WHOI-92-26

e- :2

Particle ,Contact on Flat Plates 1n Flow:

A Model forlnitlal Larval Contact'

by

Elizabeth D. Gariand and Lauren S.Mullneaux

June 1992

Technical Report.

FrJndingwas provided by the Ofice of Naval Research under Contract Nos.I\00014-89-J..1431 and N00014-89-J-11 12. .

Approved for public release; distribution unlimited. - .

/' / WHOI.92.26

Particle Contact on Flat Plates in Flow: A Model for Initial Larval Contact

by

Elizabeth D. Garland and Lauren S. Mullineaux

Woods Hole Oceanographic Institution Woods Hole, Massachusetts 02543

June 1992

Tecca Report

Funding was provided by the Offce of Naval Research under Contract Nos. NOOO14-89-J-1431 and NOO014-89-J-1l12.

Reproduction in whole or in part is permtted for any purose of the United States Government. This report should be cited as Woods Hole Oceanog. Inst. Tech. Rept., WHOI-92-26. -=------cO Approved for public release; distribution unlimited. - - .J - -ir .J - Approved for Disbution: _o ..-.J cO i_-".- -0- 0 ~or_ -r= 0 --0- rn -- o ~ (¿. a ~ ---- 90:1 c. Goldmn, ~ir =- Department of Biology

Contents

List of Figues ...... 2

List of Tables ...... 2

1 Introduction ...... 3

2 Methods ...... 4 2.1 Plate Design and Flow Measurements ...... 4 2.2 Parcle Contact Experients ...... 5 2.3 Statistical Analyses ...... 9

3 Results ...... :...... 10 3.1 Flow Over Plates ...... 10 3.2 Parcle Concentrations in the Flume ...... 13 3.3 Parcle Contact Patterns ...... 13 3.4 Parcle Contact Rates ...... 20

4 Discussion ...... 21 4.1 Parcle Contact Patterns in a Developing Bounda Layer ...... 21 4.2 Implications for Laral Settement ...... 22 4.3 Parcle Contact Rates in Non-Varg Flow ...... 23 4.4 Interpretation of Laal Settement Plates ...... 24

5 Acknowledgements ...... 25

6 Literature Cited ...... 25

1 List of Figures

Figure 1. Plate design and qualtatve boundar layer flow patterns ...... 4

Figure 2. Design of the paddle wheel flume ...... 6

Figure 3. Quantitative flow measurements over experienta plates ...... 11

Figure 4. Parcle concentratons in the flume durg replicate rus ...... 13

Figure 5a. Parcle contat patrns on faied plates ...... 14

Figue 5b. Parcle contat patrns on bluff plates ...... 15

Figue 5c. Parcle contat paterns on split plas ...... 16

Figue 6. Parcle contact rates on faied, bluff and split plates ...... 21

Figure 7. Parcle contact rate as a fuction of along-stream flux ...... 21

List of Tables

Table 1. Parcle concentraton in replicat flume rus ...... 17

Table 2. Kendall's 't correlation between parcle contat and flow ...... 17

Table 3. F ratos from ANOY A of cross-stream parcle abundances ...... 19

Table 4. F ratos from ANOY A of downstream parcle abundances ...... 20

2 1 Introduction

Predictig locatons where parcles wi fit contat a flat surace is possible in steady, unidirectional, boundar-layer flows by using momentu flux as an analog to parcle flux (e.g., Schlichtig, 1979). Predctig parcle behavior in flows over complex suaces is more dicult (Paola, 1983; Middleton and Southard, 1984). Paricle contact with a surace in flow is determed by parcle characeritics, such as siz, density and sing velocty, and flow characteritics such as fr-stream flow velocity, tubulence, shear strss and flow separatons in the boundar layer. Hydrodynamc processes that determe paricle contact with a surace also may inuence intial settement of benthc invertebrat larae in controlled labratory flows (Hanan, 1984; Butman, 1986, 1987; Mulleaux and Butman, 1991) and in the field (Duggis et al., 1990; Keen, 1987; Mulleau and Butman, 1990; Mulleau and Garlad, in prep). It is oftn difcult, however, to determe which flow characristic is responsible for the resultig settlement patrn. Furermore, the post-contac exploratory behavior of laae can alo case deviatons from settement pattrns expectd from hydrodynamc consideratons alone.

The utity of using the flat plat approach in controlled laboratory flows is that dierent configuatons of the leading edge of flat settement plates alow alteration of boundar-layer flows without changig the character of the surac. Leadg edges can be constrcted to generate a varety of boundar-layer flows diferig in boundar-layer thickness, tubulence, plae-ward advection, and shear stress. Although these characristics co-var downstream from the leadg edge in a clasical boundar layer over a th flat plat, they do not always co-var in flows over more complex surace geometres. In some of these cases, each flow chareritic can be isolated, and its effects studied independently. Because movements of relatively sma parcles ar strongly inuence by flow (e.g., Nowell, 1983), sites and rates of intial conta onto plates should be predictable from boundar-layer flow characeristics. Contat of parcles with plats in these flows is, in tu, a useful analog for initial contat of passive larae onto sim suracs.

The objective of the present study was to quanti: (1) the location, relative to the leadg edge, where smal, low-density parcles fit contat a flat surace in a developing boundar layer (based on varg downstream tubulence, shear stress and vertcal advecon); and (2) the relatonship between contact rat of parcles on flat suraces and parcle flux paralel to the surace in a uniform, fully-developed, boundar layer. These contac rates and patrn were then used to model initial contac of passive larae and propagues onto flat plates. Relatively smal (c: 100 J.m)~ low-density larae ar the focus of ths study for two reasons: (1) they ar more likely to be passive in boundar-layer flows than large larae beause their swimg speeds tend to scale with their siz; and (2) small larae, hydrodynamcaly simiar to the parcles used in ths study, are promient in the deep sea (Muleaux, 1992) where these models wi be teste. Ths study was designed to help defie the role of flow in laral contat and settement onto natual suraces in deep-sea habitas and for evaluating collecion characristics of laal settement plats. If, for example, laal contac rats can be predicte from laral concentrtion in controlled laboratory flume flows, then settlement rates onto plaes can be used to esti quantitative laral abundances in the field, assuming that larae intialy contact plates lie passive paricles (Muleaux and Butman, 1991).

