transported downstream of the second mussel cluster (Fig. 1.5C).
Flow around the coral heads had a different pattern of recirculation. In this scenario, the recirculation zone was large and extended to -50% of the modeled water depth. Moreover, large eddies were evident downstream of each reef element. These areas of recirculation would essentially isolate a large area of the reef from the rest of the flow field, creating isolated patches between the individual coral heads. Indeed, scalar released within the first coral head was retained predominantly by the downstream eddies leading to 70 ± 3.1% of the scalar being retained downstream of the first coral head, and only 24 ± 2.9% of the released scalar transported downstream of the second coral head
(Fig. 1.5D). Similar patterns were found using the 3D model at 20 cm s"1, the main difference being slightly larger areas of recirculation downstream of the modeled bed elements (data not presented).
The effect of roughness height on the pattern of downstream scalar transport was examined to determine the scalar concentration downstream of the second roughness
.35 element (Fig. 1.6). In this analysis, there was a curvilinear decline in sperm concentration from 63 ± 2.1% at a relative roughness height (HRei) of 0.1 to 18 ± 2.2% at
HRei of 0.9. There was, however, a noticeable break in the curve at HRei = 0.25 indicating a transition (i.e., 50%) from downstream scalar transport to upstream scalar retention.
Interestingly the mussel cluster (HRei = 0.15) results fall within the downstream scalar
transport region whereas the coral head scenario (HRCI = 0.75) falls on the curve in the upstream retention region (i.e., low sperm concentration downstream of the second coral head).
The relationship between bathymetry and sperm limitation was examined by relating the modeled scalar concentration (e.g., Fig. 1.5C-D) to the fertilization success vs. sperm concentration curves (Fig. 1.2). In other words, the scalar concentration recorded downstream from the point of release was converted to fertilization success using the results measured in the laboratory experiments. The downstream distance was normalized to a biological roughness element scale such that each integer value of the relative distance (XRei) represents the location of the cluster/coral head (i.e., 1 = first element, 2 = second element, 3 = third element). The fertilization success declined with
XRei for the four roughness heights examined (HRCI = 0.1; 0.15 = mussel cluster; 0.75 = coral heads; and 0.9; Fig. 1.7).
The range of relative roughness heights illustrate how the difference between downstream sperm transport (HRei = 0.1) and upstream sperm retention (HRei =.0.9) affect
fertilization success (Fig. 1.7). For HRCI < 0.25 there was a relatively smooth, curvilinear decline in fertilization success downstream of release, with a 17.083 ±0.004% drop in
fertilization success for 2 < XRei < 2.5. Conversely for HRCI > 0.25 there was a 64.01 ±
36 0.04% drop in fertilization success for 2 < XRCI < 2.5. Interestingly, there was higher
fertilization success for HRCI > 0.25 compared to HRCI < 0.25 for 1 < XRei < 2, likely due to the larger recirculation zone downstream of the first roughness element that would reduce sperm dilution. The smaller recirculation zone and greater sperm transport downstream of the first roughness element for HRei < 0.25 led to lower fertilization success for XRei <
2 but higher fertilization success downstream (i.e., XRei > 2) compared to HRei > 0.25.
Clearly fluid dynamic differences created by bottom roughness lead to significant differences in broadcast spawning.
Discussion
The high developmental success observed (Fig. 1.1) supports Sprung's (1993) suggestion of some form of polyspermy blocking mechanism in dreissenid mussels. Consequently, it is possible to conclude that fertilization success declined along with sperm concentration, indicating that dreissenid species are likely affected by sperm limitation
(Fig. 1.2). The modeled results suggest that sperm limitation occurs, given the calculated decline in fertilization success as distance from the spawning source increased. This pattern is similar to other broadcast spawners that have been reported with respect to multiple sperm concentrations (Fig. 1.8). In this case, no significant differences in fertilization success were found among the seven species examined at sperm concentrations between 10 and 10 sperm ml" . From the figure, three species
{Montipora digitata, Haliotis laevigata, and Platyhyra sinensis) suffer from polyspermy at sperm concentrations > 10 sperm ml" and thus show lower fertilization success
(Oliver and Babcock 1992; Babcock and Kessing 1999). Interestingly, the two dreissenid
37 mussel species have significantly higher fertilization success rates than the other broadcast spawners at concentrations < 103 sperm ml"1 (xs2 = 11.1, P < 0.01). Differences in fertilization success rates are especially pronounced at sperm concentrations of 101 and
9 1
10 sperm ml" , where a success rate of-10-15% was observed, compared to rates approaching 0% in the other species.
This significantly higher fertilization success rate at lower sperm concentrations suggests higher sperm potency (or virility) in dreissenid mussels, which might provide some resistance to sperm dilution by turbulent flow. This would enable dreissenid sperm to fertilize under low initial population densities and/or under more dynamic environmental conditions (i.e., more dilute sperm concentrations) than the other broadcast spawning species that have been examined. This higher potency may also play a critical role related to propagule pressure (Colautti and Maclsaac 2004), by retaining its viability for a longer period of time and thus increasing the sperm's effective residence time.
Anything that would act to extend that 'effective' time frame (i.e., longer sperm viability in terms of potency) should also act to increase the flux and thus propagule pressure.
Sperm half life provides another measure of sperm potency that can be calculated from the sperm longevity experiments. As in the case of sperm concentration, dreissenid mussels and seven other broadcast spawners show similar patterns of sperm potency across a range of sperm concentrations (note the difference in species; Fig. 1.9). In this case, a high slope would indicate that fertilization success is inversely and strongly related to sperm concentration, whereas a shallow slope would indicate less change in fertilization success with sperm concentration. The curves for the dreissenid mussels are much lower in slope than all but one (Haliotis tuberculata) of the other broadcast-
38 spawning species, which would indicate a relatively consistent and uniform pattern of fertilization success over a large range of sperm concentrations. Importantly, species with shallow sperm-half life slopes would be able to withstand a larger range of flow conditions (e.g., environmental tolerance) and still be able to fertilize successfully with sperm that are potent over a wide range of concentrations.
The role of biological traits in overcoming sperm limitation in dreissenid mussels is a possibility in the case of sperm potency, but sperm potency alone, however, does not account for the dominance of dreissenid mussels. Wide environmental tolerances with respect to external fertilization in dreissenid mussels contribute to this, but so does habitat alteration by their biological structures (i.e., shell clusters), which affect the near-bed flow. The link between hydrodynamics and external fertilization has been examined to . various degrees theoretically (e.g., Denny and Shibata 1989; Levitan and Young 1995), but the role of bathymetry on fertilization is an area that has had little attention. These studies demonstrated that fertilization success is dependant on the distance away from the sperm source, and that parameters such as population size and density can influence this dynamic. The results from this study are generally consistent with those earlier models with respect to downstream sperm dilution. However, this study indicates that the decline in fertilization success downstream is influenced by bathymetric features, specifically by roughness height.
In the case of the modeled mussel scenario (HRCI =0.15; downstream transport zone HRei < 0.25), the majority of the scalar reaches all three mussel clusters, essentially connecting them as the recirculation zone downstream of each of them is small. Mussel clusters provide enough of an impediment to the flow, however, to reduce velocities and
39 reduce sperm dilution to a degree. Although a majority of the released sperm are confined between the first two mussel clusters, a substantial amount are still transported downstream to the third cluster and beyond, as shown by the uniform fertilization success downstream in the model (Fig. 1.7). Sperm released from a particular cluster, therefore, could encounter eggs released from that same cluster or be carried downstream to the next cluster This provides a mechanism to reduce sperm dilution and to potentially increase sperm - egg encounter in the downstream recirculation zone. Given that the nature of water currents in nearshore areas of the Great Lakes where dreissenid mussels are found can be characterized as temporal unidirectional flows that change direction due to seiches (Loewen et al. 2007), this resistance to sperm dilution while maintaining sperm transportability represents an ecological trait that makes dreissenid mussels successful invasive species.
In the coral scenario (HRei = 0.75; upstream retention zone HRei > 0.25), with the large recirculation zone downstream of each of the coral heads, the best way to ensure reproduction would be to maximize propagule pressure through spawning high concentrations of sperm that have a long half-life (and thus residence time) at high sperm concentrations. Dilution effects would be restricted since sperm are not exposed to free- stream velocities found outside of coral reef elements, with the model illustrating higher
fertilization success between the first two roughness elements (XRCI < 2) in the coral head scenario compared to the mussel scenario, because the region between coral heads is essentially isolated. The released sperm are confined almost entirely between the first and second coral heads, with only a small amount transported over/around the second coral head and onto the third. The model also demonstrates a large drop (i.e., 64%) in
40 fertilization success past the second coral head, simply because the limited sperm transport downstream of the coral head. A longer sperm half-life would extend the period at which sperm, if entrained out of the recirculation zones, could successfully fertilize eggs if they were encountered. The reef coral species Platygyra sinensis provides some support for this hypothesis in that at a concentration of 105 sperm ml"1 it has a sperm half life of 3.6 h, compared to ~1.6 h for the dreissenid mussels.
Unfortunately, P. sinensis fertilization drops considerably below 10 sperm ml" with zero fertilization success below 10 sperm ml" (Oliver and Babcock 1992).
Hydrodynamic analogues for these two scenarios can be found in sand bed rivers.
In this case, submerged dunes have significant impact on the nature of turbulent flow, and it is the shape of the dune that is the major factor (Best and Kostaschuk 2002). For example, asymmetric dunes with long, gently sloping upstream stoss sides and short, steep downstream lee sides generate a well-developed zone of permanently recirculating flow (Kostaschuk 2000), whereas more symmetric bedforms with essentially identical stoss and lee slopes generate a region of only intermittent flow recirculation (Best and
Kostaschuk 2002). A similar process is also seen at a smaller scale with pebble clusters in gravel rivers (Lacey and Roy 2008). Interestingly, the characteristic symmetrical shape of most dreissenid mussel druse should produce a similar pattern, and the models investigating the affect of bedforms on flow recirculation suggest that such alterations occur for dreissenid mussels as well.
Bed roughness influences external fertilization success. Sperm limitation likely affects dreissenid mussels, but it does not appear to affect them to the same extent as other broadcast spawners that have been examined. The higher fertilization success rate
41 at lower sperm concentrations and the relatively uniform sperm half-life across different sperm concentrations indicates higher sperm potency compared to other broadcast spawners. This higher reproductive potential reveals that dreissenids have ability to withstand more variable flow conditions and may explain their success as invasive species. Their ecology, however, provides the key mechanism to reduce sperm limitation through the alteration of the near-bed flow environment via mussel clusters. In this case, the flow environment limits sperm dilution behind individual mussel clusters, yet permits sperm to travel downstream from one cluster to another. These findings should be applicable to any benthic organism that utilizes broadcast spawning or a free-swimming larval stage, as the hydrodynamic forces they experience will be similar. The use of a 2D model does have it limitations, as 3D lateral flow around mussel clusters should act to influence sperm dilution as well and possibly to a higher degree as sperm plumes would be pulled in multiple directions. By beginning to examine the impact of physical structures on flow environments at the 2D level, a better understanding of the mechanisms and controlling factors for critical life history stages can be achieved, and establishes a framework that potential future 3D predictive fertilization/dispersion models can use for many benthic organisms, invasive or otherwise.
42 References
Ackerman, J.D., Sim, B, Nichols, S.J. and Claudi, R. 1994. A review of the early life
history of zebra mussels (Dreissena polymorpha): comparisons with marine bivalves.
Can. J. Zool. 72: 1169-1179.
Ackerman, J.D., Loewen, MR. and Hamblin, P.F. 2001. Benthic-pelagic coupling over
a zebra mussel reef in western Lake Erie. Limnol. Oceanogr. 46: 892-904.
Andre, C. and Lindegarth, M. 1995. Fertilization efficiency and gamete viability of a
sessile, free-spawning bivalve, Cerastoderma edule. Ophelia 43: 215-227.
Babcock, R. and Kessing, J. 1999. Fertilization biology of the abalone Haliotis
laevigata: laboratory and field studies. Can. J. Fish. Aquat. Sci. 56: 1668-1678.
Baker, M.C. and Tyler, P.A. 2001. Fertilization success in the commercial gastropod
Haliotis tuberculata. Mar. Ecol. Prog. Ser. 211: 205-213.
Benzie, J. A. H. and Dixon, P. 1994. The effects of sperm concentration, sperm:egg
ratio, and gamete age on fertilization success in crown-of-thorns starfish (Acanthaster
planci) in the laboratory. Biol. Bull. 186: 139-152.
Best, J.L. and Kostaschuk, R.A. 2002. An experimental study of turbulent flow over a
low-angle dune. J. Geophys. Res. 107: 3135-3154.
Bishop, J.D.D. 1998. Fertilization in the sea: are the hazards of broadcast spawning
avoided when free-spawned sperm fertilize retained eggs? Proc. R. Soc. Lond. B
265:725-731.
Brady, V.J., Cardinale, B.J. and Burton, T.M. 1995. Zebra mussels in a costal marsh: the
seasonal and spatial limits of colonization. J. Great Lakes Res. 21: 587-593.
43 Chang, E.R., Veeneklass, R.M., Buitenwerf, R., Bakker, J.P. and Bouma, T.J. 2008. To
move or not to move: determinants of seed retention in a tidal marsh. Funct. Ecol.
22: 720-727.
Colautti, R.I. and Maclsaac, H.J. 2004. A neutral terminology to define 'invasive'
species. Divers. Distrib. 10: 135-141.
Denny, M.W. and Shibata, M. 1989. Consequences of surf-zone turbulence for
settlement and fertilization. Am. Nat. 134: 859-889.
Devin, S. and Beisel, J.N. 2007. Biological and ecological characteristics of invasive
species: a gammarid study. Biol. Invasions 9: 13-24.
Glynn, P.W. 1976. Some physical and biological determinants of coral community
structure in the eastern Pacific. Ecol. Monogr. 46: 431-456.
Hebert, P.D.N., Muncaster, B.W. and Mackie, G.L. 1989. Ecological and genetic studies
on Dreissenapolymorpha (Pallas): a new mollusc in the Great Lakes. Can. J. Fish.
Aquat. Sci. 46: 1587-1591.
Hedges, L.V., Gurevitch, J. and Curtis, P.S. 1999. The meta-analysis of response ratios
in experimental ecology. Ecology 80: 1150-1156.
Inamdar, M.V., Kim, T., Chung, Y-K, Was, A.M., Xiang, X., Wang, C-W, Takayama, S.,
Lastoskie, CM., Thomas, F.I.M. and Sastry, A.M.. 2007. Assessment of sperm
chemokinesis with exposure to jelly coats of sea urchin eggs and resact: a
microfluidic experiment and numerical study. J. Exp. Biol. 210: 3805-3820.
Johnson, L.E., Bossenbrock, J.M. and Kraft, C.E. 2006. Patterns and pathways in the
post-establishment spread of non-indigenous aquatic species: the slowing invasion of
North American inland lakes by the zebra mussel. Biol. Invasions 8: 475-489.
44 Kostachuk, R. 2000. A field study of turbulence and sediment dynamics over
subaqueous dunes with flow separation. Sedimentology 47: 519-531.
Lacey, R.W.J., and Roy, A.G. 2008. The spatial characterization of turbulence around
large roughness elements in a gravel-bed river. Geomorphology 102: 542-553
Lamare, M.D. and Stewart, B.G. 1998. Mass spawning by the sea urchin Evechinus
chloroticus (Echinodermata: Echinoidea) in a New Zealand fiord. Mar. Biol. 132:
135-140.
Levitan, D.R., Sewell, M.A. and Chia, F.-S. 1991. Kinetics of fertilization in the sea
urcin Strongylocentrotus franciscanus: interaction of gamete dilution, age, and
contact time. Biol. Bull. 181: 371-378.
Levitan, D.R. 1993. The importance of sperm limitation to the evolution of egg size in
marine invertebrates. Am. Nat. 141: 517-536.
Levitan, D.R. 2005. Sex-specific spawning behavior and its consequences in an external
fertilizer. Am. Nat. 165: 682-694.
Levitan, D.R. and Young, CM. 1995. Reproductive success in large populations:
empirical measures and theoretical predictions of fertilization in the sea biscuit
Clypeaster rosaceus. J. Exp. Mar. Biol. Ecol. 190: 221-241.
Loewen, M.R., Ackerman, J.D., and Hamblin, P.F. 2007. Environmental implications of
stratification and turbulent mixing in a shallow lake basin. Can. J. Fish. Aquat. Sci.
64: 43-57.
Luetjens, CM., and Dorresteijn, A.W.C 1995. Multiple, alternative cleavage patterns
precede uniform larval morphology during normal development of Dreissena
polymorpha (Mollusca, Lamellibranchia). Roux's Arch. De v. Biol. 205: 138-149.
45 Maclsaac, H.J., Johannsson, O.E., Ye, J., Sprules, W.G., Leach, J.H., McCorquodale, J.A.
and Grigorovich, LA. 1999. Filtering impacts of an introduced bivalve (Dreissena
polymorpha) in a shallow lake: application of a hydrodynamic model. Ecosystems 2:
338-350.
Manriquez, P.H., Hughes, R.N., and Bishop, J.D.D. 2001. Age-dependent loss of
fertility in water-borne sperm of the bryozoan Celleporella hyaline. Mar. Ecol. Prog.
Ser. 224: 87-92.
Marsden, J.E. and Lansky, D.M. 2000. Substrate selection by settling zebra mussels,
Dreissena polymorpha, relative to material, texture, orientation, and sunlight. Can. J.
Zool. 78: 787-793.
Marshall, D.J. and Bolton, T.F. 2007. Sperm release strategies in marine broadcast
spawners: the costs of releasing sperm quickly. J. Exp. Biol. 210: 3720-3727.
Marshall, D.J., Semmens, D. and Cook, C. 2004. Consequences of spawning at low tide:
limited gamete dispersal for a rockpool anemone. Mar. Ecol. Prog. Ser. 266: 135-
142.
Mead, K.S. and Denny, M.W. 1995. The effects of hydrodynamic shear stress on
fertilization and early development of the purple sea urchin Strongylocentrotus
purpuratus. Biol. Bull. 188: 46-56.
Misamore, M., Silverman, H., and Lynn, J.W. 1996. Analysis of fertilization and
polyspermy in serotonin-spawned eggs of the zebra mussel, Dreissena polymorpha.
Mol.Reprod.Dev. 43:205-216.
Mistri, M. 2002. Ecological characteristics of the invasive asian date mussel, Musculista
senhousia, in the Sacca di Goro (Adriatic Sea, Italy). Estuaries 25: 431-440.
46 Nichols, S.J. and Black, M.G. 1994. Identification of larvae: the zebra mussel
(Dreissena polymorpha), quagga mussel {Dreissena rosteriformis bugensis), and
Asian clam {Corbicula fluminea). Can. J. Zool. 72: 406-417.
Oliver, J. and Babcock, R. 1992. Aspects of the fertilization ecology of broadcast
spawning corals: sperm dilution effects and in situ measurements of fertilization.
Biol. Bull. 183:409-417.
Overholtzer-McLeod, K.L. 2006. Consequences of patch reef spacing for density-
dependent Mortality of coral-reef fishes. Ecology 87: 1017-1026.
Pearson, G.A., Serrao, E.A. and Brawley, S.H. 1998. Control of gamete release in
flucoid algae: sensing hydrodynamic conditions via carbon acquisition. Ecology 79:
1725-1739.
Pearson, G.A. and Serrao, E.A. 2006. Revisiting synchronous gamete release by fucoid
algae in the intertidal zone: fertilization success and beyond? Integr. Comp. Biol. 46:
587-597.
Pennington, J.T. 1985. The ecology of fertilization of echinoid eggs: the consequences
of sperm dilution, adult aggregation, and synchronous spawning. Biol. Bull. 169:
417-430.