3 2 Methods

2.1 Plate Design and Flow Measurements

hi order to correlate paricle contact onto suraces with specifc flow characeritics, flat plates were constrcted with the types of leading edge configuons: "faied", "bluff', and "split" (Fig. 1). Each leading edge configuation generated unique boundar-layer flows that could be used to determe flow characteritics that may be responsible for inuencing parcle contact with plates. All thee plates were 10 cm wide by 26 cm long and were constrcted from polycarbonate plastic sheets.

, em Falred~-- Plate------L. Figure 1. Plate design and quaitative velocity vectors drwn frm LDV meaurments of vertca Bluff Plate and horionta velocities at a along-str flow sp of 10 em s.l (meaurd at 10 em above flume ~~=-)~------~------bottom). Flume flow is from left to right in diagr.

Split Plate o ------, '-- - - I ~~=.. '- - -

The the plate types were confgued to develop the bounda layer flows descrbed in Mullieaux and Butm (1991). The "faied" plate was 0.15 cm thick along its enti length, with the leadg edge tapered to crate a faig for the oncomig flow. A classical bounda layer, descrbed by "flat plate" theory (e.g., Schlchtig, 1979) was expeted to develop along this plate. The "bluff' plate was 1.0 cm thck along its lengt. Over ths plate, the flow was expected to separate at the bluff surace of the leadg edge and reatth downstram at a location determed by the along-stram velocity and leadg edge thckness (Ruderich and Femholz~ 1986). The "split" plate was 0.15 cm thck along its lengt but had a "splitter bar", consisting of a 1.0 cm high polycarnate strp atthed norm to the flow, at the leading edge. Flume studies (Mullineaux and Butman, 1991) showed that the flow separation created by ths confguration was qualtatively simar to that of the bluf plate, but difere in size and tubulent intensity. In addition, the splittr bar caused the mean horizonta velocity and shear stress immediately downstram of the leadng edge to be substatialy reduced relative to simlar locations on the bluff and fai plates.

4 Flow over the thee plate tyes was measured using a two-axs, forward-scattr laser Doppler velocimeter (Agrawal and Belting, 1988). The plates were set up individualy for LDV measurements in an open-chanel flow (see Butman and Chapma (1989) for a description of the fle and Mulleaux and Butman (1991) for a detaed descrption of profùing above a flat plate using the LDV). The flume was 17 m long and 60 cm wide, with an experiental "test section" located 12 m downstream frm the flow strghteners. The flume was filed with frshwater to a height of 12 cm and the flow was adjusted to one of the two flow speeds (5 or 10 cm s-l at 10 cm above the flume bottom).

Single plates were placed in the flume at the center of the test section, with the plate edges 25 cm from the flume side-wals. The plates were elevated 5 cm above the flume bottom on thin plastic legs in order to raise them above the region of highest shear with 2-3 cm of the bottom. Plates were oriented horizontay with their long diension paralel to the flow. Vertcal velocity profies were measur with the ilV at successive locations downstram from the leadig edge. The two axes alowed for simultaeous measurments of instantaeous along-stream (u) and vertcal, or plate-war (~v~ities. Velocity means (E w), standad deviations (rms(u'), rms(w')), and varances (u' , w' ) were calculate frm measurements taken for 4 miutes at each position. Turbulence (rms(w')) and mean vertcal advection (w) were calculated by integrtig over the bounda layer from measurments taken at eight heights between 0.1 and 2.0 cm above the plate, as in ~~eaux and Butm (1991). Shear stress was calculate frm the tubulent kinetic energy (u' ) at a height of 0.1 cm above the plate. The downstream locations and spacing of profies vared slightly among plate types and flow speeds, and depende on the presence and location of flow separation and reattchment points. No velocity profies were measur over plates at 2 cm s-1 flows because of litations of the flume pump at slow speds. Instead, dye was released at the leadig edges of the plates at an along-stream flow spee rangig from 2 to. 13 cm s-l, in order to visualize flow separations and the bounda-layer thckness. These studies were useful for determing crtical flow spees for development and destabilzation of flow separations over each plate.

2.2 Parcle Contact Experiments

Parcle contact experiments were performed using the same experienta plates but in a different flume (Fig. 2). A raetrck-style flume (the Padde-Wheel Flume descrbed in Ij detail in Butman and Grassle, submitt) was chosen for the parcle contact studies because ;1 of its smaler volume (approxiately 1800 liters as opposed to approxitely 100,00 liters in the 17-Meter Flume). The number of parcles reuired to see the flow in the Pade- Wheel Flume was almost 20 times less than the number need to achieve simar concentrtions in the 17-Meter Flume.

The Paddle-Wheel Flume was 8.5 m long and 50 cm wide. Eight pades attched to flexible linkages revolved on a large padde wheel and drove the flow at a constat velocity. The designated "test section" where flow distubances from the chanel bend had dissipated, was located 4.2 m downstream from the fist bend following the paddes, and 1.5 m (thee channel widths) upstram from the downstram bend. Flow speeds were adjusted by a

5 B

A c

1

0'

Figue i. Schematic drwig, aproximately to sce, of the Pae Wheel flume use for the parcle contat studies. Loatons A, B, and C ar

regions where pump saples were taen to deteine pacle dittibution in the flume. The "te setion" where experienta pla were pla is loc at C. Ar\)s reprnt dition of flow in the flume. Flow separs ar loc at both end of the flume to reuce cr-sir flows generate whie water flow ard beds (Figue fr Butm and Grasle, submitt). rheostat that controlled the paddle wheel motor and were veried with a Marsh McBurey electromagnetic curent meter. The flume was filed to a height of 14 cm with 10 J. fùtere seawater at ambient room temperatu and was alowed to war up and deaerate overnght. This procedure was necessar because infowig seawater at ambient winter-tie temperatus was supersaturted with gasses. Deaeratig the water overnight prevented bubbles that interfered with parcle retention on the plates from formg durg the experienta runs.

Due to the water depth in the flume (14 cm), cross-stream flow at the bottom and along the surace was expecte (Butman and Grssle, submitted), but because the plates were near the center of the water column (5 cm above bottom), ths crss-stram flow was not expected to quantitatively affect the outcome of this study. Furermore, cross-stram flows should have dissipated at the test section.

Spores from the club moss Lycopodium sp. (Duke Scientic Company, prouct #215, Palo Alto, CA) were selected for use as parcles in this study because of their low gravitational fal velocities and their low cost. Lycopodium releases pale yellow spors tha~ when hydrated, ar 40-70 J. in diameter. The spore density in frshwater was 1.03 g cm- (Duke Scientific, pers. comm'.). The fal velocities of these spores were expeted to be in the range of passively sing, deep-sea larae and propagules, most of which ar approxiately 64-250 Jl in diameter (Mullneaux, unpub. data), and have prected settg velocities of -= 0.1 cm s-l .

A stock miture of parcles and seawater was mad by mig 0.25 - 0.50 g of dred spores with approximately 20 ml of 0.2 Jlm fùtered seawater. The mitu was shaken vigorously unti most of the spores were suspended. A few drps of Betae surgical scrb were added to relieve surace tension. The spores were then alowed to hydrte for one hour at room temperature. A small aliquot was drawn frm the stock mitu, loaded into a hemacytometer, and the average number of spores per miter determed from the mean of six sub samples.

The flume was seeded with a known concentration of the spore-contag stock mixture. The stock mitue volume was adjusted to yield a fmal concentrtion of 75 spors per millliter in the flume, and then was diluted to 4 liters and mied vigorusly before introduction into the flume. Spores were distrbute homogeneously thoughout the flume water volume by pouring the stock mixture along a full circuit of flume, in the diection opposite to the flow at 10 cm s-l to miimize deposition on the flum bottom. The spors were alowed to circulate two times around the flume before the rheostat was readjusted to attain the desired flow speed.

Durng the experiental runs, parcle concentrtions in the flume were measur at plate height (5 cm above the bottom). Parcles were collected with a peristatic pump sampler simar to that descrbed by Hannan (1984). Smal water samples, of known volume (150 - 250 ml), were drawn from the flume though a th glass tube (6 mm in diameter), bent so that the intake faced dictly into the flow. The tube, which was the only par of the apparatus disturbing the flow, was positioned vertcally with a verner calper. Spors in the

7 pump samples were retained on a 1.2 ,.m Mipore fiter, and counted with a dissectig microscope. Pror to and imedately followig the experienta rus, parcle concentrations were measured at thee heights above the bottom in thee locations arund the flume (locations A, B, and C in Fig. 2) to determne whether parcles were well-distrbute, both vertcally and horizontally, thughout the flume durg the course of an experienta run. No consistent trnds in parcle concentrtion, such as an increase in parcles near the bottom due to setting, or near the iner, lower corners of the flume due to cross-stream flows (as described by Butm and Grassle, submitted), were observed.

The the plate treatments (faid, bluff, and split) were coated with a th layer of Dow Coming 340 Heat Transfer Compound (DC 340), a white, viscous, silcon-based adesive that retained parcles in locations where they first contacte the plates. DC 340 was chosen over a varety of other coatigs on the basis of five crteria. When applied to the plates in a thin coat, it: (1) contrasted in color with the parcles; (2) formed an even, flat surace; (3) was insoluble in seawater and retaed its adesive propertes over tie; (4) was nontoxic for use in the flume, where live anis were frequently used, and (5) did not deform under high along-stram flows or shear stresses. Erosion experients demonstrted that once a parcle contacted' a coated surace, the parcle could not be dislodged even at shear strsses much higher than those occurg durng experienta rus (unpub. data).

The thee experimental plates, coated on their upper suraces with adhesive, were chiled to a temperatu that matched the seawater in the flume (8.5 - 12.50C). Flume water depth averaged about 14 cm, but ranged from 13.0 to 14.5 cm; thus, the width-to-depth ratio (3.85 - 3.45 respectively) was subopti for one-densional, open-chanel flow (Nowell and Jumar, 1987). Potential effects of wal bounda layers were adessed experientay with flow visualizations and parcle contact analyses. Plates were add to the flume afr flow spee had been established for each experient. Thee plates were placed in the test section horizontay (side-by side, in a random ordr) and elevated 5 cm above the bottom on thin legs. Dye studies were conducted to determe the horzonal spacing needed to dissipate the vortces generated at the corners of the plate leadg edges before they interfered with the flow over adjacent plates. The dissipation distace, of approximately 3 cm, resulted in 7 cm clearnce from each flume wal. Dye studies showed that this arangement miimze side- wall flow effects, while keeping the leadng-edge flow distubances from inuencing adjacent plates.

Durng an individual run, thee separte experiments were conducte, one at each of three flow spees (2, 5, and 10 cm s-l). In order to conserve spores, the flume was not emptied and reseeded between the the flow speed treatments; instead between tratments, flow speed was increased to the maimum settig (approxitely 20 cm s-I), and a squeegee was used to resuspend parcles that had setted out of suspension. Flow speed was then decreased and alowed to equilbrate to the new settg before the next set of plates was introduced. An estiate of spore concentrations durg each replicate ru was made by pumping water 5 cm above the bottom (plate height) at a location 4 m upstream of the test section (location B in Fig. 2). Nine pump samples were collected durg each replicate ru (the samples in each flow speed treatment) to determne whether parcle concentrtion in

8 the flume changed signifcantly between subsequent flow tratments, and between'replicate runs.

Parcles were allowed to accumulate on the plates for 30 miutes. A fresh set of three plates was introduced at the sta of each flow spee tratmnt. The positions of the three plate types, relative to the inner and outer wals of the flume were radomized for each flow speed treatment. In addtion, the temporal order of flow treatments was radomized for the replicate runs. Flow spees were measured at 10 cm above the bottm and veried before each flow speed treatment with a Marsh McBurney curent meter. Bounda shear velocities of the flume flow were chosen to cover a rage of natualy occurg deep-sea (slow) flows.