Ram, J.L., Crawford, G.W., Walker, J.U., Mojares, J.J., Patel, N., Fong, P.P. and
Kyozuka, K. 1993. Spawning in the zebra mussel {Dreissenapolymorpha):
activation by internal or external application of serotonin. J. Exp. Zool. 265: 587-
598.
Reidel, I.H., Kruse, K. and Howard, J. 2005. A self-organized vortex array of
hydrodynamically entrained sperm cells. Science 309: 300-303.
47 Riffell, J.A. and Zimmer, R.K. 2007. Sex and flow: the consequences of fluid shear for
sperm-egg interactions. J. Exp. Biol. 210: 3644-3660.
Robinson, H.E., Finelli, CM. and Buskey, E.J. 2007. The turbulent life of copepods:
effects of water flow over a coral reef on their ability to detect and evade predators.
Mar. Ecol. Prog. Ser. 349: 171-181.
Rogers, C.S. 1979. The effect of shading on coral reef structure and function. J.Exp.
Mar. Biol. Ecol. 41:269-288.
Sprung, M. 1993. The other life: an account of present knowledge of the larval phase of
Dreissenapolymorpha. In Zebra mussels: biology, impacts, and control. Edited by
T.F. Nalepa and D. Schloesser. Lewis Publishers, Boca Raton, Fla. pp. 39-53.
Stoeckel, J.A., Padilla, D.K., Schneider, D.W., and Rehmann, C.R. 2004. Laboratory
culture of Dreissena polymorpha larvae; spawning success, adult fecundity, and
larval mortality patterns. Can. J. Zool. 82: 1436-1443.
Stoeckmann, A. 2003. Physiological energetics of Lake Erie dreissenid mussels: a basis
for the displacement of Dreissena polymorpha by Dreissena bugensis. Can. J. Fish.
Aquat. Sci. 60: 126-134.
Styan, C.A., Kupriyanova, E., and Havenhand, J.N. 2008. Barriers to cross-fertilization
between populations of a widely dispersed polychaete species are unlikely to have
arisen through gametic compatibility arns-races. Evol. 62: 3041-3055.
Wright, D.A., Setzler-Hamilton, E.M., Magee, J.A., Kennedy, V.S., and Mclninch, S.P.
1996. Effect of salinity and temperature on survival and development of young zebra
{Dreissenapolymorpha) and quagga {Dreissena bugensis) mussels. Estuaries 19:
619-628.
48 Yund, P.O. and Meidel, S.K. 2003. Sea urchin spawning in the benthic boundary layer:
are eggs fertilized before advecting away from females? Limnol. Oceanogr. 48: 795-
801.
Yund, P.O., Murdock, K. and Johnson, S.L. 2007. Spatial distribution of ascidian sperm:
two-dimensional patterns and short vs. time-integrated assays. Mar. Ecol. Prog. Ser.
341:103-109.
49 100 H c CD E Q. JO
CD T3
(/) CO CD O O
C/) O c g o o. o
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 24.0 Time post fertilization (h) Corresponding stages of development Oh 0.5 h 1.0 h 2.0 h 3.0 h 24 h
•***rt ma 'V- '->* i ^t* * ^^ „ ' eat \ , '^- W*'* 60 um 80 urn Figure 1.1. (A) Results from the polyspermy study on dreissenid fertilization. Freshly spawned eggs and sperm were mixed and monitored over a 24 h period, with the proportion of successfully developing zygotes/embryos recorded at each time interval. (B) The stages of zygote, development at; 0 h - unfertilized egg; 0.5 - fertilization envelope; 1.0 h - 2-cell stage; 2.0 h - 4-cell stage; 3.0 h - 8-cell stage; and at 24 h - trochophore larval stage.
50 100 D. polymorpha D. bugensis 80
60 H
40 H
20 H
10"1 10° 101 102 103 104 105 106 107 108
Sperm concentration (cells ml )
Figure 1.2. Fertilization success as a function of sperm concentration for Dreissena polymorpha and Dreissena bugensis. Fertilization success increases relative to the increase in sperm concentration. Plots are means ± SE, N D. polymorpha = 8, N DMgensts = 4. There was no significant difference between species.
51 SO o D. polymorphs 96 - • • D. bugensis o R2 = 0.067 94 - O • • P = 0.47 N = 10 (J) o o • CD O O 92 - o • o^-^""" ^ • o ^^ c 2 .o 90 - o R = 0.06 P = 0.34 N o -Q______N = 17 tE CD 88 - o • LL o o o 86 - • O •
84 - 1 1 1 I I I i 2000 2500 3000 3500 4000 Egg concentration (eggs ml"1)
Figure 1.3. Fertilization success as a function of egg concentration for Dreissena polymorpha and Dreissena bugensis. There was no significant relationship between fertilization success and egg concentration for either species.
52 100 O D. polymorphs 7 10 # D. bugensis 105 80
60 10"
40
10J
20 101
—i— 0 20 40 60 80 100 120 140 Time (min)
Figure 1.4. Fertilization success in Dreissena polymorpha and Dreissena bugensis as a function of time post spawning for five sperm concentrations. Regardless of concentration, there is a decrease in fertilization success as sperm age. Values are means ± oli. IN D polymorpha O, IN T) bugensis ->.
53 A 1
™ rr„.™ ..
- — •- ---'- -'- --.- —^_--^ .- - -• -=. - .---.. •• •-= -
.5 j...... : " — -. —- —_--.-;-- r - - - — ' " '—••--
------_ _ • ' ... . — - --_---_- - - . '. •_ — _
E= r- - - --'--•.• ••'- ••.- -~---V.:.^;- —•-_—--^" --^-r-rr- • — --F-z-^<--y ~*-\ - --^" "v- ~->^" S•••':.---- n - ,^.--.— ..,•. —— —- - ^.y i'. .- ,:;•{ \c ' •- 2.5
B
12.5 25 Concentration % 100%
0.5 80%
60% 2.5 D J 40%
120%
10% 0 12.5 25 Distance downstream (m)
Figure 1.5. Results from COMSOL modeling for the mussel cluster scenario (A) and coral head scenario (B). Flow is left to right, with lines representing velocity streamlines and areas of recirculation are rotating in a clockwise direction. Areas of recirculation are indicated by the presence of vortices, examples of which can be seen behind the individual mussel and reef elements. Scalar (1 mol ml"1) which was released from the downstream slope of the first mussel cluster (C) or coral head (D) is retained in the recirculation zone or transported downstream. Note the different spatial scales for the two scenarios. White line in panel D represents an example of a transect used for scalar concentration measurements.
54 70 o HRel mussel 60 o HRel coral
c 50 O *•+-« 03 •+-» § 40 c o o -2 30 H o c/3
20
Downstream transport Upstream retention 10 0.0 0.2 0.4 0.6 0.8 1.0
Roughness (HRel)
Figure 1.6. The effect of modeled roughness element height (HRei = HR/Z) on the upstream retention vs. downstream transport of scalar to the next element (i.e., downstream of the second roughness element). The dotted line indicates the point at which 50% of released scalar switches from downstream transport to upstream retention. Note that the mussel cluster and coral head element fit the curve as well. Values are means ± SE.
55 100
Distance downstream (XRel)
Figure 1.7. The influence of bathymetry on modeled fertilization success downstream. The measured fertilization rate from Fig. 1 and the modeled scalar concentration downstream was used to calculate fertilization success. The vertical line indicates the location of the second roughness element. Distance downstream was normalized to the element diameter, where the point of scalar release was equal to 1 (i.e., from the first cluster/coral head). Values are means ± SE.
56 100 O D. polymorpha - this study • D. bugensis - this study v Strongylocentrotus franciscanus 80 T S. droebachiensis • Haliotis laevigata • Montipora digitata w O Platygyra sinensis (D o 60 - o 3 W c o '•4—» 03 40 N
20
5— O £r 1 .... i i 0 10-1 10° 101 102 103 104 105 106 107 108 Sperm concentration (cells ml"1)
Figure 1.8. A comparison of fertilization success as function of sperm concentration for seven broadcast spawning species. The two dreissenid species, represented by the circles, display much higher fertilization success than the other five species for sperm concentrations less than 10 sperm ml" . The three species Montipora digitata, Haliotis laevigata, and Platygyra sinensis are affected by polyspermy at high sperm concentrations, indicated by a drop in fertilization at 10 sperm ml" . Data from Pennington (1985); Levitan et al. (1991); Oliver and Babcock (1992); and Babcock and Kessing (1999). Values are means ± SE. Differences are significant (xs = 11.1, P < 0.01).
57 O Dreissena polymorpha • D. bugensis V Acanthaster planci 6 - T Haliotis laevigata • H. tiiberculata • Strongylocentrotus droebachiensis O S. franciscanus 5 - • S. purpuratus A Cerastoderma edule
4 - .9
03 3 -
Q. 2 -
1 - " ^-^ ^^^ /
0 - 1 T 1 1 1 1 1
Log sperm concentration (cells ml" )
Figure 1.9. A comparison of sperm half-life in hours as function of sperm concentration for nine broadcast spawning species. The sperm half-life is defined as the time it takes to reach 50% of the maximal fertilization rate achieved at any given initial sperm concentration. The two dreissenid species, represented by the circles, display a sperm half-life curve with a low slope indicating a uniform sperm decay rate regardless of sperm concentration, suggesting more potent sperm at lower concentrations. Data from Levitan (1993); Benzie and Dixon (1994); Andre and Lindegarth (1995); Babcock and Kessing (1999); and Baker and Tyler (2001).
58 CHAPTER 2: The effect of near-bed turbulence on sperm dilution and fertilization success of broadcast spawning bivalves.
Abstract
Hydrodynamic forces near the bed have been posited to influence external fertilization in broadcast spawning benthic invertebrates. This study examined how the fertilization success of two freshwater organisms (Dreissena polymorpha and D. bugensis) in laboratory flow chambers and in the field (Lake Erie) is influenced by water velocity and turbulence near the bed as influenced by bed roughness. Experimental results demonstrate that fertilization success decreased with velocity and downstream distance, indicating that dreissenid mussels are subject to sperm limitation. Detailed laser particle imaging velocimetry (PIV) measurements in the laboratory and acoustic Doppler velocimetry (ADV) in the field were used to determine the nature of small-scale turbulence in the near-bed environment, as indicated by turbulent diffusivity, Reynolds shear stress, bed shear stress, turbulent diffusivity, and the frequency of sweeps and ejections. Ejections displayed positive association with fertilization success and were more frequent in half-cylinder and patchy mussel bottom configurations. Both of these bed roughness conditions generated recirculation zones that retained sperm packets and stretched them, thus enhancing sperm-egg encounter and reducing sperm dilution.
Reynolds shear stress displayed positive associations with fertilization success at low velocities but negative associations at high velocity, suggesting that an intermediate level of turbulence provides the highest fertilization success. The connection between turbulence and bed roughness was consistent in the laboratory and field highlighting the importance of roughness to sperm transport and fertilization success.
59 Introduction
External fertilization via broadcast spawning is a key life-history strategy used by many sessile benthic organisms (Pennington 1985; Levitan 1993; Abelson and Denny 1997).
The process is influenced by near-bed hydrodynamics, as sperm swimming speeds are typically much lower than ambient water velocities (Denny and Shibata 1989; Abelson and Denny 1997). Successful fertilization is difficult to predict in dynamic flow environments for two main reasons: (1) gametes are diluted rapidly in turbulent flows; and (2) the lifespan of gametes is limited (Levitan 1993). Fertilization success is also dependent on the characteristics of the gametes and associated fluids (e.g., viscosity and density; Thomas 1994), the organism (e.g., egg brooding ascidian Diplosoma listerianum;
Bishop 1998), and the population as a whole (e.g., population density and size; Levitan and Young 1995). Given the short life spans and slow swimming speeds of sperm, the probability of sperm-egg encounters is extremely low (Allee 1931; Mortensen 1938).
Thus sperm dilution has been considered to be one of the most important limitations to broadcast spawners (i.e., 'sperm limited population dynamics'; Levitan and Petersen
1995) especially when water currents dilute sperm and eggs (Denny and Shibata 1989).
Many organisms have evolved strategies that increase fertilization success. They are largely related to increasing the probability of gamete encounter through tactics such as synchronous spawning of males and females (Levitan 2005), modified gamete release in a viscous plume seen in sea urchins (Yund and Meidel 2003), and selective filtration of sperm from the water by female ascidians that brood their eggs (Bishop 1998). Denny and Shibata (1989) found through mathematical modeling that the fraction of fertilized eggs decreased with: (1) an increase in mean velocity; (2) the distance between males and
60 females; and (3) the spatial and temporal variation in the eddy diffusivity (K; a measure of turbulent mixing) in near-bottom flows. Variations in K are caused in part by turbulent events such as 'sweeps' characterized by high-momentum fluid moving towards the bed, and 'ejections' characterized by high-momentum fluid moving away from the bed
(O'Connor and Hondzo 2008). These types of instantaneous velocity fluctuations are caused by changes in substrate roughness, depressions, or dunes (Cellino & Lemmin
2004). Theoretically, ejections are expected to play a critical role in entraining gametes into the water column, whereas sweeps are considered to be important for larval settlement (Abelson and Denny 1997). Moreover, turbulent mixing in the water column is also thought to be an important factor as it increases the encounter rates between eggs and sperm (Denny and Shibata 1989).
Beyond measurements of K, the nature of small-scale turbulence at organismal spatial scales are not well understood. Some indications are provided, however, by the local bed shear stress, % (i.e. the drag exerted by the bed on the flow on an areal basis;
Ackerman and Hoover 2001), and the Reynolds shear stresses (r) in the water column
(i.e., an indication of the turbulent fluctuations; Mann and Lazier 2006). Importantly, roughness features on the bottom have been found to affect both of these physical parameters, and Tb has been predicted to be the most important factor affecting larval settlement (Mead and Denny 1995; Crimaldi et al. 2002).
Exotic dreissenids are unique among freshwater mussels in that they have a conservative life history much like marine bivalves, with external fertilization via broadcast spawning of sperm and eggs into the water column and dispersal via planktotrophic larvae (Ackerman et al. 1994). These life history strategies and their
61 widespread occurrence in the Laurentian Great Lakes makes them ideal model organisms to study external fertilization dynamics. Whether broadcast spawning freshwater benthic organisms are also exposed to turbulent features seen in marine systems remains to be determined, given differences in the physical forcing in marine vs. freshwater systems
(Nixon 1988). Given that the turbulence created by bathymetric features in the near-bed environment has been predicted to affect the downstream transport and dilution of sperm
(Chapter 1), it is reasonable to examine these relationships empirically in the laboratory and field. The purpose of this study is to determine how near-bed fluid flow and turbulence affects downstream transport of sperm and hence fertilization success in dreissenid mussels.
Materials and methods
Laboratory - Dreissena polymorpha (Pallas, 1771) and Dreissena bugensis
Andrusov, 1897 were obtained from sites in Lake Ontario (Hamilton Harbour: D. polymorpha and D. bugensis) and Lake Erie (Evans Point and Selkirk Provincial Park: D. bugensis only), and were maintained in aerated 10-L aquaria in the Hagen Aqualab,
University of Guelph. Tanks were kept in a climate-controlled room at 13°C with a 12-h light:dark cycle, and the aquarium water was changed with fresh well water weekly.
Mussels were fed 30 ml of a prepared mixture of nanoalgae and shellfish diet (Reed
Mariculture Inc., CA) twice a week. Male and female mussels were induced to spawn by placing them in a bath of 10"3 M serotonin solution for 30 min, and then transferring them to a beaker with 20 ml of aquarium water (Ram et al. 1993). Time until spawning ranged from 15 min to 4 h, with females typically spawning later than males. Once a mussel
62 began to spawn, its sex was recorded, and the gametes were immediately removed from the beaker to determine initial gamete concentrations with four replicate hemocytometer counts. Experiments were performed at 21-25°C to maximize fertilization (Ram et al.
1993).
Experiments were undertaken in a recirculating flow chamber (10 x 10 x 100 cm long; see Nishihara and Ackerman 2006 for a detailed description), that consisted of flow straighteners in the upstream 12 cm, a 30-cm region for flow development, and a 50-cm test section. The chamber was operated at average velocities (U) of 1.5, 3.5,4.6, and 5.1 cm s"1 determined by volumetric flow measurements. These conditions corresponded to turbulent Reynolds numbers for open-channel flows (Rejh = U d^/ v where d^ = hydraulic diameter and v = kinematic viscosity) ranging from 1,499 to 5,037.
A series of sperm-transport experiments were performed in the flow chamber at each velocity to determine the effect of velocity and downstream distance from the spawning source on sperm limitation and fertilization success. A 0 cm s" flow condition was used to determine downstream sperm transport and sperm dilution due to exhalation from the siphon. In the case of the velocity experiments, 10 ml of sperm was released from a syringe 2 cm above the bottom at a rate consistent with observed siphon ejections
(e.g., 2 cm s"1; Bunt et al. 1993) at a downstream distance x = 0 (i.e., start of the test section), with the addition of 0.05 ml of fluorescein dye, which did not affect dilution or fertilization success.
The sperm cloud was sampled downstream using a pipette at x = 10, 20, 30, 40, and 50 cm based on the timed interval determined from the mean velocity at which peak concentrations were expected. At each location, a 2-ml sample was taken 2 cm above the
63 bottom through a guide plate fitted over the top of the flow chamber to standardize the sampling location. Samples were added to 2 ml of freshly spawned eggs and incubated for 60 min, after which the samples were fixed with 2% KC1. Sperm concentrations/dilutions were determined through four replicate hemocytometer counts.
Fertilization rates (%) were determined by counting the number of elevated fertilization envelopes on the first 200 eggs encountered under a compound microscope. Although not counted, early cleavage patterns were apparent for the majority of successfully fertilized eggs encountered, indicating that polyspermy is not a major issue for this species. This observation supports Sprung's (1993) assertion regarding the existence of blocks to polyspermy in D. polymorpha.
Two controls were used: (1) a control that contained spawned eggs to check for background sperm in the original egg sample; and (2) a control that contained spawned sperm to check for background eggs in the original sperm sample. In addition, a sample of freshly spawned sperm was added to 2 ml of eggs to determine maximal fertilization success. Six replicates (each consisting of a different male and different female mussel) were conducted for each of the five velocities using D. polymorpha for a total of 30 trials.
Three replicates were also conducted at the 0 and 5.1 cm s"1 velocities using D. bugensis to determine whether there were differences between mussel species.
Three turbulence-generating conditions were also examined: (1) grid turbulence - a 1 cm x 1 cm wire-mesh (0.1 cm diameter thickness) placed perpendicular to the direction of flow at (a) x = 0 cm and (b) x = -5 cm (i.e., 5 cm upstream of the test section); (2) cylinder turbulence - a half-cylinder with a 1.5 cm radius placed horizontally on the bottom of the flume at (a) x = -5 cm and (b) x = -15 cm; and (3) bed
64 roughness - mussel shells attached to the bottom of the flow chamber with Tac 'N Stik
(Elmer's Products Inc., Columbus OH, USA) in (a) low (860 mussels m"2) and (b) high
(1700 mussels m"2) densities over the 50 cm long test section, as well as (c) three high- density, 10-cm cluster "patches" placed 10 cm apart from each other {see Fig. 2.1). Each turbulence condition was examined at U= 1.5 and 5 cm s" , and involved the same sperm release and sampling technique described above. Three replicates (each a different male- female mussel pair) were conducted for each of the turbulence conditions at each velocity, providing a total of 42 trials.
Particle-imaging velocimetry (PIV) - PIV measurements were undertaken for each of the velocity and turbulence-generating flow conditions {see Nishihara and
Ackerman 2006). The laser sheet (670 nm, 20° beam angle) intersected the flow chamber in the mid-longitudinal plane, with a plane length of 13 cm. Five 13-cm long PIV measurements were taken along the 50-cm test section, with a 1.5-cm overlap at each end. Images of silver-coated, hollow glass beads (13 jum diameter) were captured at a frequency of 30 Hz through the side wall of the flow chamber. A total of 150 frames were processed using a 768 X 481 pixel grid with an 80 X 80 pixel interrogation window.
The PIV data generated a series of instantaneous u and w values in the stream wise (x) and vertical (z) directions, respectively, for each 13-cm test section. Mean velocities in the horizontal and vertical directions {u and w) in each frame were calculated and used to determine the magnitude of the fluctuations u' and w' from the mean velocities (i.e., u = u + u' and similarly for w). Products u'w' were classified using quadrant analysis to determine sweeps (quadrant 4; i.e., u 'w' < 0 where w' < 0) and ejections (quadrant 2; i.e., u 'w' < 0 where w' > 0), as well as inward (quadrant 1; u 'w' > 0 where w' > 0 ) and
65 outward interactions (quadrant 3; u 'w' > 0 where w' < 0 ). Each quadrant event was compared in a hyperbolic hole region created in the u' and w' plane, with only quadrant events outside of a threshold hyperbolic hole size (H) of 2 termed significant and used in subsequent analysis (where H = \u'wy \u'w'\; Lu and Willmarth 1973). Quadrant analysis has traditionally been applied to temporal fluctuations of u and w, but recent studies have proven successful in the identification of persistent spatial features in velocity fields
(Pokrajac et al. 2007), similar to those in the current study. PIV data were also used to determine shear velocities («*), calculated by multiplying the von Karman constant (K =
0.41) by the slope of the velocity versus ln(z) in the logarithmic portion of the boundary layer (e.g., Ackerman and Hoover 2001). Reynolds shear stress (r), or fluid stress within the water column, was calculated from the PIV data via
r = -pu'w' (1) where p is density of water.
Field - Experiments were conducted in the eastern basin of Lake Erie at Evans
Point (42° 50' N, 79° 46' W; Fig. 2.2) between July 25 and September 28, 2007. Four sites within ~50 m of each other and at a depth of 40 cm with distinct bathymetries were used, corresponding roughly to laboratory conditions: (1) flat bottom - a large section of bedrock without mussels; (2) mussel beds of low density (600 ± 26 mussels m"); (3) high mussel density (1500 ± 64 mussels m"); and (4) patches of mussels at high density.
Mussel densities were determined through four replicate 25 x 25 cm quadrant counts at each location. Only Dreissena bugensis were present in the area. A bed profiler consisting of 50-cm long pins oriented vertically and spaced 2.5 cm apart in a 1-m long frame was used to profile the bed roughness at each site along a 2-m transect. Two
66 transects spaced 5 cm apart were taken on either side of the central transect. Digital photos were taken of each profile measurement for analysis, where the top of each pin provided an indication of the depth.
An apparatus was used to release and sample sperm from mussels induced to spawn on site using methods described above. It consisted of 4-cm diameter plastic plumbing pipe with 2-ml polystyrene pipettes (0.5 cm diameter and 25 cm long) inserted through holes at x = 10, 20, 30, 40, 50, 100, 200, and 300 cm in the pipe. The height of the apparatus was adjusted before each trial so that all pipettes would sample 2 cm above the lake bed. Each pipette was connected to 5-mm diameter plastic tubing that ran along the inside of the pipe to an exit port at the midpoint of the plumbing pipe, and from there to a 10-cc syringe held above the surface by the researcher. A separate 2-ml pipette was used to release 10 ml of fresh sperm and 0.05 ml of fluorescein dye 2 cm above the bed at
JC = 0.
Before each trial, velocity measurements and current direction were taken using a propeller velocimeter (Model 2100, Swoffer Instruments Inc., Seattle) at 40% of the flow depth and with the timed release of dye. Mean velocity was used to establish sperm transit time along the transect and to determine sampling times. Trials were undertaken in the morning between 10:00 and 12:00 (local time) to restrict the influence of oscillatory flow caused by surface waves that were strongest in the afternoon. Trials were undertaken during periods of unidirectional flows and were suspended when the flow oscillated, as indicated by transport of the dye (i.e., -15% of the time). After a predetermined interval, 10 ml of water was sampled at a given distance using the
67 pipette/syringe at that location to sample the released sperm plume as it travelled along the length of the sampling apparatus.
After sampling, 2 ml of each 10-ml sample was added into one of 8 test tubes
(corresponding to the sampling locations) containing 2 ml freshly spawned eggs from one female. The sperm-egg sample was allowed to incubate for 60 min, after which the samples were fixed with 2% KC1. For each of the four sites, 10-12 replicates (each replicate consisting of a different male and different female mussel) were undertaken over a range of velocities comparable to laboratory conditions (i.e., 1 - 5 cm s"1), for a total of 43 trials. Two sperm and egg controls were also used, as well as a sample of fresh sperm to determine maximal fertilization success for each trial. Preliminary results indicated that sperm detection via the sampling technique used was limited to 2 m downstream of the release point (i.e., x = 0).
Acoustic Doppler velocimetry (ADV) -ADV vertical profiles were taken at each site along the first 50 cm of the central transect at each of the sampling locations using a vertical, hand-held stand (e.g., Lane et al. 1998). The ADV was used to acquire vertical profiles starting as close to the bed as possible (within 2 cm) and ending just below the surface, giving from 5 to 7 points per profile (each point sampled at 25 Hz for 2 min).
These data were analyzed using quadrant analysis and used to calculate shear velocities
(w*) in a similar manner to that of the PIV data in the laboratory. Shear velocity was used to calculate bed shear stress (r*), a measure of stress at the bed, using
Tb = put (2)
In addition, u* was also used to calculate depth-averaged turbulent diffusivity (Kavg) given by
68 K U H 3 a*g=T > ( ) o where H = 0.4 m is depth (Edwards et al. 2005), which is the mixing intensity due to shear on the bed, the formulation based on a modification of the Prandtl's mixing length model (Fischer et al. 1979).
Results
Laboratory flow chamber - The maximum measured distance that sperm traveled under the no-flow condition, where advection was due only to siphonal ejection, was 40 cm
(Fig. 2.3). Fertilization success decreased with increasing velocity and downstream distance in the flow chamber (Fig. 2.3). Statistically significant differences among velocities over the first 20 cm were small, but by x = 30 cm notable differences were observed between low (1.5 - 3.5 cms") vs. high (4.6 - 5.1 cms") velocities. No significant differences were detected between species with respect to fertilization success and distance, as indicated by separate two-way ANOVAs (species and distance) on a set of six trials for each dreissenid species at both the 0 cm s"1 (Fi,i6 = 0.13, P = 0.95) and 5.1 cms"1 velocities (Fi^ = 0.38, P = 0.72). No significant interactions were found. A strong, significant, positive association was found between fertilization success and sperm dilution (R = 0.99, P < 0.0001), determined through a linear regression of the log- transformed sperm concentrations on fertilization success with data from all experiments
(Fig. 2.4).
Similar patterns of decreased fertilization success with increased velocity and downstream distance were observed under the various turbulence-generating conditions
(Fig. 2.5). Significant differences in fertilization success were found with respect to turbulence condition (including data from flat bottom in Fig. 2.3), velocity, and distance
69 using a three-way ANCOVA, with distance as the covariate (F7,63 = 9.8, P < 0.0001; Fi_63
= 103, P < 0.0001; and Fi,63 = 500, P < 0.0001, respectively). A significant two-way interaction was also found between turbulence condition and velocity (F7,63 = 2.4, P <
0.04). Significant differences in fertilization success were also found with respect to turbulence condition, velocity, and distance using a three-way ANOVA (F7J60 = 102, P <
0.0001; Fi.ieo = 900, P < 0.0001; and F4,i6o = 1240, P < 0.0001, respectively). Significant two-way interactions were also found between all factors (P < 0.0001 in all cases), and one-way ANOVAs were used to determine how fertilization success differed among conditions at each downstream location (10, 20, 30, 40, and 50 cm) and velocity (1.5 and
5 cms"1).
In general, fertilization was significantly higher for the flat bottom vs. all other turbulence-generating conditions. Several patterns were evident at low velocity: (1) there was significantly higher success for grid turbulence at 0 cm vs. -5 cm for x = 10 to 30 cm;
(2) significantly higher fertilization success in low-density mussels vs. high density and patch configurations for x = 10 to 50 cm; and (3) there was significantly higher fertilization success for the grid-generated turbulence versus the other turbulence conditions, with the exception of the low-density mussel configuration. Patterns in fertilization success were more similar at the high velocity with fewer significant pairwise differences: (1) fertilization success for the grid at x = -5 cm was significantly greater than the high and patch mussel configurations and the cylinder at x = -5 cm; and
(2) as in the case of low velocity, fertilization success was higher at the low mussel density vs. the high density and patch configurations.
70 PIV was used to evaluate the chamber flow environment and specially the small- scale turbulent structures generated by each turbulence configuration (see Fig. 2.6).
Dimensionless velocity measurements from PIV data in the flat-bottom configuration generated velocity profiles that indicated reasonably well behaved flow and well- developed boundary layers, especially for the higher-velocity trials (Fig. 2.7A). The high-density mussel cluster provided an example of skimming flow, as indicated by the relatively fast flow across the top of the cluster, and low flow between the individual mussels (Fig. 2.6A). In the open-patch configuration there was a transition from skimming flow to a recirculation zone, indicated by slow flow and an eddy-like pattern in the velocity vectors downstream of the mussel cluster atx = 3 cm (Fig. 2.6B). The half- cylinder created a flow field somewhat similar to the open patch, with acceleration over the cylinder and a large region of recirculation at x = 4 cm close to the bed downstream of the cylinder, as indicated by the pattern of the vectors (Fig. 2.6C). This recirculation zone persisted downstream, with evidence of two smaller eddies centered at x = 8 and 14 cm. There were numerous small eddies immediately downstream of the wire-mesh placed at x = 0 cm, as indicated by the varied direction of the vectors. They persisted to x
= 7 cm downstream, where the flow appeared well behaved (Fig. 2.6D).
Small-scale structures in the flow field were classified by quadrant analysis to identify turbulent ejections (Q2; u 'w' < 0 where w' > 0) and sweeps (Q4; u 'w' < 0 where w' < 0) as well as inward (Ql; u 'w' > 0 where w' > 0 ) and outward interactions (Q3; u 'w' > 0 where w' < 0 ). The spatial frequency of these events provides an indication of the nature of the turbulence in the flow field under the different configurations (bars in
Fig. 2.5). In general, the frequency of Q2 and Q4 events were higher for the cylinder and
71 grid-generated turbulence versus the mussel configurations, which experienced skimming flow. Sweeps tended to be more frequent at downstream distances x > 30 cm with some evidence of higher ejection frequency for the high velocity at x < 30 cm. Significant differences were not found for Ql and Q2 events among the turbulence-generating conditions, although they were found using a one-way ANOVA for Q3 and Q4 events
(F772 = 3.7, P < 0.002; F772 = 3.5, P < 0.003, respectively). Pairwise differences were detected predominately between high-density mussel and the other configurations.
The PIV velocity profile data (see Fig. 2.7A) were also used to calculate the shear velocity («*) at 10-cm intervals in the downstream direction for each of the turbulence generation conditions (R ranged from 0.64 to 0.96; P < 0.01 in all cases). Little difference was seen in u* among the turbulence configurations at the low velocity, which had a mean u* of 0.24 ± 0.01 cm s"1, corresponding to a bed shear stress (T^) of 0.063 ±
0.006 Pa and a turbulent diffusivity (K) of 9.7 ± 0.42 x 10" m s" , using equations (2) and
(3). The bed shear stresses (xb) determined from these u* values follow a similar pattern with respect to downstream distance but are about an order of magnitude higher than the theoretical laminar and turbulent values (using eqs. 7.25 and 7.44, respectively, in White
1994; Fig. 2.7B). The Reynolds shear stress (r) based on equation (1) was 0.003 ± 0.001
Pa. There were, however, some differences at the high velocity: (1) significantly lower u* and K in the flat versus the turbulent configurations (w*: 0.48 ± 0.02 vs. 0.62 ± 0.01 cm s"
5 5 2 1 '; F7,39 = 7.91, P < 0.0001; and K: 1.92 ± 0.07 x 10" vs. 2.47 ± 0.10 x 10" m s" ; F7;39 =
7.91, P < 0.0001) and (2) significant differences were detected among the three mussel configurations for w*(low: 0.57 ± 0.03 cm s"1; high: 0.65 ± 0.02 cm s"1; and patch: 0.53 ±
1 5 2 1 0.03 cm s" ; F2,14 = 4.33, P < 0.04) and K (low: 2.3 ± 0.2 x 10" m s" ; high: 2.60 ± 0.09 x
72 5 2 1 5 2 1 10" m s" ; and patch: 2.1 ± 0.1 x 10" m s" ; F2, H = 4.33, P < 0.04); T was also significantly higher in the mussel patch compared to all other conditions (0.09 ± 0.03 vs.
0.000 ± 0.005 Pa; F7,39 = 4.79, P < 0.0001).
Association between fertilization success and physical parameters was examined with linear regression in the entire data set and subsets of the full data set to identify relationships within a particular downstream location, at a particular velocity, and under specific turbulence conditions. Downstream distance had the highest association with fertilization success, regardless of whether the entire data set was compared (R2 = 0.67, P
< 0.0001) or subsets of the data set were used (R ranged from 0.60 for grid at x = 0 cm to
0.82 for flat bottom, P < 0.001), and all associations were negative. Shear velocity and velocity had significant but small and negative associations with fertilization success when examined under the full data set (R =0.16 and 0.14, respectively). They were stronger when examined at specific downstream locations (e.g., R2 = 0.76 at x = 10 cm vs. 0.56 at x = 50 cm for u* and R2 = 0.62 at x = 10 cm vs. 0.57 at x = 50 cm for U).
Ejections (Q2 events) were not associated strongly with fertilization success (Rz = 0.05, P
= 0.04) in the full data set, but they provided the only positive significant correlation observed. However, Q2 events also had a significant positive association with
•y fertilization success when examined in the cylinder turbulence condition (R = 0.60, P =
0.008), as did Reynolds shear stress at 1.5 cms"1 (R2= 0.13; P = 0.025).
Field experiments - Fertilization success decreased with distance downstream in the field under both low (1.75 cm s"1) and high (5.4 cm s"1) velocities (Fig. 2.8). There was a much larger reduction in fertilization success within the first 10 cm downstream in the field (i.e., 71.0 ± 0.7 %) compared to the laboratory (28 ± 1 %). Fertilization success
73 was significantly lower for the high mussel density and significantly higher for the flat bottom when examined under high velocity. A three-way ANOVA revealed significant differences in fertilization success with respect to bottom configuration, velocity, and downstream distance (F3,245 = 19.4, P < 0.0001; FU245 = 5.49, P < 0.05; and F6j245 = 498,
P < 0.0001, respectively). There were significant two-way interactions for velocity x distance (F6,245 = 6.19, P < 0.000.1) and velocity x configuration (Fi8>245 = 1-79, P < 0.05).
Subsequent one-way ANOVAs revealed significantly higher fertilization success at low velocity forx < 20 cm (atx = 10 cm: Fi;42 = 5.22, P < 0.05; atx = 20 cm: Fi,42 = 6.77, P <
0.05) and significantly higher fertilization values at high velocity for x = 100 cm (F]i42 =
9.92, P < 0.005). No effect of species was examined as all mussels collected on site were identified as Dreissena bugensis.
ADV was used to acquire the vertical flow profiles at various downstream locations over the four roughness configurations (Fig. 2.9). In general, the velocity increased with distance from the bottom and many of the profiles, especially at high velocity, were approximately logarithmic. It was also evident that some of the profiles were affected in the near-bed region by the roughness elements upstream (e.g., Fig 2.9B,
C). These types of differences in profile shape will lead to differences in shear velocity
(see below). Areas of recirculation were also apparent at high velocity, particularly in the patch configuration, as indicated by the regions of negative velocity (i.e., flow in the opposite direction) observed close to the bed at x = 10 and 50 cm (Fig. 2.9D).
The frequency of ejections (Q2) and sweeps (Q4) were calculated over a 50-cm transect for each bottom configuration, and compared to fertilization success (Fig. 2.10).
At low velocity, the mussel patch configuration had the highest frequencies of both Q2
74 and Q4 events, followed by the low-density mussel configuration, and the high-density mussel configuration. The differences in Q2 and Q4 frequency among configurations decreased at high velocity, but the patch configuration still had the highest frequencies of
Q2 and Q4 events and the high-density mussel configuration had the lowest. Significant differences in configuration with respect to all turbulent events were found under one way ANOVA (F3J91 ranged from 41.8 to 69.0, P < 0.0001 in all cases), and all pairwise comparisons were significant with the exception of the flat and high density conditions.
Bed shear stress (rj determined eqn. 2; R ranged from 0.62 to 0.96; P < 0.01 in all cases) and Reynolds shear stress (r determined from eqn. 1) were calculated at each
10-cm interval downstream within each configuration (Fig. 2.11). In the flat bed, zj, field values were similar to tb values found in the laboratory flat-bottom condition (Fig 2.7B).
In general, Tb was greater than rfor both velocities (low U: 0.099 ± 0.004 vs. -0.11 ± 0.03
Pa; high U: 0.54 ± 0.01 vs. 0.05 ± 0.02 Pa), and was more uniform over distance. A tripling of velocity corresponded to similar increase in Tt within the mussel configurations. At high U, the mussel roughness configurations had significantly higher
Tb than the flat bed (one-way ANOVA: F3, u = 12.7, P < 0.0005); r appeared to be somewhat less variable at high U, and for both velocities was more variable at x = 10 cm, but these differences were not significant. Shear velocities used to determine bed shear stress were also used to calculate K (Fig. 2.9). Of the four configurations, the high mussel-density configuration had the highest mean lvalues at both velocities (high: 2.15 x 10"4 m2 s"1; low: 9.65 x 10"5 m2 s"1), followed by the low mussel density and patch configurations, with the flat configuration the lowest (high: 1.40 x 10"4 m2 s"1; low: 5 49 x
10"5 m s"1). Significant differences in-between the mussel roughness configurations vs.
75 the flat bed were consistent with those found in Tb mentioned above (one-way ANOVA:
F3,14= 12.7, P< 0.0005).
Discussion
Velocity and fertilization success - The relationship between velocity and fertilization success has been studied before, but in these cases velocity has been treated either theoretically (Denny and Shibata 1989; Lauzon-Guay and Scheibling 2007) or in a non-experimental manner using simple measures of U (Pennington 1985; Levitan et al.
1992). For example, Denny and Shibata's (1989) fertilization model predicted that the fraction of eggs fertilized would decrease with increasing u, as shear would disperse particles relative to one another thus increasing the rate of dilution. Lauzon-Guay and
Scheibling (2007) incorporated distance and population size into this type of model, which predicted fertilization would decrease with velocity. For a given distance with sufficient population density, however, eggs would still have a high probability of fertilization. Field experiments revealed that fertilization rates of sea urchins were significantly higher at U < 0.2 m s"1 compared to > 0.2 m s"1 (Pennington 1985). Levitan et al. (1992) found similar changes at lower /7 (0.2 to 4.7 cm s"1), based on the tracking of water-filled bags 10 to 50 cm above the bed. None of these studies, however, manipulated U experimentally nor did they provide the detailed examination of turbulence needed to assess the mechanisms affecting fertilization success above the bed.
To the best of our knowledge, this is the first study to show experimentally that fertilization success decreases with velocity and to show how it is affected by turbulence.
In this case, increased flow chamber velocity was found to increase sperm dilution and thus decrease fertilization success. Whereas turbulence can also affect fertilization
76 success through sperm-egg encounter and bonding ability, it is relevant to note that for sperm limitation to occur, fertilization success must decrease with sperm dilution and dilution should increase with increasing distance from the spawning source (Denny and
Shibata 1989; Levitan 1993). This was clearly the case for both D. polymorpha and D. bugensis in the. laboratory and for D. bugensis in the field, and thus we conclude that these dreissenid species are sperm limited. These results are also consistent with computational fluid dynamic models of sperm dilution and transport, which predicted an increase in dilution with distance (Chapter 1). These are the first freshwater broadcast spawners that have been demonstrated to be sperm limited.