After a 30-min replicate run, plates were removed from the flume. Parcle accumulation patterns within 3 cm of the leading edge of each plate were quantied by counting spores under a dissectig microscope in successive 0.5 cm bands downstram frm the leadng edge. Only the central 6 cm in a band were counte to avoid regions near the edge that could have been affected by seconda flows (e.g., roll vortces in Ruderich and Fernholz, 1986). Between 4.5 and 20.5 cm from the leadg edge of the plates, spoes were counted in 1.0 cm bands, centered at 2.5 cm intervals frm the leadg edge. The bands were narower near the leading edge because strong downstream gradents in flow charteristics were expected in that region, whereas flow gradients were expeted to be more gradual furer downstream. Each 0.5 or 1.0 cm band was furer subdvided into six 1.0 cm cells across the plate to quantify cross-stram parcle distrbutions. Parcle abundaces on the plates from each flow treatment, Pp(i)' were adjusted to normalze values P'pfi) by multiplying them by the ratio of tle mean paricle concentrtion in the flume from al replicate runs /P wi to the flume parcle concentrtion durg the tie interval of the replicate run (P w(i;'

P'p(i) = Pp(i) . /P wil/P w(i)l

This normization permtted comparson of parcle abundaces on plates between flow speeds and between replicate runs.

2.3 Statistical Analyses

Comparsons between parcle abundaces and flow charcteristics were conducte with Kenda's 't corrlation coefficient (Siegel, 1956). For each plate tye in the two faster flow tratments (5 and 10 cm s -1), mean parcle abundaces in the eight bands between 0 - 8 cm were calculated using the thee replicate runs. These mean abundaces were raned for comparson with raned values of the thee flow characteristics, (vertcal (plate-ward) advection, turbulence and shear stress), from comparable locations on the plates. In cases where flow measurements had not been made at the same location as parcle counts, values were interpolated from the closest measurements. Simiar comparsons between parcle contact and flow characteristics at 2 cm s-l flow spees were not performed because detaed flow characteristics had not been measured at ttis flow speed.

. 9 Addtional analyses were conducted for the sole purose of investigatig possible flow biases in the experients. Cross-stream varations in parcle abundace were analyze with a one-way ANOVA (Systat, version 2.1) on each of the 27 plates (thee plates in each of three flow spee tratments in th replicate runs). The analyses were done separately for the leadg edge region (0-8 cm) and the mid-plate region (10-20 cm) of each plate, to isolate effects in the developing bounda layer frm those in the relatively unor bounda layer downstram. Cross-stram varations in parcle contact in the leadg-edge region were of interest to evaluate the potential inuences of seconda flows (e.g., roll vortces) prouced at the leadng edges of the plates. Cross-stream varations in the mid-plate region were expeted to be less inuenced by leadg-edge distubances, but were examed to determe if cross- stram flows in the flume channel influenced parcle contat.

Downstram varations in parcle abundace (i.e., at the leadg edge versus the mid- plate regions of the 27 plates analyze for cross-stram effects) were also analyze separtely for each plate tye, using an one-way ANOV A. Replicate plates were not used as replicates in the analysis because they had been placed in a dierent position in the flume (i.e., iner, middle, outer) durg each replicate ru. Downstram parcle abundaces were expeted to co-var with flow grdients in the leadig edge region of plates, but parcle abundaces were not expected to var in the mid-plate region, where there was only moderate downstram varation in bounda-layer flow characteristics. The one-way ANOV As tested ths prediction so that parcle abundaces in the mid-plate region could be pooled for furer analyses. Since the reliabilty of al above analyses depended on the assumption that parcle abundances in contiguous bands on a plate were independent, the results were interpreted cautiously.

3 Results

3.1 Flow Over Experienta Plates

Bounda-layer flows measured by the LDV over the thee plate tyes at 5 and 10 cm s-1 flow speeds (Fig. 3) corresponded to those predcted by classical bounda-layer theory and measured in open-chanel flows (Kiya and Sasak, 1983; Ruderich and Fernholz, 1986). On the faid plate, no flow separtion was detected. At both 5 and 10 cm s-1 flow spee, a classical bounda layer developed, with bounda thckness increasing with distace downstram from the leadg edge. Vertcal advection (away from the plate) deeased downstram from the leadg edge as the bounda layer grw, but did not drop below zero (i.e., where vertcal advection would be toward the plate). Turbulence and shear strss tende to be elevated slightly at the leadg edge where the bounda layer was th and a slight bow wave may have existed. The exception was shear stress at the leadig edge in the 5 cm s-1 flow, which was lower than over the rest of the plate. Ths anomalous result may have been due to the measurement being taen slightly above the bounda layer, rather than with it (Mullineaux and Butman, 1991). Furer downstram, tubulence and shear stress deased to lower, relatively constat values. Turbulence and shear strss tended to be higher in 10 cm s-1 than in 5 cm s-1 along-stream flow spees, and were higher over the bluff and split plates

10 5 em 8-1 10 em 8-1 4 4 V-ertical Advection (W) 3 , Vertical Advection (W 3 ,. ~ 2 2 ,\. e\ 1 'l~ "~ \'\ l?"a_ . I--t ------~------o -..~~-- o ~.;..i:~-~' -1 -1 -2 -2 -3 0246 8 W~UWU~ -3 0246 8 W~UWU~ 5 5

Turbulence (rms (wi)) ...... "" Turbulence (rms (Wi)) 4 4 ,!, ,,_, "'''''-, 3 3 /...... -__,__

2 ß'...t------._____ 2 /~/.-"'--'-..."'---..,,-.,-'------., ------~ ~,i,;.."--,.----".--,---~~-=.:::=====-:,:,::,~-~-=::::-:,:: . . __ 0 o 0 2 4 6 8 10 12 14 18 18 20 0246 8 w~uww~ 8 8 Shear Stress (U,2) 7 7 . Shear Stress (ur2) 6 6 ! .. 5 5 -, 4 4 . 3 3 , ,,,, 2 2 J /.....-----:~:__- If' ~::~~::~'~1l______

..:.....;... ":'_ ~.;;.~.;;.=..:-=.= o o 2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20 cm cm . Faired ------e-----, Bluff nn+-n_ Split

Figure 3. Bounda-layer flow charctenstics caculate frm 2-axIs las Doppler velocimetr meaurments. Mea fluctuting flows were meaured over fai, bluff, and split plates at along-str velocities of 5 and 10 cm s-1 meaured at 10 cm abve flume bottom. Vertcal (nonnal tq the pla) advection (w) and tuulence (rms(u'~ere integrte over vertca prfies between 0.2 and 2.0 cm abve the plat surace. Bounda shea stress (u' ) was meaured at 0.1 cm above eah plate. Data symbols not connecte by lies reprent suspt meaurements (see Results).