Turbulence and fertilization success - As mentioned above, small-scale turbulent events also affected fertilization success. Specifically, ejections have been predicted to resuspend larvae off the bottom, and Reynolds shear stress (r) has been used as a proxy for these types of turbulent events (Crimaldi et al. 2002). In this analysis, we have used both quadrant events and r to describe the connection between turbulence and fertilization success. For example, flow through a wire mesh (grid) produced small eddies related to the mesh size, which resulted in a well-characterized turbulent flow field created by the interaction of the individual wire elements with the flow (Nowell and
Jumars 1984). These small turbulent eddies created by the mesh (Fig. 2.6D) produced
Reynolds shear stress (z) that reduced fertilization success. Specifically, grid turbulence was the only condition where there was a significant negative association between shear velocity and fertilization success. Shear between each of the small and more uniformly distributed turbulent eddies created by the mesh and their corresponding x would effectively break apart the released sperm packet at a higher rate compared to the other
77 turbulence conditions. Conversely, the half cylinder-generated turbulence was the only turbulence condition to show a significant positive association with fertilization success via ejection (Q2 frequency). Ejections effectively stretch the sperm packet in the downstream direction and increase encounter rates in the narrowed lateral plane.
Fluid shear has been shown to both enhance and reduce fertilization success in broadcast spawners (purple sea urchin Strongylocentrotus purpuratus and the red abalone
Haliotis rufescens), but these studies relied on Couette cells in the laboratory to generate fluid shear stress (Mead and Denny 1995; Riffel and Zimmer 2007). Specifically, low shear enhanced fertilization success by facilitating sperm-egg encounters, whereas high shear interfered with sperm-egg encounter and broke the bonds between sperm and eggs
(Mead and Denny 1995; Riffel and Zimmer 2007). In this study, Reynolds shear stress
(r) enhanced fertilization success at low velocity in the laboratory and field but reduced fertilization success at high velocity in the field. These results are consistent with the laboratory Couette cell results mentioned above, which is not surprising given that the ranges of rwere consistent to the Couette cell studies (median values: 0.089 to 0.45). It is reasonable to suggest that a similar mechanism is responsible in this study. When bed shear stress (r$) and Reynolds shear stress (r) were compared, however, bed shear stress had higher, more consistent values indicating that it was a better parameter to use in describing turbulence in the near-bed environment. Field and laboratory Tb values were comparable for the flat-bottom configuration (Fig 2.7B and 2.11), which indicates that fluid dynamic conditions were similar in the laboratory flow chamber and the field. This conclusion is also supported by the separation of bed shear stress values between the flat bottom configuration and the roughness around the mussels at high U. That such
78 separation was not apparent at low U suggests that a minimum velocity is needed before the effect of roughness on the bed extends into the water column.
It is also possible that fertilization success in the field was affected by the direction of flow. In this case, the lateral velocity (v) was on average 6.9 ± 2.3% of u based on the ADV measurements at z = 20 cm in the field. It is, therefore, possible that under lateral advection of the sperm packet it could be stretched, thus increasing its dilution and directing it away from the point of sampling, essentially adding a third dimension. Lateral stretching of the sperm packet would be more pronounced at lower U because at higher U, the sperm packet would be stretched predominately downstream. It is also likely that these types of directional effects would occur close to the roughness elements in the near-bed region.
Bottom roughness and fertilization success - Bottom roughness influenced the fluid dynamic environment of the near-bed region, which has been predicted to affect larval settlement and fertilization success (Crimaldi et al. 2002; Chapter 1). In this study, abrupt changes in bottom roughness (e.g., space between patches) led to downstream recirculation, which enhanced fertilization success. This pattern was seen in the laboratory half-cylinder and the patchy mussel configuration (see Fig. 2.6B and C) and in the field with the mussel patch (Fig. 2.9D). The increased resistance of the larger obstacle caused the flow over the half cylinder to separate from the bottom (Nowell and
Jumars 1984), resulting in a recirculating zone on the downstream side (Fig. 2.6C). With symmetrical obstacles (e.g., pebble clusters and dunes) this recirculating flow would not be permanent and would result in vortex shedding downstream, creating large eddies
(Best and Kostaschuk 2002; Lacey and Roy 2008). A sperm packet, however, could
79 become entrained in the eddy within the recirculation zone reducing dilution and promoting stretching, and therefore potential egg-sperm encounters. However, if it was entrained in the flow over the cylinder, it would be transported downstream in the skimming flow. The result is similar for ejections, which essentially make up one component of an eddy. This relationship between changes in bottom roughness and fertilization success provides a novel role for population patchiness and mussel-induced bathymetry (e.g., Chapter 1). Turbulent diffusivity values (AT) provide an indication of the overall mixing rate present at each study location, with eddies contributing to the mixing rate. For the current study, K values fell within the low range of typical values seen in lakes (10~4 to 10° mV; Wiiest and Lorke 2003), confirming that mixing occurred but at relatively low rates compared to other lake environments. This was expected, given the shallow water depths at the study site and the conditions at the time of day when the experiments were performed.
In conclusion, both water velocity and bottom roughness features have important affects on the spawning ecology of freshwater benthic organisms and can influence the extent of sperm limitation for a broadcast spawner. The type of bottom roughness features present will consequently determine the nature of turbulence seen near the bed, with ejections being beneficial to fertilization success. This bottom roughness-turbulence relationship may also explain the aggregating population structure of certain species. In addition to external fertilization, another critical life-history stage for sessile organisms is larval transport, which is also controlled by hydrodynamics. Future work should examine the relationship between these two life-history strategies and near-bed turbulence.
80 References
ABELSON, A., AND M. DENNY. 1997. Settlement of marine organisms in flow. Annu. Rev.
Ecol. Syst. 28:317-339.
ACKERMAN, J. D., AND T. M. HOOVER. 2001. Measurement of local bed shear stress using
a Preston-static tube. Limnol. Oceanogr. 46: 2080-2087.^
, B. SIM, S. J. NICHOLS, AND R. CLAUDI. 1994. A review of the early life history
of zebra mussels (Dreissena polymorpha): comparisons with marine bivalves. Can. J.
Zool. 72: 1169-1179.
ALLEE, W. C. 1931. Animal Aggregations: A Study in General Sociology. University of
Chicago Press, Chicago.
BEST, J. L., AND R. A. KOSTASCHUK. 2002. An experimental study of turbulent flow over
a low-angle dune. J. Geophys. Res. 107: 3135-3154.
BISHOP, J. D. D. 1998. Fertilization in the sea: are the hazards of broadcast spawning
avoided when free-spawned sperm fertilize retained eggs? Proc. R. Soc. London B
265:725-731.
BUNT, C. M., H. J. MACISSAC, AND W. G. SPRULES. 1993. Pumping rates and projected
filtering impacts of juvenile zebra mussels {Dreissena polymorpha) in western Lake
Erie. Can. J. Fish. Aquat. Sci. 50: 1017-1022.
CELLINO, M, AND U. LEMMIN. 2004. Influence of coherent flow structures on the
dynamics of suspended sediment transport in open-channel flow. J. Hydraul. Eng.
130: 1077-1088.
81 CRIMALDI J. P., J. K. THOMPSON, J. H. ROSMAN, R. J. LOWE, AND J. R. KOSEFF. 2002.
Hydrodynamics of larval settlement: the influence of turbulent stress events at
potential recruitment sites. Limnol. Oceanogr. 47: 1137-1151.
DENNY, M. W., AND M. SHIBATA. 1989. Consequences of surf-zone turbulence for
settlement and fertilization. Am. Nat. 134: 859-889.
EDWARDS, W. J., C. R. REHMANN, E. MCDONALD, AND D. A. CULVER. 2005. The impact
of a benthic filter feeder: limitations imposed by physical transport of algae to the
benthos. Can. J. Fish. Aquat. Sci. 62: 205-214.
FISCHER, H. B., E. J. LIST, R. C. Y. KOH, J. IMBERGER, AND N. H. BROOKS. 1979. Mixing
in inland and coastal waters. Academic Press, New York.
LACEY, R.W.J., AND A. G. ROY. 2008. The spatial characterization of turbulence around
large roughness elements in a gravel-bed river. Geomorphology 102: 542-553
LANE, S. N., P. M. BIRON, K. F. BRADBROOK, J. B. BUTLER, J. H. CHANDLER, M. D.
CROWELL, S. J. MCLELLAND, K. S. RICHARDS, AND A. G. ROY. 1998. Three-
dimensional measurement of river channel flow processes using acoustic Doppler
velocimetry. Earth Surf. Process. Landforms 23: 1247-1263.
LAUZON-GUAY, J.-S., AND R. E. SCHEIBLING. 2007. Importance of spatial population
characteristics on the fertilization rates of sea urchins. Biol. Bull. 212: 195-205.
LEVITAN, D. R. 1993. The importance of sperm limitation to the evolution of egg size in
marine invertebrates. Am. Nat. 141: 517-536.
. 2005. Sex-specific spawning behavior and its consequences in an external
fertilizer. Am. Nat. 165: 682-694.
, AND C. PETERSEN. 1995.Sperm limitation in the sea. Trends Ecol.Evol.lO:228 -231.
82 , AND C. M. YOUNG. 1995. Reproductive success in large populations: empirical
measures and theoretical predictions of fertilization in the sea biscuit Clypeaster
rosaceus. J. Exp. Mar. Biol. Ecol. 190: 221-241.
, M. A. SEWELL, AND F.-S. CHIA. 1993. How distribution and abundance influence
fertilization success in the sea urchin Strongylocentrotus franciscanus. Ecology 73:
248-254.
Lu, S.S. & WILLMARTH, W. 1973. Measurement of the Structure of the Reynolds Stress
in a Turbulent Boundary Layer. J. Fluid Mech. 60: 472-478.
MANN, K. H., AND J. R. N. LAZIER. 2006. Dynamics of marine ecosystems. Blackwell.
MEAD, K.S., AND M. W. DENNY. 1995. The effects of hydrodynamic shear stress on
fertilization and early development of the purple sea urchin Strongylocentrotus
purpuratus. Biol. Bull. 188: 46-56.
MORTENSEN, T. 1938. Contributions to the study of the development and larval form of
echinoids. Kong. Danske Vidensk. Selsk. Nat. Math. Afd. 9 raekke 4: 1-39.
NISHIHARA, G. N., AND J. D. ACKERMAN. 2006. The effects of hydrodynamics on the
mass transfer of dissolved inorganic carbon to the freshwater macrophyte Vallisneria
americana. Limnol. Oceanogr. 51: 2734-2745.
NIXON, S. W. 1988. Physical energy inputs and the comparative ecology of lake and
marine ecosystems. Limnol. Oceanogr. 33: 1005-1025.
NOWELL A. R. M., AND P. A. JUMARS. 1984. Flow environments of aquatic benthos.
Annu. Rev. Ecol. Syst. 15: 303-328.
O'CONNOR, B. L., AND M. HONDZO. 2008. Dissolved oxygen transfer to sediments by
sweep and eject motions in aquatic environments. Limnol. Oceanogr. 53: 566-578.
83 PENNINGTON, J. T. 1985. The ecology of fertilization of echinoid eggs: the consequences
of sperm dilution, adult aggregation, and synchronous spawning. Biol. Bull. 169:
417-430.
POKRAJAC, D., L. J. CAMPBELL, V. NIKORA, C. MANES, AND I. MCEWAN. 2007. Quadrant
analysis of persistent spatial velocity perturbations over square-bar roughness. Exp.
Fluids 42: 413-423.
RAM, J. L., G. W. CRAWFORD, J. U. WALKER, J. J. MOJARES, N. PATEL, P. P. FONG, AND K.
KYOZUKA. 1993. Spawning in the zebra mussel {Dreissena polymorpha): activation
by internal or external application of serotonin. J. Exp. Zool. 265: 587-598.
RIFFELL, J. A., AND R. K. ZIMMER. 2007. Sex and flow: the consequences of fluid shear
for sperm-egg interactions. J. Exp. Biol. 210: 3644-3660.
SPRUNG, M. 1993. The other life: an account of present knowledge of the larval phase of
Dreissena polymorpha. In Zebra mussels: biology, impacts, and control. Edited by
T.F. Nalepa and D. Schloesser. Lewis Publishers, Boca Raton, Fla. pp. 39-53.
THOMAS, F. I. M. 1994. Physical properties of gametes in three sea urchin species. J.
Exp. Biol. 194: 263-284.
rd WHITE, F. M. 1994. Fluid mechanics. 3 ed. McGraw-Hill, New York.
WUEST, A., AND A. LORKE. 2003. Small-scale hydrodynamics in lakes. Annu. Rev. Fluid
Mech. 35:373-412.
YUND, P. O., AND S. K. MEIDEL. 2003. Sea urchin spawning in the benthic boundary
layer: are eggs fertilized before advecting away from females? Limnol. Oceanogr.
48:795-801.
84 Figure Legends
Fig. 2.1. Examples of turbulence generating conditions: mussel shells attached to the flume bottom in (A) low mussel density (860 mussels m"2) configuration over the entire
50 cm sampling area; (B) high mussel density (1700 mussels m") over the entire 50 cm sampling area; (C) high mussel density clusters (1700 mussels m") 10 cm in length and placed 10 cm apart from each other; (D) 1.5 cm radius half cylinder (indicated by white arrow) placed upstream of sampling area; and (E) 1-cm2 grid (indicated by white arrow) placed upstream of sampling area. Each white square on flume bottom is 1 cm .
Direction of flow is bottom to top in all cases.
Fig. 2.2. Location of Lake Erie field site and individual sampling configurations (1-4).
Location of study area is indicated by the black circle in the insert. Image of Lake Erie from the National Geophysical Data Center, National Oceanic and Atmospheric
Administration, U.S. Department of Commerce, http://www.ngdc.noaa.gov/.
Fig. 2.3. Laboratory fertilization success as a function of distance downstream for
Dreissena polymorpha at five different flow chamber velocities. Plots are means ± SE, N
= 6.
Fig. 2.4. The relationship between fertilization success and sperm dilution, determined through a linear regression of the log-transformed sperm concentrations on fertilization success using data from all experiments. N = 108.
85 Fig. 2.5. Laboratory fertilization success and ejection (Q2) and sweep (Q4) events as a function of distance downstream for Dreissena polymorpha at seven different turbulence configurations and two flow chamber velocities. Fertilization plots are means ± SE, N =
3, and quadrant plots are based on 150 PIV frame-averaged means.
Fig. 2.6. PIV vector plots indicating direction and magnitude of the overall spatial flow fields for: (A) a mussel cluster; (B) an open patch between two mussel clusters; (C) a half cylinder 5 cm upstream (seen on left hand side of panel); and (D) a wire grid placed at point 0 (indicated by blue vertical bar on left hand side of panel). Y-axis represents height in cm, and x-axis downstream distance in cm. The length of the vector represents higher relative velocity.
Fig. 2.7. (A) Dimensionless velocity measurements for five downstream locations based on local velocities («) measured by particle image velocimetry (PIV) normalized to the free-stream velocity (U). Black symbols are 5 cm s" and white symbols are 1.5 cm s" .
Values are means ± SE, N = 5. (B) Measured bed shear stresses at 5 cm s"1 (black) and
1.5 cm s"1 (white) and their corresponding theoretical laminar (theory - L) and turbulent
(theory - T) calculated shear stresses (based on eqs. 7.25 and 7.44, respectively, in White
1994).
Fig. 2.8. Fertilization success as a function of distance downstream for Dreissena bugensis at four different bottom roughnesses and two different velocities in Lake Erie.
Plots are means ± SE, N = 5.
86 Fig. 2.9. Bottom roughness for the four-field configurations determined from a roughness profiler along a 1 -m transect, and vertical flow profiles from ADV measurements taken at 10, 30, and 50 cm along the sampling transect (data at 20 and 40 cm not illustrated).
Fig. 2.10. Fertilization success and ejection (Q2) and sweep (Q4) events as a function of distance downstream for Dreissena bugensis at four different bottom roughnesses and two different velocities in Lake Erie. Fertilization success plots are means ± SE, N = 5, and quadrant plots are based on the means of 120 s of data recorded at 25 Hz.
Fig. 2.11. Bed shear stress and Reynolds shear stress as a function of distance downstream for ADV data collected at four different bottom roughnesses and two different velocities in Lake Erie. Symbols represent means of 120 s of data recorded at
25 Hz.
87 Fig. 2.1.
88 2 3
Fig. 2.2.
89 100 0 cm s"1 1.5 cms"1 3.47 cms"1 80 4.60 cm s"1 5.06 cms"
60
40
20
—i— 10 20 30 40 50 Distance downstream (cm)
Fig. 2.3.
90 IU"
R2 = 0.99 107 - P< 0.0001 X
"^ 106 - V) £ 105 - JXr c o • := 104 - •^r TJ E • K 103- • . *W" CO • 102 - ^ * &• 1 10 - i I 1 1 20 40 60 80 100 fertilization success (%)
Fig. 2.4.
91 U = 1.5 cms" U = 5 cm s"1 100 100 Cylinder-generated turbulence I I -15cm I I -5cm
100
Q. T3 C ns a> -1 c g r 40 o rt i ?n ID g o Q. o 100 100 Mussel-generated turbulence I I low density I I high density ^H patch ejection - open bars sweep - lined bars 20 30 50 0 10 20 30 40 50 Distance downstream (cm) Fig. 2.5. 92 Velocity Magnitude [cm/s] A \X h-s? 6 € . . . i«*; .,',,^}p-i- -Njj. ' -i> -•=** •*-». :—*».-. —*- —> —* -*> — ' -*• —•- ~^- —*• *- _^-' —j> —*• —S> .tr~S£> -~s>. *•*£• —*• —5> ~^- *> •"- "?> •r*# .•-*' -* —s> -* -* -*• — —> '—* -»•. -* —> —J> ••?>£> ~S> T^ _•*> •?•?••. —s. ~» --ti. -* -* — — ->. — -^ -*»• -* -* -* -*• -» •„;-^ri .~.j^ _.«> '.-». >* -^ ~^sv >* . ~""* —s> ^*> -^ —*J>. — —^ -* -» -» ->- -*- -* _rt .-*>: ~-*& .<*. >^ -*-*>. -w>^.- ' -^si -» ->-** -H* —J, ~* —•*• ~~ — — -» -* •* —f —*• —s> •»>* —»• 3 . .^ —* ~* "^ ^* ~*• —*• ^" —* ~* -* ^* —-> _* S "Wto. v • 1 V ':- • y -Tto ^ - ' .» % V "•». "•^ - - ' -.:. -» '-» -» -" ~* — X .„> "na. •-> '^ ., ":.. ^ S •Si ~* -"'' • - %>• . —*• -* -^ — ~~ — -?fc • -^ -. - , -?•; A-.- T . -» -^v TV- • :-' ' - .-*. — -— -?• -*. - i_':.. '' iU. **• • * i- - ' '-4- '" J ' ••—' .-.—,_'. * _ '~' /-, .:.- :. _ '. **.. _ 0 .— ...__ wii, J D °r 10 15 Downstream distance (cm) Fig. 2.6. 93 dimensionless velocity (u/U) 1 0 10 10 1 0 3.5 3.0 E 2.5 ^ E g 2.0 o n E o 1.5 D) ' 0.5 -\ 0.0 1 B 03 0.1 Q. o- 0.001 a theory - T £H =B= =Q -1 5 cm s theory - L theory - T 0.0001 —i— 10 20 30 40 50 downstream distance (cm) Fig. 2.7. 94 U = 1.75 ±0.06 cms"1 50 100 Distance downstream (cm) U = 5.4 ± 0.2 cm s"1 Distance downstream (cm) Fig. 2.8. 95 Distance downstream (cm) 40 50 60 100 i T 1 1 r 0 2 4 6 8 40 low mussel density 30 H 9 5 1 20 Ka.glow = 8.29x10- mV 4 1 KavgHigh = 2.01x10- mV 10 O T^—r^ n 1 1— r o 0 2 4 6 8 JC 10 0 2 4 6 0 2 4 6 Velocity (cm s ) Fig. 2.9. 96 U = 1.75 ±0.06 cms" U = 5.4 ±0.2 cms"1 flat I I low density ejection - open bars sweep - lined bars [-] 7 ri n 7 [-] - 7 t / / 7 / i 7 / / • / / / • / / / / 10 20 30 40 50 high density patch ejection - open bars sweep - lined bars ri 7 7 / / 1-1 ri / 1-1 ri / / 7 / / / / 7 / / * / / / 7 / / / / 7 / rfl 7 / 7 10 20 30 40 50 Distance downstream (cm) Fig. 2.10. 97 U = 1.75 ±0.06 cms1 U = 5.4 ± 0.2 cm s1 40 50 0 10 Distance downstream (cm) Fig. 2.11. 98 CHAPTER 3: Effects of near-bed turbulence on settlement and resuspension of freshwater mussel larvae. Abstract Larval settlement and transport in benthic invertebrates has been theorized to be influenced by bottom roughness and the hydrodynamic forces it creates near the bed. This study of the freshwater invasive organism Dreissena bugensis examines larval and model larval transport in laboratory flow chamber experiments and larval settlement in Lake Erie field experiments. Both studies examined the influence of flow velocity, turbulence near the bed and differing bed roughness on larval transport or settlement. This is the first empirical study to examine turbulence parameters and larval movement together. Detailed laser particle imaging velocimetry (PIV) measurements in the laboratory, and acoustic Doppler velocimetry (ADV) in the field, determined the nature of small-scale turbulence in the near-bed environment, namely the frequency of sweeps and ejections and the extent of skimming flow. Larval transport was higher in bottom roughness configurations that generated skimming flow, such as the high density mussel configuration. Larval settlement in the field was higher in bottom roughness configurations with higher frequencies of sweeps and ejections, namely the patchy mussel configuration, generated by variations in bottom roughness due to the presence of mussels. The spatial configuration of mussel roughness influenced the creation and magnitude of skimming flow and wake interference, which can inhibit or enhance larval settlement, respectively. Introduction Successful larval settlement is critical for the long-term viability of sessile benthic invertebrate populations (Eckman 1983; Abelson and Denny 1997; Hadfield and Koehl 2004). Larval settlement can be separated into two key stages, both of which are affected by physical forces: the mass transport of larvae to the substratum (termed 'settlement'), and the subsequent establishment on the substratum through some means of attachment and subsequent metamorphosis (termed 'recruitment'; Abelson and Denny 1997). In terms of mass transport, turbulence and related fluid dynamic forces near the substratum are examples of 'just enough, but not too much' phenomena (Denny and Shibata 1989; Nishihara and Ackerman 2008). In the case of a larval transport, turbulent flow and mixing increases the probability that larvae will encounter a suitable substratum on which to settle (Crimaldi et al. 2002), but high turbulence has been shown to increase larval mortality (Rehmann et al. 2003) and inhibit larvae from successfully attaching to a substratum (Crimaldi et al. 2002). The relative strength of turbulence depends on characteristics of the bed, the state of the flow (which are strongly associated), and the morphology (e.g., ovoid or spherical) of the larva (McNair et al. 1997; Crimaldi et al. 2002). One such example is seen in coral reefs, where brief bursts of high velocity at the upper region of the reef structure determined where larvae were most likely to settle (Reidenbach et al. 2009). Established adult populations may affect larval settlement, although results have been somewhat inconsistent and contradictory due to interspecific variation (Pawlik 1992; Andre et al. 1993). For example, Thrush et al. (1996) found that adult Macomona liliana bivalves either increased or reduced larval settlement, depending on whether the 100 substrate was sandy or muddy, indicating that hydrodynamic constraints on larval settlement were generated at least in part by the presence of adult populations. Evidence to support this conclusion comes from a study on models of the Asian clam, Potamocorbula amurensis, which demonstrated that the probability of successful larval settlement decreased with a reduction in the spacing between adults (Crimaldi et al. 2002). This is because the resultant skimming flow (review in Schindler and Ackerman, 2009) reduced the probability that larvae would be transported to the bed and increased the probability that larvae would be swept off the bed before they had sufficient time to attach. Not surprisingly, the role of turbulence in larval recruitment has been a subject of much interest (Nowell and Jumars 1984; Denny and Shibata 1989; Crimaldi et al. 2002). Much of this work, however, has focused on theoretical modeling based on data from laboratory flow chambers. Some of the effort has been extended to the field, where coherent structures of the turbulent bursting process (Robinson 1991) have been found to create sweeps that hit the bed and spread the suspended particulate matter, including larvae, horizontally and ejections that may entrain larvae from the bed and carry particulate matter into the water column (Cellino and Lemmin 2004; O'Connor and Hondzo 2008). Topographic forcing by benthic bedforms (e.g., subsurface dunes or depressions; Best and Kostaschuk 2002) and by bottom roughness (O'Connor and Hondzo 2008) have been found to generate these coherent turbulent structures. Whether the bottom roughness created by the presence of benthic organisms generates similar turbulent events remains to be determined. Freshwater dreissenid mussels provide an excellent model system to examine the influence of near-bed hydrodynamics on larval settlement. Their natural history is similar 101 to those of many sessile marine invertebrates (Ackerman et al. 1994) that are the basis of current theoretical systems (Abelson and Denny 1997). The freshwater systems in which they exist display many similarities to marine systems with the exceptions of tidal forcing and high short-term variability including seiching are present (Wiiest and Lorke 2003). It is reasonable to suggest that the bottom roughness created by mussel populations generates coherent turbulent structures that influence larval deposition. The purpose of this study is, therefore, to determine whether the roughness generated by dreissenid mussel populations influence near-bed turbulence and consequently, larval settlement. Specifically, bed roughness features that generate high sweep frequencies and reduced incidence of skimming flow should have higher larval settlement than bed roughness features that generate high ejection frequencies and high incidence of skimming flow. Materials and Methods Laboratory - Larvae were collected with 100 um mesh plankton nets at Evan's Point, Lake Erie (42° 50' N, 79° 46' W), and maintained in the Hagan Aqualab, University of Guelph, using the techniques of Wright et al. (1996). Whereas mussel populations in the immediate sampling area consisted of only Dressena bugensis Andrusov, 1897 adults, it was not possible to identify the larvae to species. Larvae were monitored daily for their stage of development, and the settling pediveliger stage (~ 200 - 300 um and presence of foot; Ackerman et al. 1994) was isolated and used for laboratory experiments. The experiments were performed in a recirculating flow chamber {see Nishihara and Ackerman (2006) for a detailed description). The chamber was 10 cm wide with a 40-cm long region for flow development and a 3 5-cm long test section, with a water 102 depth of 7 cm. Experiments were performed at average velocities (U) of 3.1, 6.5, 8.3, 9.4, and 11.0 cm s"1 determined by volumetric flow measurements. These velocities correspond to chamber Reynolds numbers (Redh = Udh/v where dh = hydraulic diameter and v = kinematic viscosity) ranging from 3,058 to 10,956 (i.e., turbulent flow chamber conditions). Velocity profiles in the empty flow chamber at the beginning of the test section were determined from particle image velocimetry (PIV). The velocity profiles were logarithmic and uniform above a height of 3 cm, indicating a properly developing boundary layer (see Chapter 2). Larval transport experiments were performed in the flow chamber at each velocity to determine the influence of flow on larval motion at the bed. Dreissenid larvae were placed, using a pipette, at five locations on the bed centerline at x = 0, 10, 20, 30, and 35 cm downstream in the test section. Larval movements were observed at each location using a Nikon SMZ-2T stereomicroscope placed against the side wall of the flow chamber. Larval movements were classified using methods of Pawlik and Butman (1993) as either bedload transport, wherein larvae moved horizontally along the bed without ever leaving the bed, or suspended transport, wherein larvae left the bed for any period of time. For each trial, four larvae were placed at each location, and three replicates were used. Data were recorded as a percentage of larvae that moved from the four that were placed in the flow chamber for each replicate at each turbulence configuration and location (see below). Polystyrene beads (Cat # 19825, Polysciences Inc., Warrington, PA, USA) of a comparable size (250-310 urn) and density (1050 kgm" 3) to larvae were used as passive models to determine whether larval behaviour contributed to movement at the bed. Beads were placed at the same locations as larvae 103 and examined under the same velocities. Four to nine beads were placed at each location, and three replicates were used as in the larval experiments described above. In addition to the flat bottom of the flow chamber, three spatial configurations of mussel populations were examined using mussel shells attached to the bottom with Tac 'N Stik (Elmer's Products Inc., Columbus OH, USA) in low (860 mussels m"2) and high (1700 mussels m"2) densities over the 3 5-cm long test section, as well as in three high- density patches with 10-cm long clusters spaced 10 cm apart. The first cluster was placed at x = -35 cm so that x = 0 was 5 cm upstream of the second patch. In addition to the mussel shells, two other forms of well-characterized turbulence were created as reference points, grid turbulence, with a 1 cm x 1 cm wire-mesh of 0.1 cm diameter placed perpendicular to the direction of flow at x = 0 cm and x = -5 cm (i.e., 5 cm upstream of the test section), and cylinder turbulence with a half-cylinder (1.5 cm radius) placed horizontally across the width of the channel on the bottom at x = -5 cm and x = -15 cm. A total of 8 different mussel and turbulence-generating configurations resulted. Larval transport experiments, as described above, were undertaken at each turbulence configuration at U= 3.1 cms'1 and 11.0 cm s" . Three replicates were used for larvae and for the polystyrene beads. Two-way ANOVAs and linear regressions were used in STATISTICA (version 6.0, StatSoft Inc., Tulsa, OK, USA) to examine differences and associations between larval/model transport and velocity, distance, and turbulence configurations. Particle imaging velocimetry (PIV) - PIV measurements were performed for the 3.1 and 11.0 cm s"1 velocities in each of the different turbulence-generating flow configurations. 104 The laser sheet intersected the flow chamber in the mid-longitudinal plane, with a plane length of 13 cm {see Nishihara and Ackerman 2006; Chapter 2). Four 13-cm long PIV measurements were taken along the 3 5-cm test section, with a 2 cm overlap at each end of the sections. Images of silver-coated, hollow glass beads (13 jam diameter) added to the flow chamber were captured at a frequency of 30 Hz through the side wall of the flow chamber. A total of 400 images were collected, of which 150 image pairs (i.e., 300 images) were processed using PIVview 2C batch processing software (PIVTEC GmBH, Germany) to generate a series of instantaneous u and w velocity values in the streamwise (x) and vertical (z) directions, respectively, for each 13-cm long section. Mean velocity {u and w) in each frame was calculated and used to determine magnitude of the fluctuations u' and w' from the mean velocity (i.e., u = u + u' and w = w + w', where u and w are the instantaneous velocities). Product u 'w' were classified using quadrant analysis to determine ejections (quadrant 2; i.e., u 'w' < 0 where w' > 0) and sweeps (quadrant 4; i.e., u'w' < 0 where w' < 0), as well as inward (quadrant l;u'w'>0 where w' > 0) and outward interactions (quadrant 3; u V > 0 where w' < 0). Quadrant events larger than the threshold hole size of 2 (H = \u' w'\ l\u' w'\) were considered significant and were counted to determine the frequency of quadrant events. Quadrant analysis has been used predominantly to determine temporal fluctuations of these velocity components, but recent studies have shown that this technique can also identify persistent spatial velocity disturbances (Pokrajac et al. 2007; Chapter 2), similar to those in the current study. The PIV data were also used to determine shear velocities (u*) from the law of the wall, calculated by multiplying the von Karman constant (K = 0.41) by the slope of the velocity versus ln(z) in the logarithmic portion of the boundary layer (e.g., Ackerman and Hoover 105 2001). Reynolds shear stress (x) was calculated from the PIV data 2 cm above the bottom of the flow chamber via T = -pu'w' (l) where p is density of the water. Field - In order to determine the role of bottom roughness in larval settlement, experiments were conducted in the eastern basin of Lake Erie at Evans Point. Four sites within -50 m of each other and at a depth of 40 cm with distinct mussel bottom roughness were used, corresponding loosely to the conditions examined in the laboratory: (1) flat bottom; (2) low mussel density (600 ± 26 mussels m" ); (3) high mussel density (1500 ± 64 mussels m"); and (4) high mussel density patches separated by open flat sections about 10 to 15 cm long. Only Dreissena bugensis were present in the area. A bottom profiler consisting of 50-cm long pins oriented vertically and spaced 2.5 cm apart in a 1 -m long frame was used to profile bed roughness at each site along a 2-m transect. Two transects spaced 5 cm apart were taken on either side of the central transect. Digital photos were taken of each profile measurement for analysis, where the top of each pin provided an indication of the roughness height {see Chapter 2, for detailed description). These data were used to calculate average bed roughness, kavg, and its coefficient of variation (CV = standard deviation/mean) at each bottom roughness. Larval settlement was determined through methods modified from Martel (1993). Larval collecting pads consisted of 12 x 11 cm, 0.6 cm thick nylon kitchen scouring pads (Hero/Sun-Glo Products Inc., Mississauga, ON), which have been shown to yield high numbers of dreissenid larvae over short periods (Martel 1992, 1993). Collecting pads 106 were soaked in fresh well water 2 to 3 d prior to use to remove any chemicals remaining from the manufacturing process that might inhibit larval settlement. Four collectors were anchored to the bed at each site with stainless steel tent pegs at both ends of the pads to ensure they remained flat and parallel to the lake bed. To estimate the flux of larvae in the water column, four other collectors per site were attached to vertical lines anchored to the lake bed and kept just below the surface (i.e., ~ 35 cm above the bottom) using small, air-filled plastic floats. Collectors were deployed for 72 h on five occasions between 9 August and 9 September 2008. Once retrieved, pads were washed with tap water for 15 s per side into a large, shallow pan to remove attached larvae, and this procedure was repeated three times. Removed larvae were sieved through a 200-um mesh and counted using a Nikon SMZ-2T stereomicroscope. Acoustic Doppler velocimetry (ADV) - Vertical profiles of velocity were taken at each site using an ADV during the larval settlement study. On each occasion, profiles were taken along the first 50 cm of the central transect at each of the sampling locations with a vertical, hand-held stand (Chapter 2). The profiles started as close to the bed as possible (within 2 cm) and ended just below the water surface (~ 37 cm in depth), giving from 5 to 7 points per profile (each point sampled at 25 Hz for 2 min). This data set was analyzed with quadrant analysis. Shear velocities (u*) were calculated from the law of the wall in a similar manner to that of the PIV data in the laboratory. 107 Results Flow chamber experiments - The frequency of both larval and polystyrene bead bedload and suspended transport increased with velocity (Fig. 3.1). This pattern was consistent for the three different mussel bed configurations and the two reference turbulence configurations (e.g., grid and cylinder). In general, at least 80% of larvae and models experienced bedload transport at the highest velocity (U) of 11 cm s" , whereas suspended transport did not exceed 26% for either group. Bedload transport appeared to be higher for polystyrene models than larvae for 4 < U< 10 cm s"1, with model transport exhibiting a logarithmic-like curve leveling out at the highest velocities, whereas larval transport exhibited an exponential-like pattern at the higher velocities. There was no significant difference, however, between overall larvae or model bedload transport (larval = 58 ± 4% [mean ± SE]; model = 57 ± 4%; t79 = 0.33, P = 0.74), but larval bedload transport had a higher significant positive association with velocity than model bedload transport (i.e., R2 = 0.87 vs. R = 0.44, respectively; Table 3.1). The suspended transport of larvae (10 ± 2%) was significantly higher than that for models (2.9 ± 0.8%; t79 = 4.68, P < 0.0001). Both had significant positive associations with velocity, but to a lesser degree than bedload transport (Table 3.1). The three mussel-bed configurations and the two reference turbulence configurations exhibited similar patterns for the different velocities with respect to bedload transport, with a clear separation between larvae and models (Fig. 3.1). The exception was larval bedload transport over the low mussel density configuration, which resembled that for the model (Fig. 3.1). There were noticeable differences, however, 108 with respect to suspended transport among the 8 mussel-bed and turbulence configurations examined, with the highest larval suspended transport occurring in the grid and mussel configurations. Two-way ANOVAs for both larval and model bedload transport indicated significant differences with respect to velocity (larval: Fi64 = 862, P < 0.0001; model: FIJ64 = 238, P < 0.0001) and configurations of mussel bed and turbulence (larval: F7;64 = 3.61, P = 0.0024; model: F7>64 = 2.69, P = 0.02), as well as a significant velocity x configuration interaction (larval: F7i64 = 3.68, P = 0.002; model: F7]64 = 3.36, P = 0.004). Subsequent one-way ANOVAs revealed significant bedload transport differences with respect to velocity (larval: F4j35 = 67.3, P < 0.0001; model: F4>35 = 47, P < 0.0001), but no significant differences were seen with respect to configuration (larval: F7>72 = 0.27, P = 0.96; model: F7j72 = 0.59, P = 0.76). Two-way ANOVAs for larval and model suspended transport also indicated significant differences in velocity (larval: Fi64 = 29.2, P < 0.00001; model: F1>64 = 8.81, P < 0.005), but significant differences in configuration were found only among the models (F7>64 = 2.49, P = 0.03). A significant velocity X configuration interaction was also detected among the models (F7j64 = 2.39, P = 0.03). Subsequent one-way ANOVAs revealed significant differences in suspended transport with respect to velocity for larvae (F435 = 13.4, P = 0.96), with no significant differences seen with respect to configuration for either larvae or model groups (F7-72 = 0.64, P = 0.72 and F7]72 = 2.00, P = 0.07, respectively). Larval bedload transport generally declined with downstream distance with the grid and cylinder configurations, but increased slightly with the flat-bottom and mussel- bed configurations at 3.1 cm s"1 (Fig. 3.2 A - D). There was only a small degree of suspended transport at 3.1 cm s"1 in the grid and cylinder turbulence configurations (data 109 not presented), with the highest frequencies of 17% seen in the larvae at x = 0 and 10 cm in both grid configurations (grid location x = -5 and x = 0 cm). At 11 cm s"1, however, all mussel and turbulence configurations regardless of downstream distance were either approaching or were at 100% bedload transport, with no significant differences or associations (data not presented). Suspended transport tended to decline with downstream distance in a more consistent manner (Fig. 3.2 E - H). Differences were observed for the low mussel-density configuration, which had higher suspended transport at x > 30 cm. Larval suspended transport tended to be higher than the model group, a trend seen in all the mussel and turbulence configurations. Two-way ANOVAs for larval and model bedload transport revealed significant differences in velocity (larval: Fu0 = 626, P < 0.0001; model: Fi>70 = 179, P < 0.0001), but a significant difference in distance was found only among larvae (F470 = 3.10, P = 0.02). There was no significant velocity x distance interaction. At the lowest velocity (3.1 cm s"1), only the grid generated turbulence configuration (where the mesh x = -5 cm) had significant association with larval bedload transport (Table 3.1). Two-way ANOVAs for larval and model suspended transport again indicated significant differences in velocity (larval: Fij70 = 52.2, P < 0.0001; model: Fi,70 = 7.19, P < 0.01), and significant differences in distance of larval suspended transport (F^o = 11.7, P < 0.001). There were no significant velocity x distance interactions. There was significant negative association between suspended larval transport and distance when data from all 8 turbulence configurations were examined (Table 3.1). As indicated above, there was no significant relationship between the type of turbulence configuration and suspended larval transport when the full data set was 110 compared, but there was a small, significant association between turbulence configuration type and suspended larval transport at x > 10 cm. Three of the reference turbulence configurations had significant negative associations between distance downstream and suspended larval transport, the strongest association being seen with the grid-generated turbulence with the mesh placed at x = 0 cm upstream (Table 3.1). PIV data was compared to the hole size, and significant turbulent events were found and classified by quadrant analysis to identify turbulent ejections (Q2) and sweeps (Q4), as well as inward (Ql) and outward (Q3) interactions (Fig. 3.3). Significant differences were found among the mussel and turbulence configurations using one-way ANOVA for all four quadrants (F7J2 ranged from 3.47 to 6.09, P < 0.003 in all cases), yet there was not a very strong association between quadrant events and larval/model transport. There were distinct areas of skimming flow, indicated by the steep velocity gradients and relative vector magnitudes above the mussel beds in the high mussel- density configuration and the high-density mussel patch configuration (Fig. 3.3C and D). This area of skimming flow in the patch configuration, however, began to diminish before the next patch (Fig. 3.3D). The flat-bottom configuration exhibited a uniform spatial flow profile, with most of the flow at the mean velocity (Fig. 3.3 A). The low mussel-density configuration had a small area of skimming flow well above the mussels, but there was evidence of wake interference between individual mussels, indicated by the larger velocity vectors and gradients between various mussels (Fig. 3.3B). Not surprisingly, shear velocity (u*) determined from the PIV data for each of the mussel and turbulence-generation configurations had a strong, significant association with {/(Table 3.1, Fig. 3.4; note that R2 ranged from 0.61 to 0.94; P < 0.05 in all cases). ill At 3.1 cm s" , u* was generally uniform with respect to downstream distance {u* = 0.5 ± 0.02 cm s" ), but there were some differences with respect to the mussel configurations. The high-density mussel configuration exhibited the highest u* over all five downstream locations (u* = 0.74 ± 0.04 cm s"1), whereas the flat-bottom configuration had the lowest (u* = 0.39 ± 0.02 cm s"1). At 11 cm s"1, however, u* was much more variable with downstream distance (w* = 1.3 ± 0.03 cm s"1), particularly in the adult mussel-bed configurations. As with velocity, however, shear velocity did not differ significantly = across the different mussel and turbulence configurations (one-way ANOVA, F7j72 0.54, P = 0.80). Shear velocity had the strongest significant positive associations with larval bedload transport, followed by model bedload transport, with a minor association seen with larval suspended transport (Table 3.1). Reynolds shear stress (x) did not show any significant associations, but one-way ANOVA indicated there was a significant difference across the mussel and turbulence configurations (Fjj2 = 3.04, P < 0.01), namely the high-density mussel-patch configuration had x significantly higher than in all other configurations (0.10 ± 0.05 and 0.003 ± 0.01 Pa, respectively). Field experiments - There was noticeable differences in mean larval settlement among the four mussel configurations in the field, with the highest larval settlement in the mussel-patch configuration (Fig. 3.5B). This difference was significant as revealed by one-way ANOVA (F376 = 3.50, P < 0.02), withpairwise differences between the mussel- patch configuration and the flat-bottom (P < 0.03) and high-density mussel configurations (P < 0.0005). The flux of larvae found in the water column above each configuration was 9 1 not, however, significantly different (mean = 3,386 ± 195 larvae m" day" ; one-way ANOVA: F3,54 = 0.10, P = 0.96; Fig. 3.5B). Bed roughness, kavg, calculated from the 112 bottom profiles, indicated that the flat bottom had the lowest roughness, whereas the mussel patch configuration had the highest. The low-density mussel configuration exhibited the most variable roughness, as indicated by the highest CV, followed by the . mussel-patch and high-density mussel configurations (Fig. 3.5A). Quadrant turbulent events determined from ADV flow profiles (for individual flow profiles see Chapter 2) had the highest frequencies of ejection (Q2) and sweep (Q4) events in the mussel patch configuration, whereas outward interaction (Q3) events were lowest in that configuration (Fig. 3.6). Significant differences in Q2, Q3, and Q4 events across the 4 configurations were revealed by one-way ANOVAs (Q2: F3;g = 5.08, P = 0.03; Q3: F3;8 = 14.1, P = 0.001; and Q4: F3,8 = 4.07, P < 0.05). Only Q3 and Q4, however, had significant associations with configuration type. In terms of significant associations between turbulent events and larval settlement, Q3 had the strongest negative association, with the strongest positive association seen with ejections (Q2) followed by sweeps (Q4) (Table 3.1). No other physical parameters displayed any significant association with larval settlement. Discussion The results of this study demonstrate that roughness created by the presence of mussels on the bottom influences small-scale, near-bed turbulence, and thereby the larval transport and settlement of freshwater dreissenid bivalves. This observation is consistent with the predictions of previous studies that have modeled turbulence measured near the bed (Crimaldi et al. 2002; Hendriks et al. 2006; Reidenbach et al. 2009) and those that have used hydrodynamic arguments to explain larval transport and behaviour near the 113 bed (Pawlik and Butman 1993; Dobretsov and Wahl 2008). To the best of our knowledge, ours is the first empirical study to examine turbulence parameters and larval movement together. The strong positive correlation between ejection (Q2) and sweep (Q4) events in the field with larval settlement was predicted in the earlier theoretical studies mentioned above. Hendriks et al. (2006) examined the induced turbulence from a grid and a live oyster ridge that were introduced into a flow chamber, and related the resultant turbulence to the probability of larval settlement and resuspension. They suggested that the induced turbulence caused by the higher frequency of sweeps and ejections led to erosion of the viscous sublayer. Further they suggested that the decreased thickness of the sublayer would promote transport into the area and increase the probability of settlement (Hendriks et al. 2006). Regardless of the mechanism, there was moderate significant association between larval settlement in the field and frequency of ejections and sweeps in the field (R2 > 0.5 in both cases) in the current study. In terms of the extent of potential induced turbulence in the field, the mussel-patch configuration had the highest average bed roughness (kavg), the highest frequency of ejections and sweeps and the highest larval settlement. Conversely, the high-density mussel configuration exhibited the second highest kavg, but the lowest frequency of ejections and sweeps and the lowest larval settlement. This pattern is consistent with the predictions by Crimaldi et al. (2002) that decreased spacing between model clams would reduce larval settlement rates by creating turbulence that would disrupt the ability of larvae to "anchor" to the bottom rather than affect the flux of larvae to the bottom. Support for this prediction was given by estimates of the 114 Reynolds shear stress, which captures the structure of the episodic sweep-ejection cycle in the near-bed region. Crimaldi et al. (2002) argued that this prediction was inconsistent with the prediction of Eckman's (1990) one-dimensional model that larval flux and settlement would increase with increases in roughness element density. The current study provides evidence to support both predictions in that larval settlement was higher in the patchy mussel density configuration compared to the flat bottom configuration in support of Eckman (1990), but larval settlement was lowest in the high-mussel density configuration in support of Crimaldi et al. (2002). The reason for this is that it is not simply roughness element density that influences larval settlement, but rather the spatial configuration of the roughness elements and the local flow regimes those configurations create, namely the extent of skimming flow and wake interference (Schindler and Ackerman 2009). In the current study, the high-mussel density configuration led to high velocity skimming flow, whereas less evidence of skimming flow was seen with the flat-bottom configuration, and wake interference was observed in the low-density configuration (Fig. 3.3). Indeed the high-mussel density configuration had both lower larval settlement rates in the field and higher suspended larval transport rates in the laboratory compared to their respective flat-bottom configuration. The low mussel-density configuration, however, exhibited higher larval settlement rates and higher suspended transport rates than the flat- bottom, in support of Crimaldi et al. (2002) but contradictory to Eckman (1990). In this case, with wake interference flow, more flow between the individual mussels would result in higher encounter rates with the bottom and thus higher settlement rates, despite the fact that there appears to be more skimming-like flow relative to the flat bottom. This 115 may be a reason why Eckman (1990) found that his model could not be applied to sparse roughness density arrays. That the highest larval settlement rates were observed in the high density mussel patch configuration is consistent with Eckman (1990). In this case, the skimming flow immediately above the mussels dissipates over the open patch before the next mussel cluster leading to the highest frequency of sweeps and ejections. Based on the high larval suspended transport rates in the patch configuration in the laboratory, it appears that this skimming flow may help to transport larvae from one patch to another, but sweeps in the open patch may provide an opportunity for the larvae to settle onto the bed. Further evidence of the importance of local spatial resolution comes from a recent study on larval settlement in coral reefs (Reidenbach et al. 2009). Microhabitats were found just a few centimeters below the reef surface that were protected from high free-stream velocities in a flume study of larval nudibranchs, Phestilla sibogae. It was these small differences in larval habitat and position that influenced the type of instantaneous sweep and ejection events that the larvae experienced, with the maximum forces at the top of the reef being three times greater that just 5 cm below the top of the reef. Moreover there was a much higher frequency of these larger forces at the reef top. Based on the magnitude of these forces, larval attachment strength and speed become the primary determinants of where larvae would settle in this particular system; only larvae with strong and fast attachment being able to successfully recruit to the top of reefs. Since P. sibogae have relatively weak attachment strength, they were predicted to settle only in the sheltered regions below the reef tops (Reidenbach et al. 2009). In the current study, evidence for localized small-scale structures determining suitable settlement areas 116 was seen with the differences in larval resuspension in the turbulence-generating grid and cylinder configurations with distance downstream. Larval resuspension, which would act to prevent larval settlement, occurred approximately 50 to 60% of the time in the area of the flow chamber immediately downstream of the respective roughness features, but dropped to 0% only 10 to 20 cm further downstream. PIV images confirm localized areas of sweeps and ejections behind the grid and cylinder elements {see Chapter 2). The mechanisms described above illustrate the passive transport of larvae either away or towards the bed. Previous studies have shown, however, that there may be active components of larval behaviour that can influence both settlement and recruitment (Pawlik and Butman 1993). Two such behaviours described are 'balloonist' or 'ping- pong ball' behaviour in which larvae explore a particular substrate and then reenter the water column, and a drifting behaviour wherein larvae appear to drift just above the substratum (Jonsson et al. 1991; Andre et al. 1993). The differences in resuspension rates between larvae and polystyrene models in the current study suggest that resuspended transport may also have an active larval component in dreissenids, although no active behaviour was observed. Other mollusc species have been shown to utilize active settlement behaviour, such as the gastropod Lacuna spp., which secretes a mucous thread which is pulled by water currents to transport the gastrod and initiate drifting behaviour (Martel and Chia 1991), and dreissenid and blue mussels Mytilus edulis that use drifting threads to accomplish a similar drifting behaviour and have been shown to settle in response to low shear velocity (Ackerman et al. 1994; Dobretsov and Wahl 2008). In terms of passive settlement or transport in the current study, shear velocity did not appear to be a factor affecting larval settlement in the field, but it was the major factor 117 controlling larval transport in the laboratory after they had settled. Bedload transport was observed at u* = 0.50 cm s"1 for all larval configurations and all model configurations with the exception of the high mussel-density configuration. Resuspension was initially observed at u* = 0.50 cm s"1 for the grid and cylinder turbulence configurations, but was not seen in the flat-bottom configuration until u* = 1.03 cm s"1. These results were similar to work by Pawlik and Butman (1993), who observed bedload transport for model polystyrene spheres at u* = 0.47 cm s"1 and initial resuspension at u* = 1.03 cm s"1. In conclusion, larval settlement in dreissenid mussels is influenced by turbulent sweeps and ejections, which are generated by variations in bottom roughness due to the presence of mussels. The spatial configuration of mussel roughness plays a major role in the determining the character of skimming and wake interference flow, which can inhibit or enhance larval settlement, respectively. These similar turbulent events also appear to influence larval resuspension off the bed, but to a much lesser degree. There appears to be an active component involved in the resuspension of dreissenid larvae, that controls their ability to move from one region of the bed to another. Further detailed studies linking physical forces and larval behavior will help increase understanding of the processes controlling benthic organisms in general, regardless of habitat and species, and will hopefully lead to accurate predictive models that will help to control invasive species and enhance the survival of threatened species. 118 References Abelson A and Denny M. 1997. Settlement of marine organisms in flow. Annual Review of Ecology and Systematics 28: 317-339. Ackerman JD, Sim B, Nichols SJ, and Claudi R. 1994. A review of the early life history of zebra mussels (Dreissena polymorpha): comparisons with marine bivalves. Canadian Journal of Zoology 72: 1169-1179. Ackerman J D, and Hoover TM. 2001. Measurement of local bed shear stress using a Preston-static tube. Limnol. Oceanogr. 46: 2080-2087. Andre C, Jonsson PR, and Lindegarth M. 1993. Predation on settling bivalve larvae by benthic suspension feeders: the role of hydrodynamics and larval behaviour. Marine Ecology Progress Series 97: 183-192. Best J and Kostaschuk R. 2002. An experimental study of turbulent flow over a low- angle dune. Journal of Geophysical Research 107: 3135-3153. Cellino M and Lemmin U. 2004. Influence of coherent flow structures on the dynamics of suspended sediment transport in open-channel flow. Journal of Hydraulic Engineering 130: 1077-1088. Crimaldi JP, Thompson JK, Rosman JH, Lowe RJ, and Koseff JR. 2002. Hydrodynamics of larval settlement: the influence of turbulent stress events at potential recruitment sites. Limnology and Oceanography 47: 1137-1151. Crisp DJ. 1955. The behaviour of barnacle cyprids in relation to water movement over a surface. Journal of Experimental Biology 32: 569-590. 119 Denny MW and Shibata MF. 1989. Consequences of surf-zone turbulence for settlement and external fertilization. The American Naturalist 134: 859-889. Dobretsov S and Wahl M. 2008. Larval recruitment of the blue mussel Mytilus edulis: the effect of flow and algae. Journal of Experimental Marine Biology 355: 137- 144. Eckman JE. 1983. Hydrodynamic processes affecting benthic recruitment. Limnology and Oceanography 28: 241-257. Eckman JE. 1990. A model of passive settlement by planktonic larvae onto bottoms of differing roughness. Limnology and Oceanography 35: 887-901. Hadfield MG and Koehl MAR. 2004. Rapid behavioral responses of an invertebrate larva to dissolved settlement cue. Biological Bulletin 207: 28-43. Hendriks IE., van Duran LA and Herman PMJ. 2006. Turbulence levels in a flume comparedto the field: implications for larval settlement studies. Journal of Sea Research 55: 15-29. Jonsson PR, Andre C, and Lindegarth M. 1991. Swimming behaviour of marine bivalve larvae in a flume boundary-layer flow: evidence for near-bottom confinement. Marine Ecology Progress Series 79: 67-76. Lane SN, Biron PM, Bradbrook KF, Butler JB, Chandler JH, Crowell MD, McLelland SJ, Richards KS and Roy AG. 1998. Three-dimensional measurement of river 120 channel flow processes using acoustic Doppler velocimetry. Earth Surf Processes and Landforms 23: 1247-1263. Martel, A. 1992. Scouring pad collectors for zebra mussels. Dreissenapolymorpha Information Review 3: 2-3. Martel, A. 1993. Dispersal and recruitment of zebra mussel {Dreissena polymorpha) in a nearshore area in west-central Lake Erie: the significance of postmetamorphic drifting. Canadian Journal of Fisheries and Aquatic Sciences 50: 3-12. Martel, A., and Chia F-S. 1991. Foot-raising behaviour and active participation during the initial phase of post-metamorphic drifting in the gastropod Lacuna spp. Marine Ecology Progress Series 72: 247-254. McNair JN, Newbold JD, and Hart DD. 1997. Turbulent transport of suspended particles and dispersing benthic organisms: how long to hit bottom? Journal of Theoretical Biology 188: 29-52. Nishihara GN and Ackerman JD. 2006. The effects of hydrodynamics on the mass transfer of dissolved inorganic carbon to the freshwater macrophyte Vallisneria americana. Limnology and Oceanography 51: 2734-2745. Nishihara GN and Ackerman JD. 2008. Mass transport in aquatic environments, pp. 299 - 326 In: C. Gualtieri and D.T. Mihailovic (eds.) Fluid Mechanics of Environmental Interfaces. Taylor & Francis. Nowell ARM and Jumars PA. 1984. Flow environments of aquatic benthos. Annual Review of Ecology and Systematics 15: 303-328. 121 O'Connor, B. L., and M. Hondzo. 2008. Dissolved oxygen transfer to sediments by sweep and eject motions in aquatic environments. Limnology and Oceanography 53: 566-578. Pawlik JR. 1992. Chemical ecology of the settlement of benthic marine invertebrates. Oceanography and Marine Biology Annual Reviews 30: 273-335. Pawlik JR and Butman CA. 1993. Settlement of a marine tube worm as a function of current velocity: interacting effects of hydrodynamics and behavior. Limnology and Oceanography 38: 1730-1740. Pokrajac, D., L. J. Campbell, V. Nikora, C. Manes, and I. McEwan. 2007. Quadrant analysis of persistent spatial velocity perturbations over square-bar roughness. Experimental Fluids 42: 413-423. Rehmann CR, Stoeckel JA, and Schneider DW. 2003. Effect of turbulence on the mortality of zebra mussel veligers. Canadian Journal of Zoology 81: 1063-1069. Reidenbach, M. A., J. R. Koseff, and M. A. R. Koehl. 2009. Hydrodynamic forces on larvae affect their settlement on coral reefs in turbulent, wave-driven flow. Limnology and Oceanography 54: 318-330. Robinson, SK. 1991. Coherent motions in the turbulent boundary layer. Annual Reviews of Fluid Mechanics 23: 601-639. Schindler, R.J. and J.D. Ackerman. 2009. The environmental hydraulics of turbulent boundary layers. In: D.T. Mihailovic and C. Gualtieri (eds.) Advances in Environmental Fluid Mechanics. World Scientific, London. 40 pp. 122 Thrush SF, Hewitt JE, Pridmore RD, and Cummings VJ. 1996. Adult/juvenile interactions of infaunal bivalves: contrasting outcomes in different habitats. Marine Ecology Progress Series 132: 83-92. Wainman BC, Hincks SS, Kaushik NK, and Mackie GL 1996. Biofilm and substrate preference in the dreissenid larvae of Lake Erie. Canadian Journal of Fisheries and Aquatic Sciences 53: 134-140. Wright DA, Setzler-Hamilton EM, Magee JA, and Harvey HR. 1996. Laboratory culture of zebra (Dreissena polymorpha) and quagga (D. bugensis) mussel larvae using estuarine algae. Journal of Great Lakes Research 22: 46-54. Wiiest A and Lorke A. 2003. Small-scale hydrodynamics in lakes. Annual Reviews of Fluid Mechanics 35: 373-412. 