11 than over the faied plate.

The highest shear strsses occurd on the bluff plates in a region extendig from 1-5 cm downstram of the leadg edge; these were substatialy higher than those over a simar region on the split plates. Reduced shear stress on the split plate, downstream of the splittr bar, was liely due to protection of the plate from incident flows. These patterns were intensified at the higher along-stream flow speed. Both tubulence and shear strss values decreased to a moderate, but constant level in the mid-plate regions of the split and bluf plates, although the values were consistently higher than over the faied plates.

Vertcal advection away from the plate was high at the leadg edge of both the bluff and split plates where flow separtions formed, but deceased sharly (Le., vertcal advection toward the plate increased) as the flow reattched to the plates, 2 to 5 cm downstram The reattachment position on both plates vared with changes in along-stram flow speeds. On the bluff plate at 5 cm s-l, flow reattched 4.5 - 5.0 cm downstream frm the leadg edge, but at 10 cm s-1 the reattachment point was located between 3.5 and 4.5 cm downstream. Generay, the reattachment point on the bluff plates was located between 0.5 and 1.0 cm upstram from the reattachment point on the split plates in comparable along-stream flow speeds. Dye studies indicated that the downstream extent of the separation eddy increased with increasing flow spee up to a critical spee, where the separtion edy became unstable. The observation that reattchment in 10 cm s-1 flow was closer to the leadg edge than in 5 cm s-1 flow suggested that at 10 cm s-1 the separtion edes beame unsteady and appeared smaller because they were being shed more frequently (as in Ruderich and Fernolz, 1986). The reattachment point corresponded with the region of highest shear strss on both the split and bluff plates.

Flow visualzations with dye on bluff and fai plates corroborated the ilV measurements at higher flow speeds, and contrbuted insight into behavior of bounda-layer flows over plates in 2 cm s-1 along-stream flows. Dye injecte along the leadg edfe of a bluff plate followed the contour of the plate in along-stream flows of 2.0 - 2.8 cm s- , outlining a bounda layer qualitatively simiar to that over a faied plate, but with a higher boum.tar-layer growth rate. Visualizations at along-stram flows spees between 5 and 13 cm s-1 indicated that a separtion eddy formed and grw in height and lengt with incrasing flow spee up to 9 cm s- . Separation eddies in faster flow spees were unstable and were shed frequently.

Downstram flow varations were most pronounced in the region 0-8 cm from the leadig edge of all thee plate tyes, where the bounda layer was developing and flow separations occured. Flow disturbances produced at the leadng edge dissipated downstram, and the boundar layer at distances greater than 10 cm from the leadig edge grw so slowly that a change in bounda-layer thickness was not noticeable on regions downstram from ths point. Because of this qualtative difference, downstream varations in parcle contat on the "leading edge" region only were compared to downstreåm varations in vertcal (plate-ward) advection, turbulence and shear stress. In contrast, parcle abundaces on the "mid-plate" region (10-20 cm) were pooled and analyzed relative to along-stream flow speeds and parcle

12 fluxes. The flow in this region was considered "non-varg", although the bounda layer was sti developing slightly.

3.2 Parcle Concentrations in the Flume

Prelimiar measurments indicated that parcles were well-distrbuted thoughout the flume durng experimenta rus, so subsequent parcle concentration measurments were taken at plate height in the section of the flume where the plates were placed. Parcle concentrtions in the flume were consistent between replicate rus and, with the exception of the first run, decreased only slightly (4.7 to 11.1 %) over the coure of the thee flow treatments (speed adjustments) with each run (Fig. 4). Mean parcle concentrations, measured durg each flow tratment (Table 1), were used to adjust corrspondig parcle abundance on the plates, so they could be compar between flow speeds and between replicates. All plate abundaces were normalize to a mean flume parcle concentration of 44.50 ml-1.

- 70 ..I- . Figure 4. Parcle concentrtions S 65 . Run 1 + Run 2 in the flume durg thee replicat Ö . i: . Run 3 runs detennined by countig - 60 . number of spres per ml from i: 0 pump saples drwn from ..... 55 (f Loction B (Fig. 2). Mea "" +. .. 50 numbers collected durg each of vi: t) . the th flow sp trents in i: . 0 45 . + . each replicate ru wer used to u . + +" . . . ~ # nonnalze parcle abundaces on v 40 ...... -t) . . plat (Table 1)...... "" 35 i:(f .

30 , , o 40 80 120 160 200 240 Time (s)

3.3 Parcle Contact Patterns

Parcle accumulation on the plates in the leading edge regions was used as a measure of parcle contact, since previous experients had shown that erosion was negligible. Parcle contact, normalze by parcle concentration in the flume, vared downstram from the leadig edge of all plates, and also vared slightly between plate tyes and flow speed (Fig. 5).

Parcle contact on the faied plate showed a strong pattern relative to the leadg edge at all the flow speeds, with contat increasing with distace downstream in the leadg edge region and becoming relatively constat in the mid-plate region (Fig. 5a). Contact near the

13 80 Faired 2 em s-l 60

40

20

.. 0 C\ 2 4 6 8 10 12 14 16 18 20 18 I; 80 0 Faired 5 em s-l ¡: -- 60 Q) I; ¡: co "t 40 ¡: :: ..~ 20 Q) -I; ...-i 0 i. 2 4 6 8 10 12 14 16 18 20 co 0. 80 Faired 10 em s-1 60

40

20

0 2 4 6 8 10 12 14 16 18 20 Distance From Leading Edge (em)

Figure Sa. Parcle distrbution (nonnalized by flume parcle concentrtion) on faied pla at flume flow spes of 2,5, and 10 cm s-l. Means and stadad deviations are shown frm th replicate flume ru. Values from 0-3 cm represent abundaces in 0.5 cm interval; values from 5-20 cm represent abundace in 1.0 cm interval, centered at the marked distace.