123 Table 3.1. Significant associations with R values greater than 0.100 for laboratory and field data using linear regression analysis. Comparison R2 P value Laboratory: Larval bedload transport with U 0.87 < 0.0001 Model bedload transport with U 0.77 < 0.0001 Larval suspended transport with U 0.44 < 0.0001 Model suspended transport with U 0.13 0.02 Larval bedload transport with grid turbulence 0.54 0.015 Larval suspended transport with x 0.20 < 0.0001 Larval suspended transport with x > 10 cm: full data set 0.15 0.03 Larval suspended transport with x > 10 cm: grid x = 0 0.61 <0.01 Larval suspended transport with x > 10 cm: grid x = -5 0.54 <0.02 Larval suspended transport with x > 10 cm: cylinder x = -5 0.52 <0.02 Shear velocity with U 0.86 < 0.0001 Shear velocity with larval bedload transport 0.74 < 0.0001 Shear velocity with model bedload transport 0.58 < 0.0001 Shear velocity with larval suspended transport 0.21 < 0.0001 Field: Q3 with bottom configuration 0.35 <0.05 Q4 with bottom configuration 0.48 <0.05 Q2 with larval settlement 0.58 < 0.005 Q3 with larval settlement 0.67 < 0.005 Q4 with larval settlement 0.51 <0.01 124 Figure Legends Fig. 3.1. Percent of individual Dreissena sp. larvae and model polystyrene spheres that exhibited bedload transport (panels A - D) and suspended (panels E - H) movement along the flow chamber bottom for three mussel bed configurations and the two reference turbulence configurations (e.g., grid and cylinder) at five different velocities ranging from 3.1 cm s"1 to 11 cm s"1. Plots are means ± SE, N = 15 (5 locations x 3 replicates). Fig. 3.2. Percent of individual Dreissena sp. larvae and model polystyrene spheres that exhibited bedload transport at a velocity of 3.1 cm s"1 (panels A - D) and suspended movement at a velocity of 11 cm s"1 (panels E - H) along the flow chamber bottom for three mussel bed configurations and the two reference turbulence configurations (e.g., grid and cylinder) at five different locations downstream. Plots are means ± SE, N = 3. Fig. 3.3. Particle image velocimetry (PIV) velocity flow profiles for U= 11 cm s"1 over four 13-cm long sections of flow chamber bottom roughness configurations; (A) flat bottom, (B) low mussel density, (C) high mussel density, and (D) high mussel density patches 10 cm apart. Areas of high velocity swimming flow are represented by the dark grey contours and larger vectors. Fig. 3.4. Shear velocities determined from PIV data using the law of the wall, calculated by multiplying the von Karman constant (K = 0.41) by the slope of the velocity versus ln(z) in the logarithmic portion of the boundary layer, for eight different turbulence 125 configurations over five different locations downstream at velocities of 3.1 cm s" and 11 cm s"1. Values are means ± SE, N = 3. Fig. 3.5. (B) Larval settlement for Dreissena sp. larvae onto 12x11 cm, 0.6 cm thick nylon scouring pads deployed over mussel beds of different configurations at Evan's Point in the eastern basin on Lake Erie, and in the water column above each of the different configurations. Plots are means ± SE, N = 20. (A) The roughness of each configuration is represented by kavg, with the coefficient of variation (%) of each plot in brackets. Fig. 3.6. Quadrant analysis of the turbulence measured over mussel beds of different configurations at Evan's Point in the eastern basin on Lake Erie. Quadrant plots are three separate 120 second time-series means ± SE from three different ADV profiles between 2007 and 2008. Depth of mussel beds = 40 cm. 126 Bedload transport Suspended transport 100 -i A: Flat T 80 • 60 - r 15 40 - \ 10 20 - j_ • larvae O model 0 n • . 30 • B: Grid F: Grid 80 • / / 40 • ^* T T 10 • larvae x ~ -5 cm •e 20 O model x = -5 cm J o o • larvae x = 0 cm Q. a. CO A model x = 0 cm w c c CD o 2 •a 30-S 1 100 C: Cylinder G: Cylinder c -o a> CD Q. CO 25 3 CO 15 • larvae x =-15 cm o model x = -15 cm • larvae x = -5 cm \/\ > A model x = -5 cm • -JY '.: : 100 A D: Mussel shells 40 A • larvae low density o model low density • larvae high density A model high density • larvae patch a model patch 10 12 U(cms"') Fig. 3.1. 127 Bedload transport: 3.1 cm s Suspended transport: 11 cm s' E: Flat • larvae 0 model —9 9 ^^r^^ <•^= ^ > • larvae x = -15 cm O model x = -15 cm T larvae x = -5 cm A model x = -5 cm 30 40 0 10 distance (cm) Fig. 3.2. 128 13 velocity (cm s'1) '.=-.*^S *r*%»' ;:"^>- i> j>.. —1^.:'' •'-.-^•'.. ;^™*> -^^' '«™*> -js, •--jh. —^> —^* ...... -.•• • 1 —rS*. • -~3>. —i». —i> _j> —i> —*• ' •-*"- —s>. ^ —*• »—»»-: —*•• —* —?•'.' ••T?," ;';-*; !> "* - -*• -*»•• •">£*. 1 -*. .*=4?> =«*> —*-' 5*ft ..~^>. ; ™*. ,*,-^ ^,, ^ . -* -"*• *__^>. 'rt*>'\ -":"r4£> ^* -* — _i> ~j» —$• —»£> -*&> ^* -*- —**• —9- —J> —*> *~^J» '-*• «^ '.-:=>*>- -*i> -*• —*> -* —5> _-<> ~^> *^J5- —S> *• —s> —*• —*• Fig. 3.3. 129 1.8 1.6 ^~* 1.4 - '(/) -1 E 11 cm s ^ 1.2 - • flat low density * o T high density 3 ^ 1.0 - A patch o • cylinder x = -5 cm o • cylinder x = -15 cm CD • grid x = 0 cm i>— 0.8 - grid x = -5 cm CD o CD szV) 0.6 - 3.1 cm s 0.4 0.2 —i— 0 10 20 30 35 distance (cm) Fig. 3.4. 130 E o > 10 (14%) 8 - w (14%) 6H c (25%) D) 4 (10%) 2 2 T3 CD B "° 3500 I 3000 ••• Benthic i i Pelagic •o CN 2500 5 2000 CO 1500 H 1000 low density higih densit y patch bottom roughness Fig. 3.5. 131 30 H I I I I I flat low density high density patch Configuration Fig. 3.6. 132 CHAPTER 4: The role of bottom roughness parameters on the transport of sperm and larvae of benthic organisms. Abstract Bottom roughness elements have been shown to influence gamete and larval transport of benthic organisms. The ratio of the roughness-element spacing (X) and roughness- element height (k) in 2D models determines the type of flow regime created: (a) X/k < 8 results in skimming flow; (b) X/k ~ 8 results in wake interference flow; and (c) XIk > 8 results in isolated roughness flow. The computational fluid dynamic modeling program COMSOL in a 2D model was used to determine the accuracy of the X/k ratio in predicting the flow regime for a series of manipulated bottom roughness parameters and for three different bottom roughness 2D geometries (square, rounded, and triangular). A continuously released scalar was also used as a proxy for gametes and larvae to determine relative transport under the different ratios and geometries. The model flow regimes fit closely with the predictions from the ratios with the square and triangle geometries, but the rounded geometry required a lower ratio than expected for skimming flow. Relative transport (RT) of the scalar confirmed the model flow-regime results, with significant differences in RT seen among the three flow types at each geometry, and significantly lower RT values for skimming flow in the rounded geometry. In general, the ratio of X/k appears to provide an accurate means of classifying roughness flow regimes, and the addition of a released scalar incorporating bottom roughness geometry will allow greater accuracy in the determination of near-bed flow regimes. These results indicate that the spatial configuration of bottom roughness is an important determinant of gamete/larval transport in terms of whether the scalar will be retained or transported downstream. 133 Introduction Hydrodynamics are critical for mass transport of gametes and larvae for many aquatic organisms, benthic or otherwise, primarily due to the fact that most gametes or larvae act largely as passive particles entrained in the flow (Abelson and Denny 1997; Nishihara and Ackerman 2008). Much of the focus on benthic organisms has been on the role of turbulence in gamete/larval transport (Denny and Shibata 1989; Crimaldi et al. 2002; Chapters 2 and 3), as turbulence facilitates the encounter between eggs and sperm, and the dispersal and/or settlement of larvae onto the bed. Recent studies have shown that turbulence caused by bed roughness is one of the primary factors controlling flow and transport near the bottom (Crimaldi et al. 2002; Hendriks et al. 2006). Specifically, it is the small-scale spatial configuration of the roughness elements on the bottom or in the near-bed region that generate this turbulence (Eckman 1990; Hendriks et al. 2006; Chapters 2 and 3). Earlier models have reached contradictory conclusions concerning the effect of bottom roughness on larval settlement. Eckman's (1990) one-dimensional model theorized that larval flux and settlement would increase with roughness element spatial density, whereas Crimaldi et al. (2002) predicted that increased roughness element density, caused by decreased spacing between model clams, would reduce larval settlement rates. Quinn (Chapter 3) suggested that it was the nature of the spatial configuration of bottom roughness elements that determined the type of flow near the bed (i.e. skimming flow), which affected larval settlement. Simple measures such as density, discussed by Eckman (1990) and Crimaldi et al. (2002) do not account for different flow regimes, which are likely created by the spatial configurations of roughness elements 134 (review in Schindler and Ackerman 2009). The connection between gamete or larval transport and bottom roughness has been examined recently, although the importance of local small-scale hydrodynamics to these processes has been recognized (Yund et al. 2007; Reidenbach et al. 2009). Moreover, the physical relationship between spatial configuration of bottom roughness and the nature of flow has been modeled extensively, with some well-established principles related to the way flow regimes can be described. Flow regime classification Three types of flow regimes have been identified over a two-dimensional rough surface, based on the principal parameters of roughness height (k), water depth (d), the longitudinal distance between roughness elements, or roughness spacing (A), and roughness groove width (/) (Morris 1955; Fig. 4.1). These parameters help to determine whether the flow regime is characterized as: (a) isolated roughness flow, where the coherent flow structure (i.e., recirculation zone) created behind one roughness element is dissipated before the next element downstream; (b) skimming flow, where the spaces between roughness elements are small enough to prevent mixing between the fluid in the spaces and the fluid travels at a higher velocity above the spaces than in sparser arrays; and (c) wake interference flow, which acts as a transition between isolated roughness flow and skimming flow and where the wake structures of one element impinge on the next element, creating additional turbulence (Fig. 4.1). A fourth type of flow was added by Davis and Barmuta (1989) with chaotic flow found in shallow, rough flows where d < 3k. 135 Of the parameters mentioned above, roughness spacing (X) and roughness height (k) appear to be of particular importance. In 2-dimensional (2D) transverse bars in particular, perhaps due to ease of manipulating the modeling environment, the influence of the differing Xlk (or Roughness Index; Morris 1955) on flow is understood to some extent (Schindler and Ackerman, in press). For flows where Xlk < 8, the fluid in the space between bottom roughness elements becomes disconnected from the flow above, resulting in skimming flow (Leonardi et al. 2003; 2004). With flows of Xlk ~ 8, the wakes created by each roughness element interact with each other, which satisfy the definition of wake interference flow. Finally, for flows with Xlk > 8, wakes from the roughness elements do not interact with each other and the results are areas of intermittent recirculation in the regions immediately downstream of each bottom roughness element, or isolated roughness flow (Djenidi et al. 2008; Schindler and Ackerman, in press). The purpose of this study is to examine the Xlk prediction for flow over different types of roughness geometry and to examine the downstream transport of a released scalar under these flow regimes. The released scalar is used to model the transport of gametes or larvae from a benthic population, such as the dreissenid mussels of the Laurentian Great Lakes. By relating gamete/larval transport to bottom roughness, the role of physical parameters on biological processes and benthic population dynamics can be better understood. Methods Flow environment 136 The COMSOL multiphysics (version 3.4, COMSOL Inc.) computational fluid dynamic modeling program was used to examine how different bottom roughness parameters influence benthic hydrodynamics and subsequent scalar transport, as a proxy for sperm and larval transport. A 2D k-s turbulence model was used, with model solutions performed using the built-in segregated solver GMRES. Subdomain flow field parameters were set to fresh water at 20°C to simulate lake conditions (density (p) = 1,000 kgm" and dynamic viscosity (u.) = 1 x 10' Pas). The modeling environment consisted of a spatial domain 5 m long and 1 m high above the bottom. The bottom boundary was set to the wall logarithmic boundary layer condition, the left-most boundary was set to the inlet condition (U) and the right-most boundary was set to the outlet condition. All other boundary domains were set to the symmetrical boundary condition (i.e., to restrict the model domain without affecting the hydrodynamic characteristics on either side of the boundary). Three roughness element geometries were examined: (a) square, (b) rounded hemispheres, and (c) triangular transverse bars (see Figs. 4.1, 4.2, and 4.3 for examples, respectively). Roughness elements were placed on the bottom at various X apart and with various k, to obtain the different values of X/k under consideration, as described below. The influence of the number of bottom roughness elements on the flow regime was also examined using 3, 4, and 5 elements. The first roughness element, however, was always a minimum of 2 m downstream to allow for adequate flow development, evaluated through visual observations of the modeled velocity gradient indicating a well-developed boundary layer.. The spatial domain was modeled using a free mesh with 751 mesh elements that were generated with COMSOL. The free mesh approach concentrates mesh 137 elements around complex boundaries (Fig. 4.2). Five velocities (U= 10, 20, 30, 40, and 50 cm s"1) were examined as they are typical of the fully turbulent flow conditions characteristic of a freshwater mussel habitat (Ackerman et al. 2001). Bottom roughness X/k values The square bottom roughness geometry was the primary geometry examined, as the square form has been used classically to examine different flow regimes (Davis and Barmuta 1989; Young 1992). In terms of the roughness spacing, X, seven values at 10- cm intervals from 20 to 80 cm were used, combined with seven values of the roughness height, k, at 2-cm intervals from 2 to 14 cm, for a total of 49 combinations of XIk ranging from 1.43 to 40. This range of parameters was used to match the biologically-relevant roughness spacings and heights that have been observed for freshwater dreissenid mussels (Chapters 2 and 3). A biologically relevant subset of 16 out of the 49 X/k values was examined for the rounded hemisphere and triangle roughness geometries in order to determine the influence of roughness shape. In this case, four X values at 10-cm intervals from 20 to 50 cm and four k values at 2 cm from 2 to 8 cm were used. Scalar environment A 2D convection and diffusion model was applied to the velocity data provided from the 2D k-s turbulence models described above to examine the transport of a continuously released scalar over the roughness elements. In this case, the scalar was released from a point 1 m upstream of the upstream-most roughness element at the same k as the roughness element under examination (i.e., if roughness A: was 2 cm, then the scalar was 138 released at 2 cm). The point of scalar release was a flat boundary perpendicular to the flow and 5 cm high (see Fig. 4.3 for an example). A relative scalar concentration of 1 mol m" was used with a diffusion coefficient of 10" m s" , which has been reported for the sperm of other broadcast spawners (Reidel et al. 2005; Inamdar et al. 2007). The scalar concentration for a particular trial was determined by recording the modeled results at five points along a downstream transect that bisected each roughness element and taking an average of those five points (Fig. 4.3). The relative scalar transport (RT) for each trial was determined through the following expression, rc^ 'P RT = CQ (1) vQy vQy where Co is the initial scalar concentration at the point of release, C/ is the scalar concentration between the first and second roughness elements, and Q is the scalar concentration between the fourth and fifth roughness elements. A high RT(> 10) indicates that the majority of the scalar is being carried downstream (i.e., skimming flow) whereas a low RT(~ 1) indicates that the scalar is being trapped within the spaces of the roughness elements (i.e., isolated roughness). A slightly higher RT (~2) would indicate wake interference flow, as the majority of scalar is still trapped, but a larger portion would escape and be carried downstream, lvalues were compared among the different velocities and roughness geometries through one and two-way ANOVAs, and the relationship between 1/k and RT was determined through linear regression, using the statistical program STATISTICA version 6.0 (StatSoft Inc., Tulsa, Ok, USA). 139 Results Flow regime classification The number of roughness elements in tandem did not appear to affect the nature of the flow regime, as the flow regime behaved in a similar manner regardless of whether there were 3, 4, or 5 elements present (see Fig. 4.4 for an example). Consequently, 5 roughness elements were used for all remaining model trials. Increasing velocity did not appear to change the qualitative aspect of the flow regime (e.g., skimming, wake interference or isolated roughness flow), as the flow regimes remained similar based on visual observations of model results from models run from 10 to 50 cm s"1 (data not provided). All subsequent trials, therefore, used U= 10 cm s"1, corresponding to turbulent Reynolds numbers ranging from 2000 to 14000 based on Re = Ul/v where / = length scale (in this case k) and v = kinematic viscosity. The modeled flow regimes for the square geometry were compared visually to the predicted flow regimes based on the bottom roughness parameters (Fig. 4.5). Skimming flow predominated, found in the right side of the figure in gray colour, with the highest velocity skimming flows (i.e., the flow in the space was more isolated from the flow above the roughness) found in the bottom right corner. Isolated roughness flow was the next most frequent type observed, found on the left side of the figure in black colour, with a narrow diagonal band of wake interference flow around X/k of 8, acting as a transition between skimming and isolated roughness flows (see Fig. 4.1 for examples of model results at each flow regime). Generally, the model results followed the predicted flow regimes closely (Schindler and Ackerman 2009). In general, all three geometries agreed with the predictions of X/k, and had similar model results with respect to isolated roughness flows and wake interference flows, in that there was a separation between the flow regimes at X/k ranging from 8 to 9 (Fig. 4.6). Skimming flows were found for X/k up to 7 or 8, wake interference flows between X/k of 7 to 9, and isolated flows for X/k greater than 8 to 9 for the square and triangular geometries. The rounded roughness element geometries did not, however, appear to follow this pattern for skimming flow regimes, which were seen for X/k up to 6 (see Fig. 4.7). Scalar transport Velocity did not appear to have a significant effect on scalar RT values (Fig 4.8). When the three different flow regimes were compared, there was no difference among the five velocities within any of the flow types. No significant difference was found for the RT values among the five velocities under a two-way ANOVA (F^ 224 = 0.58, P = 0.93), but there was a significant difference among RT values across the three flow regimes (F2,224 = 7.28; P < 0.0001). There was no significant velocity x flow regime interaction. Note that 9 X/k ratios (i.e., the lowest values of X and k, ranging from 3.33 to 20) were used per velocity. RT values for scalar transport over the square roughness geometry were contoured in a similar manner as the flow regime classification. In this case, there was a clear separation between skimming and the other two flow regimes (Fig. 4.9). Skimming flow was found predominantly on the right half of the contour, matching with the X/k ratio predictions for flow regimes. The separation between wake interference and isolated 141 roughness flows was less obvious, a result of the relatively close R rvalues between the two flow regimes. This similarity is also evident in the comparison of RT values among the flow regimes (Fig. 4.10). A significant difference was observed for the RT values among the three flow regimes under a one-way ANOVA (F2,244 = 79, P < 0.0001). Pairwise differences were found between all three flow types (Fig 4.10). There was also a minor significant negative association between the X/k and RT(R2 = 0.23, P < 0.0001). Rrvalues for the triangular and rounded geometries were compared for the different flow regimes (Fig. 4.11A and B, respectively). Results for the triangular geometry were similar to those for the square geometry, in that a clear separation was evident between skimming flow and the other two flow regimes. The difference in RT values between wake interference and isolated roughness flows was not large, and the separation between the two flows regimes was not clear. Scalar transport over the rounded geometry followed the pattern for the skimming flow region. However, a larger region of isolated roughness and wake interference flow was evident, extending past a X/k ratio of 8, with skimming flow confined to the bottom right corner of the contour where X/k ratios were lowest. This would suggest that skimming flow occurs at a lower X/k for rounded versus triangular or square geometries. Significant differences in lvalues were found among the flow regimes for the rounded and triangular geometries under one-way ANOVAs (F279 - 19.8, P < 0.0001; and F2J9 = 12.6, P < 0.0001, respectively). Pairwise differences were detected between each combination of flow regimes (Fig 4.10). A two-way ANOVA of the RT values comparing roughness geometry and flow regime indicated a significant difference in RT values among geometries ^2,239 = 141, P < 0.0001) and among the flow regimes (F 15,239 = 142 515, P < 0.0001). There was no significant geometry x flow regime interaction. Pairwise differences were seen between the rounded and triangular geometries (P = 0.014) and between the round and square geometries (P = 0.03). Discussion The ratio of X/k appears to provide an accurate means of classifying roughness flow regimes as skimming, isolated roughness, or wake interference flows. COMSOL model results confirmed the predicted flow regimes for most of the bottom roughness parameters, especially the square and triangular forms. To the best of our knowledge, ours was the first study to examine the effect of different bottom roughness geometries and X/k ratios on released scalar, which allows a comparison to be made to aspects of gamete and larval transport of benthic organisms. This comparison builds on models used to examine sperm transport over roughness elements of different roughness height (Chapter 1). When roughness height (k) was manipulated systematically, the modeled results indicated that above a certain k, scalar (sperm) released within the space between two elements was retained within the space rather than being transported downstream. The manipulation of both roughness height and spacing, X/k, in this study helps to explain this result. Specifically, when the ratio of X/k is less than 7, skimming flow results, and the fluid between the spaces becomes isolated from the flow above, essentially trapping the released scalar (sperm) within that space. If the sperm is released outside of these isolated between-element spaces, however, skimming flow will transport that sperm at higher rates downstream than was observed for isolated roughness, resulting in relatively higher RT values. 