14 80 Bluff 2 em s-1 60

40

20

-- 0 N 2 4 6 8 10 12 14 16 IS 18 20 Q 80 -1 0 Bluff 5 em s i: -- 60 l1 Q i: cd "' 40 i: :: .Q~ 20 -l1 ...Q .¡ 0 s. 2 4 6 8 10 12 14 16 18 20 cd 0. 80 Bluff 10 em s-1 60

40

20

0 2 4 6 8 10 12 14 16 18 20 Distance From Leading Edge (em)

Figre Sb. Parcle ditrbution (nonn by flume parcle concentron) on bluf plates at flume flow spes of 2, 5, and 10 cm s-l. Meas and stadad devions ar shown from th replicate flume rus. Values from 0-3 cm represent abundaces in 0.5 cm interval; values frm 5-20 cm repreent abundace in 1.0 cm interval, centered at the marked distace.

15 Figure Sc. Parcle distrbution (nonn by flume parcle concentrtion) on split plates at flume flow sp of 2,5, and 10 cm s.l. Meas and stadad devitions ar shown from th replicate flume rus. Values from 0-3 cm represent abundaces in 0.5 cm interval; values from 5-20 cm represent abundaces in 1.0 cm inteals, centered at the marked distace.

16 leadig edge of plates in 2 cm s-1 and 5 cm s-l flows was reduced only in the 0-1 cm bands, whereas contact on the plates in 10 cm s-l flows was reduced thoughout the 0-3 cm bands. Because the flow characteristics of turbulence, vertcal advection and shear strss al decreased with distace from the leadg edge at 5 cm s-l and 10 cm s-l, contact was negatively correlated (Kendal's t correlation coefficient, p '" 0.05) with al the flow characteristics (Table 2).

Flume Flow Spe 2 cm s-l 5 cm s-l 10 cm s-l (no. mr1) (no. mr1) (no. mr1)

Replicate #1 38.69 42.65 58.71 Replicate #2 41.3 41.77 46.20 Replicate #3 42.51 45.19 43.61

Table 1. Mea parcle concentrtions the flume for each flow sp trtment durg each of thee replicate rus. Meas were calculat from parcle concentrons meaur in at leat thee pump samples (Fig. 4), except for Replicate #2 at 5 cm s.l (n=2). These meas were use to nonnal pacle contact data to a standad flume concentrtion of parcles (arbitry chosen to be the grd mea of 44.50 parcles mr1).

Pla Tretment Vertca Turulence Shea and Flow Spe Advection Strs Fai 5 em s-l -0.714 .. -0.643 .. -0.714 .. 10 cm s-l -0.857 .. -1.00 .. -0.857 .. Bluf 5 cm s.l 0.00 0.071 -0.214 10 cm s.l 0.143 -0.143 -0.500 .. Split 5 cm s-l -0.500 .. 0.429 0.429 10 em s-l 0.214 -0.286 -0.214 .. P '" 0.05

Table 2. Nonparetrc correlation (Kendal's 't) between positions of parcle contat and flow charteritics on th pla treatments. Values raed frm eight positions downstr frm the leadig edge. Parcle contact ra were assigned to the average of th replicate rus at eah flow intervals.sp. Ra for the flow charctestics were assigned to' mea values frm 4-min saplig

17 Contact on the leadg edge of the bluff plates appeared to peak between 1-2 cm downstram of the leading edge at 5 and 10 cm s-1 flow spees (Fig. 5b). The absence of this peak at 2 cm s-1 flows, where no flow separation was observed, suggested that enhanced contact in the faster flows was a consequence of the flow se~artion. The effect of the separation on contact was much more prominent at 10 cm s. than at 5 cm s-1. Contact patterns on bluff plates in 5 cm s.1 flows were dicult to interpret because varations in the mid-plate region were as great or grater than varations near the leadig edge.

No correlation was found between contact and vertcal advection on the bluff plates at 5 or 10 cm s-1 flow speeds. Although contact was consistently low at the leadng edges, where vertcal advection was diected away from the plate, it was also low at the reattachment point, where vertcal flows were towars the plàte. No consistent corrlation was found between contact and tubulence at 5 and 10 cm s- 1, or between contact and shear strss at 5 cm s.1; the peaks in turbulence and shear strss intensities tended to be locate furer downstream than the peaks in contact. The shear strss maxum at 10 cm s-1 corresponded in location to the contact minimum, resulting in a signcant negative correlation.

Contact patterns over the bluff plates at 2 cm s-1 were simar to those over the faid plates at 2 cm s-1. Since no separtion over the bluff plate was observed at 2 cm s-1, it is likely that bounda-layer development was simar to that over the fai plate. The simlarty in flows may account for the similarty in contact pattrns between the two plate types.

Parcle contact on the split plates was low at the leadg edge in al flows, and increased with distace downstream in the 2 and 5 cm s-1 flow spees (Fig, 5c). In 10 cm s-1 flows, contact peaked under the separation eddy and deceased at the reattchment point. No consistent corrlations between contact and flow characteristics were observed, and the only signicant correlation was with vertcal advection at 5 cm s.l. Ths corrlation reflecte a slight increase in contact with distance from the leadg edge in the 0-3 cm region, where vertcal advection was dereasing.

Downstram patterns of parcle abundance, and their correlations with flow patterns were of priar interest in ths study, but cross-stream patterns were also quantied. These cross-stream analyses were used only to test the assumptions that flow distubances generated at the edges of plates did not influence parcle contact on neighboring plates, and that cross- channel flows in the flume did not influence parcle accumulation pattrns.

Only thee of the plates showed significant varation in cross-stram parcle abundances in the leading edge region (Table 3). The thee plates included a bluff plate in 5 cm s.1 flow positioned near the inner wall of the flume, another bluff plate in 5 cm s.1 flow positioned in the center of the flume, and a split plate in 2 cm s-1 flow positioned near the inner walL. Consistent trnds in parcle contact in the Cross-stream diection did not occur for any plate type at any flow speed. Visual inspections of these graphed data (not shown) also revealed no consistent trends of increased or decrased parcle abundaces along the

18 edges of plates adjacent to other plates. These results suggest that seconda flows produced at a plate leadng edge did not inuence parcle contact on adjacent plates. The graphs also revealed no consistent increase in parcle abundance from the outer to iner sides of a plate, as would be expected if cross-channel flume flows were accumulatig parcles along the inner walls (as in Butman and Grassle, submitted). Six different plates showed signcant cross-stream varation in the mid-plate region but visual inspection of these data agai revealed no consistent trnds.

Plate Position 2 em s-l 5 em s-l 10 em s.l Treatment Led Mid Led Mid Le Mid Faied Run 1 Middle 0.11 1.1 0.46 0.55 0.04 1.55 Run 2 Inner 0.54 0.19 0.63 2.73 * 0.18 0.36 Run 3 Oute 0.64 0.12 0.17 0.41 0.26 227 Bluff Run 1 Inner 1.7 2.78 * 2.72 * 1.28 1.2 2.65 * Run 2 Outer 0.02 1.54 2.21 1.2 0.82 1.28 Run 3 Midde 1.00 0.68 2.54 * 0.91 0.42 0.08 Split Run 1 Oute 0.22 1.0 0.67 1.38 0.94 3.02 * Run 2 Middle 0.69 1.84 2.07 2.82 * 0.97 028 Run 3 Inner 6.06 t 0.05 0.72 2.86 * 0.75 0.05

* P c: 0.05 t P c: 0.01

Table 3. F ratios from ANOV A of cross-str parcle abundaces on the plates in the region 0-8 em downstr of the leag edge (L), and the region 10-20 em downstr (Md). The parcles in six I-em by O.5-em cells were counte from eight O.5-em-deep bands oriente pendicul to the flow in the leading edge region (Ld; df=5, n=8), and the parcles in six I-em by l-cm cell wer counte from five I-em-deep bands furer downstrea (Md; df=5, n=5).

Downstram parcle abundances were analyzed with parametrc statistics (ANOV A) to determne whether varations in parcle contat in the mid-plate region were smal relative to those in the leading edge region. Near the leadng edge, signcant downstram varations (P oe 0.05) were found on 23 of the 27 plates (Table 4), whereas fuer downstream, signifcant varations were found on only eight of the plates. Visual inspection of the pattrns on these eight plates revealed no consistent trends in parcle contact related to distace frm the leadng edge. The eight plates showing significant varation include al plate tyes (th faied, two bluff and th split), flow speeds (two at 2 cm s-l, thee at 5 cm s-l, and th at 10 cm s-l) and positions in the flume (thee near inner wal, the in the midde, and two near the outer wall). These results suggest that downstram varations in parcle contat on the mid-plate regions may have been independent of any of the measured flow charcteristics.

19 Plate Position 2 em s-1 5 em s-1 10 cm s-1 Tretment Led Mid Led Mid Led Mid Faied Run 1 Middle 11.36 t 0.77 13.78 t 3.33 * 28.55 t 3.21 * Run 2 Inner 9.64 t 2.09 11.34 t 1.65 59.99 t 2.01 Run 3 Outer 4.04 t 17.48 t 20.70 t 1.79 23.93 t 2.71 Bluff Run 1 Inner 7.19 t 1.63 1.01 9.20 t 5.67 t 1.45 Run 2 Outer 13.81 t 0.69 0.87 1.13 4.43 t 0.07 Run 3 Middle 2.76 * 0.11 1.7 0.71 14.40 t 10.35 t Split Run 1 Outer 10.62 t 0.58 2.61 * 3.57 * 10.10 t 0.66 Run 2 Middle 2.91 * 1.98 2.27 * 0.69 9.24 t 2.34 Run 3 Inner 1.43 5.00 t 2.64 * 0.82 8.70 t 5.00 t

* P 0: 0.05 t P 0: 0.01

Table 4. F ratios from ANOV A of downstr parcle abundace on the plates in the region 0-8 cm downstr of the leag edge (L), and the region 10-20 em downstr (Md). The parcles in eight 0.5-em by I-em cells were counte from six I-em-wide strps rug parel to the flow in the leadig edge region (L; df=7, n=6), and parcles in five I-em by I-em cell wer counte in the si I-em-wide strps furer downstr (Md; df=4, n=6).

3.4 Paricle Contact Rates

Parcle abundace was measurd away from the effects of the plate leadig edges (10 - 20 cm downstream) to calculate contact rates in non-varg bounda-layer flows. The absence of flow-related vanations in parcle contat in this mid-plate region (Table 4) indicated that these parcle contact data could be pooled for subsequent analyses. Mean contact rates were then compar between plate tyes and between along-stream flow spees.

Parcle contact rates on the mid-plate region vared strngly with along-stram parcle flux (Fig. 6). Contact rates did not, however, var strongly with along-stream flow spee (Fig. 7). A two-way ANDY A revealed no signifcant (P '" 0.05) vanation with either flow speed or plate tye in the "non-varg" flow region of the plates. Thus, the observed increase in parcle contact rate on plates in increasing along-stream parcle fluxes appears to have been due almost entily to increases in parcle concentrtion in the flume.