143 Under isolated roughness and wake interference flows, scalar (gametes) released upstream would enter spaces between the roughness elements at a higher rate, and the resulting RT value indicates that a relatively higher proportion of scalar was retained versus transported downstream. RT rates were lowest in the isolated roughness flows, indicating that most of the scalar remained in the space downstream of the first few roughness elements and was not transported downstream. RT rates were slightly, yet significantly, higher under wake interference flow, indicating that the majority of scalar remained downstream of the first few roughness elements, but a higher proportion of scalar was transported downstream relative to isolated roughness flow. In terms of gamete release, a wake interference flow should be ideal, allowing gametes to enter areas (i.e., gaps between roughness elements) where a higher concentration of other gametes would exist, yet allowing gametes to be transported downstream where there would be an opportunity to encounter gametes from individuals further downstream. In the field, however, local populations appear to exist under conditions of skimming flow, i.e., roughness heights of 10 to 15 cm and roughness spacing of ~ 30 cm (Chapter 2). Whether gamete release occurs under particular flow conditions that would enhance gamete encounter via wake interference, such as those seen for marine algae (Pearson et al. 1998), remains to be determined. The relationship between bottom roughness and larval transport/settlement has led to some contradictory findings in the past. Eckman (1990) predicted an increase in larval settlement with roughness density, whereas Crimaldi et al. (2002) predicted a decrease. When the combined influence of roughness height and spacing on flow regime type is considered, however, their predictions can be interpreted in a different way. Eckman's 144 (1990) prediction holds true when one considers that increasing the number of roughness elements on the bottom would increase the opportunity that different flow regimes would be created. Further there would be an opportunity for areas of recirculation under isolated roughness and wake interference flow that would enhance settlement relative to a flat-bottom configuration. On the other hand, Crimaldi et al. (2002) used three adult clam spacings (analogous to X) of 3.0, 4.4, and 6.1 cm, and two heights (analogous to k) of 0.9 and 1.8 cm corresponding to X/k ratios ranging from 1.69 (3.0:1.8) to 6.78 (6.1:0.9). They found that larval settlement probability was lowest at the lowest spacing and highest height, i.e., X/k ratio of 1.69, which would correspond to a strong skimming flow regime, and high RT. Such high relative transport would carry larvae downstream before settlement could occur, confirming Crimaldi et al.'s (2002) predictions but ignoring the fact that lower X/k ratios would result in isolated roughness and wake interference flows, which would increase-settlement. In fact, Quinn (Chapters 1 and 3), found that mussel shells placed uniformly at high density on the bottom of a flow chamber led to skimming flow, in contrast to wake-interference flow observed under a low mussel-density configuration. It is important to note that we are assuming that similar scalar transport processes are responsible for the transport of gametes and larvae. The bottom roughness geometry that differed most from the predicted flow regimes characterized by X/k was the rounded geometry. The primary difference was that wake interference flow occurred over a wider range of X/k values than for the other two geometries, where such ratios would have predicted skimming flow for square and triangular shapes. Flows over sharp edges can lead to flow separation and areas of recirculation (Nowell and Jumars 1984). These areas of recirculation would be persistent 145 and would be restricted to the immediate upstream portion of the sharp edge (Nowell and Jumars 1984). The mechanism involved may be revealed from sediment transport studies in river environments. Best and Kostaschuk (2002) found that the symmetry of dunes in rivers was important in the type of recirculation zone found immediately downstream of the dune crest. Asymmetrical dunes demonstrated persistent recirculation zones immediately downstream of the lee slope of the dune, whereas symmetrical dunes demonstrated intermittent recirculation zones immediately downstream of the lee slope, with evidence of vortex shedding (larger eddies escaping the recirculation zone and continuing downstream). It was hypothesized that the angle of the lee slope (the downstream slope) was the main determinant of the recirculation zone created, and that symmetrical features typically had lower lee slope angles (Best and Kostaschuk 2002). In the current study, the rounded roughness would appear to have the lowest lee slope angle, and be most similar to a symmetrical dune. This would suggest that an area of intermittent recirculation would be found downstream of each rounded roughness element, and the presence of vortex shedding could be enough to disrupt the flow such that areas that normally produce skimming flow would act more like isolated roughness or wake interference flows. As the roughness height increases and the roughness spacing decreases (i.e., X/k drops), the likelihood that these vortices could disrupt the flow regime would diminish, and skimming flow would eventually be seen. The use of an idealized 2D model environment versus a more realistic 3D roughness model does have it limitations. However, it does allow for the relative ease of manipulating the physical model environment, an important aspect considering the number of roughness parameters tested, and the acquisition of model results as COMSOL 146 K-e model would not converge for velocities less than 1 m s~ , an unrealistic value for the natural conditions that the model was meant to simulate. Future work using 3D models is needed and will help advance findings from 2D models and determine whether similar ideas can be developed from the ratios of X and k. The inclusion of roughness geometry and roughness height and spacing in models of scalar transport provide more realistic depictions of near-bed flow regimes (Young 1992). Results from this study will prove valuable for understanding transport processes near the bed, which may be relevant to the management and conservation of benthic environments. For example, gamete and larval transport are critical life-history phenomena for most benthic organisms, so evaluating the impact of bottom roughness on these stages is essential. Results from this study can also be applied to invasive species where control of dispersal is important to limit potentially harmful access to an ecosystem. The approach used by this study may prove useful in a wide variety of applications related to benthic habitats. 147 References Abelson, A., and M. Denny. 1997. Settlement of marine organisms in flow. Annu. Rev. Ecol. Syst. 28:317-339. Ackerman, J.D., B. Sim, S.J. Nichols, and R. Claudi. 1994. A review of the early life history of the zebra mussel (Dreissena polymorpha): Comparisons with marine bivalves. Can. J. Zool. 72:1169-1179. Ackerman, J.D., Loewen, M.R., and P.F. Hamblin. 2001. Benthic-pelagic coupling over a zebra mussel bed in the western basin of Lake Erie. Limnol. Oceanogr. 46: 892- 904. Best, J., and R. Kostaschuk. 2002. An experimental study of turbulent flow over a low- angle dune. J. Geophys. Res. 107: 3135-3153. Crimaldi J. P., J. K. Thompson, J. H. Rosman, R. J. Lowe, and J. R. Koseff. 2002. Hydrodynamics of larval settlement: the influence of turbulent stress events at potential recruitment sites. Limnol. Oceanogr. 47: 1137-1151. Denny, M. W., and M. Shibata. 1989. Consequences of surf-zone turbulence for settlement and fertilization. Am. Nat. 134: 859-889. Davis, J. A., and L. A. Barmuta. 1989. An ecologically useful classification of mean and near-bed flows in streams and rivers. Freshwater Biol. 21: 271-282. Djenidi L., R. A. Antonia, M. Amielh and F. Anselmet. 2008. A turbulent boundary layer over a two-dimensional rough wall. Exp Fluids 44: 37-47. Eckman, J. E. 1990. A model of passive settlement by planktonic larvae onto bottoms of differing roughness. Limnol. Oceanogr. 35: 887-901. Hendriks, I. E., L. A. van Duran, and P. M. J. Herman. 2006. Turbulence levels in a 148 flume comparedto the field: implications for larval settlement studies. J. Sea Res. 55: 15-29. Inamdar, M.V., Kim, T., Chung, Y-K, Was, A.M., Xiang, X., Wang, C-W, Takayama, S., Lastoskie, CM., Thomas, F.I.M. and Sastry, A.M.. 2007. Assessment of sperm chemokinesis with exposure to jelly coats of sea urchin eggs and resact: a microfluidic experiment and numerical study. J. Exp. Biol. 210: 3805-3820. Leonardi, S., P. Orlandi, R. J. Smalley, L. Djenidi and R. A. Antonia. 2003. Direct numerical simulations of turbulent channel flow with transverse square bars on one wall. J. Fluid Mech. 491: 229-238. Leonardi S., P. Orlandi, R. J. Smalley, L. Djenidi and R. A. Antonia. 2004. Structure of turbulent channel flow with square bars on one wall. Int. Jour. Heat Fluid Flow 25: 384-392. Morris, H. M. 1955. Flow in rough conduits. Trans. Am. Soc. Civil. Eng. 120: 373-398. Nowell A. R. M., and P. A. Jumars. 1984. Flow environments of aquatic benthos. Annu. Rev. Ecol. Syst. 15: 303-328. Pearson, G.A., Serrao, E.A. and Brawley, S.H. 1998. Control of gamete release in flucoid algae: sensing hydrodynamic conditions via carbon acquisition. Ecology 79: 1725-1739. Reidel, I.H., Kruse, K. and Howard, J. 2005. A self-organized vortex array of hydrodynamically entrained sperm cells. Science 309: 300-303. Reidenbach, M. A., J. R. Koseff, and M. A. R. Koehl. 2009. Hydrodynamic forces on larvae affect their settlement on coral reefs in turbulent, wave-driven flow. Limnol. Oceanogr. 54:318-330. 149 Schindler, RJ. and J.D. Ackerman. 2009. The environmental hydraulics of turbulent boundary layers. In: D.T. Mihailovic and C. Gualtieri (eds.) Advances in Environmental Fluid Mechanics. World Scientific, London. 40 pp. Young, W. J. 1992. Clarification of the criteria used to identify near-bed flow regimes. Freshwater Biol. 28: 383-391. Yund, P.O., Murdock, K. and Johnson, S.L. 2007. Spatial distribution of ascidian sperm: two-dimensional patterns and short vs. time-integrated assays. Mar. Ecol. Prog. Ser. 341: 103-109. 150 1 • i i *• J •~~^<^ZJLZZ^ZZJZZ3 LlllJ L^ULJ-- IJI^J^LJ 'Sum HI ' Lw^tefr^—^w^^^.^^.^rr ^ni-- —•nri-—iTi i--— n -- -n --- Figure 4.1. COMSOL flow streamline plots illustrating the three main flow regime types over 2D transverse square roughness elements: (A) isolated roughness over A/k =12; (B) skimming flow over A/k =3.3; and (C) wake interference flow over A/k = 8.3. Also shown on (A) is the roughness parameters of roughness height (k), water depth (d), the longitudinal distance between roughness elements, or roughness spacing (A), and roughness groove width (J). Note that the panels represent a portion of the model that illustrates the flow around the element. 151 Figure 4.2. COMSOL plot of the modeling environment for 2D transverse round roughness elements of A/k= 6.3 and the model free mesh structure, which adds more mesh elements around a complex boundary, indicated by the smaller and smaller mesh elements seen around the round roughness elements. 152 Sj-'dtt;: QiiLU'lC-dliji, c. jnu'/'n j Figure 4.3. COMSOL plot of the released scalar concentration over 2D transverse triangular roughness elements. Scalar was released at a point 1 m in front of the first roughness element and at the same height of the first roughness element to a distance of one roughness height above it. The scalar was released from a 5-cm high flat interior boundary perpendicular to the flow. The Afk ratio is 6.25, corresponding to a predicted flow regime of skimming flow, which was confirmed by the scalar carried downstream with limited fluid entering the spaces between roughness elements. The white dashed line between the first two elements indicates the location of the R rvalue transect. -"-'•'r^.c—ir—--'. ^i—T--.-~'~ " "'r-r^~-_ "" "~"i—r~.--~ Figure 4.4. COMSOL plot comparing the number of bottom roughness elements (A; three elements, B; four elements, and C; five elements) downstream on the nature of the flow regime. The parameters shown are A, = 70 and k = 6, giving a X/k ratio of 11.7 for all three plots. The predicted flow regime of isolated roughness is illustrated, which matches with the flow regime generated for all plots. 154 80 IsoUted roughness flow i ll_k-S 70 f j 60 o C Wake intej^j o ro w 50 w w • .<:•:•' 30 w 20 f 1 i i i 6 8 10 12 14 roughness height (cm) Figure 4.5. The flow regimes created from the ratio of roughness spacing (A) to roughness height (k). Isolated roughness flows are found to the left of the A/k ratio = 8, skimming flows are found to the right of that line, whereas wake interference flows are found right around the line A/k = 8. 155 - A L_- -^V^-V-,V\-^~^.-;--r_-:^r7-:l:7^:,-'---r-: Figure 4.6. COMSOL plots indicating a /1/fe ratio of 10 for the (A) triangle, (B) square, and (C) round roughness element geometries. Isolated roughness flow was observed for each geometry as indicated by the small flow recirculation regions behind most of the elements that dissipate before the next element downstream. 156 Figure 4.7. COMSOL plots indicating: (A) X/k ratio of 6.25 illustrating wake interference flow over 2D transverse round roughness elements; and (B) XJkxdXxo of 6.25 over 2D transverse square roughness elements illustrating skimming flow; and (C) X/k ratio of 2.5 over 2D transverse round roughness elements illustrating skimming flow. 157 B 10 cm s"1 I 1 I I ?0 cm s" 4 A 1 i i r^n cm s" 1 I I 40 cm s" 1 XL I I 50 cm s" o c 05 i_ ••-» i ^ 2 skimming wake interference isolated roughness flow regime classification Figure 4.8. The relationship between velocity and scalar relative transport within three flow regimes over square roughness elements: skimming flow QJk < 7); wake interference flow (7 < \/k < 9); and isolated roughness flow (klk > 9). Values are means ± SE, N = at least 5. 158 Relative transport (RT) 6 8 10 12 14 roughness height (cm) Figure 4.9. RT contours for the ratios of bottom roughness spacing (k) to roughness height (k) over square roughness elements. The lighter contours indicate higher RT values. The dashed white line represents the region of predicted wake interference flow (i.e., A/k = 8), with skimming flow to the right of the solid line and isolated roughness to the left. The solid white line represents the RT= 16 threshold value observed for skimming flow. The white dotted lines in the bottom left corner indicate the subsection of A/k ratios and contour plots used for the triangle and round roughness geometries. 159 60 I ••I round i i triangle 50 H i-;=vt-':-i square - 16 ratio set i i square - full data set 40 A 30 H 20 I 10 i P1.BB D skimming wake inference isolated roughness flow regime classification Figure 4.10. The relationship between roughness element geometry and scalar relative transport for skimming flow (A/k < 7), wake interference flow (7 < A/k < 9) and isolated roughness flow {A/k > 9). Note the different values for the square geometries using the 49 A/k ratios, and the 16 A/k ratio subset. Values are means ± SE, N = at least 10. 160 50 S -•''•'"' A ;; 5 ft l! ^*;^ '" Relative transport (RT) t jj '! /:. i • f / :, j ffgffl@s@/ j 0) c .c en / / :. 'j :.' I ^o 30 - / : .i: J :' / / ,j ff '! /. 20 0 i j 1 jj . •oughness neight (cm) *^n ou B 1 ^ V P I F . J '/ II i 0 " 1 J ^ ', Relative transport (RT) • 2 30- «? r~" 0 i 0 i 3 i 1 • 161 GENERAL CONCLUSIONS From the conclusions drawn from the results of the four chapters, the hypothesis that bottom roughness is important to fertilization success and larval settlement of freshwater dreissenid mussels appears to be valid. Chapters 1 and 2 demonstrated that bed roughness influences external fertilization success. Bottom roughness elements influenced the extent of sperm limitation that a broadcast spawner might encounter, and the configurations or types of roughness elements determine the nature of turbulence seen near the bed, with ejections being beneficial to fertilization and skimming flow possibly reducing its success. On the other hand, the dreissenid mussels appear to have higher sperm potency compared to other broadcast spawners. This higher reproductive potential reveals that dreissenids have the ability to withstand more variable flow conditions and may explain their success as an invasive species. Ultimately, the influence of bottom roughness on the near-bed flow environment appears to be the primary factor controlling sperm dilution. The bottom roughness features responsible for influencing fertilization also influenced the extent of larval transport and settlement. Spatial bottom roughness configurations due to the presence of mussels play a major role in producing skimming or wake interference flows and the related turbulent ejections and sweeps, which can inhibit or enhance larval settlement, respectively. These similar turbulent events also appear to influence larval resuspension off the bed, but to a much lesser degree. The ratio of X/k appears to provide an accurate means of classifying roughness flow regimes as skimming, isolated roughness, or wake interference flows, as the COMSOL model results confirmed the predicted flow regimes for most of the bottom roughness parameters, especially the 162 square and triangular forms. With larval transport, there also appears to be an active component involved in the resuspension of dreissenid larvae, which increases their ability to move from one region of the bed to another. The spatial bottom roughness configuration and turbulence relationship may explain another aspect not addressed in the previous chapters, the aggregating population structure of certain species. The influence of established adult populations on larval settlement is another aspect of bed roughness that has been associated with hydrodynamics in determining population structure (Andre et al. 1993; Thrush et al. 1996; Crimaldi et al. 2002). With the dreissenid mussels, however, the impact of established adult populations may be an example of ecosystem engineering. One dramatic effect of invasive species such as the dreissenid mussels is the alteration of ecosystems, and one way this can occur is through a change of the physical structure of the ecosystem. Invasive species can affect the availability or quantity of physical resources, and ecosystem engineering represents the primary means by which physical resources can be controlled directly or indirectly (Crooks 2002). The flux of larvae entrained within water currents could be considered a physical resource, and by modifying this resource through changing the physical structure of the bed, the dreissenid mussels can be classified as ecosystem engineers. Previous computational fluid dynamic modeling has shown that the influence of physical structures on water currents is related to the roughness element height of those structures on the bed. The size of the mussel cluster was manipulated, and the larger the cluster, the more particles were retained within the roughness elements (Chapters 1 and 2). 163 These results confirm the previous models, confirming that changes in bed structure caused by mussel clusters can enhance larval settlement and anchoring probability, a possible example of "extended phenotype engineering" as these ecosystem changes benefit the fitness of the species (Jones et al. 1994). These hydrodynamic elements should also act to limit the extent of larval export by reducing the ambient velocity. Results from this thesis will greatly enhance understanding of the relationship between physical hydrodynamic forces and ecology. Identifying the importance of the spatial bottom roughness configuration on critical life-history stages, such as gametes undergoing external fertilization and larvae undergoing transport and settlement, will hopefully lead to improvements in conservation efforts of benthic species and prevention/reduction of harmful species invasions. References Andre C, Jonsson PR, and Lindegarth M. 1993. Predation on settling bivalve larvae by benthic suspension feeders: the role of hydrodynamics and larval behaviour. Marine Ecology Progress Series 97: 183-192. Crimaldi JP, Thompson JK, Rosman JH, Lowe RJ, and Koseff JR. 2002. Hydrodyanmics of larval settlement: the influence of turbulent stress events at potential recruitment sites. Limnology and Oceanography 47: 1137-1151. Crooks, J. A. 2002. Characterizing ecosystem-level consequences of biological invasions: the role of ecosystem engineers. Oikos 97:153-166. Jones, C. G., J. H. Lawton, and M. Shachak. 1994. Organisms as ecosystem engineers. Oikos 69:373-386. Thrush SF, Hewitt JE, Pridmore RD, and Cummings VJ. 1996. Adult/juvenile interactions of infaunal bivalves: contrasting outcomes in different habitats. Marine Ecology Progress Series 132: 83-92. 164