20 ..- 0.04 1~i Figure 6. Parcle contat rate on N IS . mid-plate region as a fuction of () . & 0.03 . along-str flux (caculte frm Ô . the product of parcle --i: . concentron in the flume and ..II . II . along-str flow sp). Contact i: 0.02 '\ .. . , rate is exprse as number of () . ..II ~. . parcles (not nonned) i: . 0 accumulting on five l-cm bands u 0.01 . in the region 10-20 cm -II . Faired () :: . Bluff downstr frm the leang edge .. Split II . durg a 30-min interval. ci 0 o 200 400 600 Along-Stream Particle Flwi (no. ciñ2 iI1)

.. 0.05 1~I D Faired N IS o Bluff 0.04 () IE Sp Ii t Ô i: Figure 7. Parcle contat rates -- ..II 0.03 (normal) on the experienta i:II plate. Values plott ar mea .. () II 0.02 ...... :.:.:.:.:.:.:-:.:. (and stadad devition) of contat ...... i: .....-:.:.:.:.:.:.:.:.:...... rates for th replicate rus. 0 ...... :.:.:.:.:.:.:.:.:...... u ...... :.:.:.:.:.:.:.:.:...... :.:.:.:.:.:.:.:...... II 0.01 ...... 'õ ...... ::::::::::::::::::...... :: ...... ;.:.:-:.:.:.:.:. li!1 ...... II ...... ci ...... 0 ~~~\~~ If I;, 2 5 10 Flow speed (cm s-l)

4 Discussion

4.1 Parcle Contact Patterns in a Developing Bounda Layer

The lack of consistent corrspondence between parcle contat pattrns on the leadig edge region of plates and individual flow characteristics (vertcal (plate-ward) advection, turbulence, shear stress) suggests that several aspects of the flow ar interatig to produce

21 the observed parcle patterns. We had predicted that increasing plate-ward adveCtion and turbulence in the boundar layer would increase parcle contat by incrasing the flux of parcles though the bounda layer (as in Eckman, 1990). A signcantly positive relationship between parcle contact and plate-ward advection or tubulence was not observed, however, on any of the plates. One explanation for this unanticipated result is that elevated values of plate-war advection and turbulence are often associated with enhanced shear strss very near the plate (Fig. 3), which may inhibit the parcles frm settg (Eckman, 1990). It is importnt to note that effects of shear stress ar actig on parcle motions in the water column, and ar not causing parcle erosion. No evidence of post- contact parcle motion or erosion (e.g., pits or trails in the adesive) were observed on the plates.

A closer examiation of parcle patterns within plate tyes reveals that high parcle contact rates occur only in regions with both a relatively thck bounda layer and reuced shear strss. Regions with only a thck bounda layer (e.g., the reattchment points on the bluff and split plates) or reduced shear stress (within 0.5 cm of the leadg edge of bluf and split plates) did not collect parcles, hence the non-signcant corrlations with parcle contact on these plates (Table 2). We suggest that mig in the thcker bounda layer downstram from the plate leadg edges enhances the flux of parcles from the water column into the bounda layer, but once in the bounda layer, parcle contact is enhanced only in regions with low shear stress. Plate-ward advection may also increase the flux of parcles into the boundar layer, although parcle contact does not appear to be enhanced at reattachment points due to elevated shear strsses.

4.2 Implications for Laral Settement in a Developing Bounda Layer

Results of these parcle contact experiments can be applied to models of laral settlement, but only if a hydrodynamc simlarty between parcles and sing larae exists. Near bottom shear stress will only inhibit parcle (or laral) contact if the ratio of parcle settling velocity to shear velocity (a correlate of shear stress) is low (Ekm, 1990). In the present study, the parcles used had a low specific gravity and low settg velocities. Assumptions were made that (1) parcles followed streames, (2) parcle-parcle interactions were negligible, and (3) parcle settg velocities were smal relative to tubulent intensities and shear velocities. No evidence of parcle clumping was observed, and the results suggested that gravitational and inertial forces were less importt than bounda-layer motions, except very near the plate. These same assumptions are probably vald for sma larae, such as many of those in the deep sea, but ar not likely to be vald for large larae such as baracle cyprids.

Evidence that swimng speeds, inerta or settlng velocity of a lara may be importt in boundar-layer flows can be found in settement data for baracles in controlled flume flows. In a series of experiments quantifying cyprid contact, exploration and settement on plates in flows identical to those used in this parcle study, intial cyprid contact was found to be highest at the reattachment points on the bluf and split plates (Muleaux and Butman, 1991). This result suggests that cyprid settling velocities or swig spes are

22 suffciently large to overcome the tubulence and shear stress at the flow reatthment points.

Small parcles with low settlg velocities ar expected to contact vertcal plates in patterns simiar to those observed on horionta plates because the bounda-layer prssur and turbulence parameters are equivalent. Ths assumption was not tested in the flume experiments presented here because the vertcal plates were not easily elevated above the strngly sheard flows near the bottom of the flume. High settg velocities of large larae, however, may cause them to contact upper suraces of horizonta plates more frequently and in different patterns than they contact vertcal plates. Thus, it may be importt to test contact patterns of large parcles on both horizonta and vertcal suraces.

An additional caveat for the use of parcle contact pattrns in laal settement studies is that they are intended to predct intial contact patterns, which are not always maitaed in subsequent settlement patterns. Especialy in shalow-water habitats, where larae may be relatively large and have diverse behaviors, post-contact laral behaviors may substatialy influence settlement patterns. Although some of these behaviors ar flow-dependent (e.g., Crisp 1955; Mullneaux and Butman, 1991), they ar not well documented for most larae. Thus it seems reasonable to caution that behavior (ie. swig, phototas, chemotas, geotaxis, rheotas, etc.) and hydrynamc charcter of the lara (relate to size, density and shape) may signifcantly influence laral settement in the field.

4.3 Parcle Contact Rates in the Non-Varng Flow Region

Parcle fluxes to the mid-plate region (downstram of the developing bounda layer) vared only slightly (and not signcantly) with along-stram flow speed. Our intial expectation was that parcle flux would increase with incrasing flow spee~ as a result of increased turbulent mixing of parcles though the bounda layer towar the surace. The absence of an increase in parcle contat with incrasing flow speed coroborates our observations, from developing bounda layers, that enhanced tuulence does not necessary increase parcle contact. It is likely that, in both the developing bounda layer and the non- varing flow region, parcle contact depends on the relative values of parcle settg velocity, integrted tubulence though the bounda later and bounda shear strss. Over the range of flows used in this experient (2-10 cm s- ), these factors appear to balance each other in such a way as to maitan a relatively constat contact rate, independent of along- stram flow speed. Rather than depending on flow spee, parcle contact appears to depend most strongly on parcle abundace in the water column.

The empircal results from ths study can be used to predict the flux of parcles to a plate, Fp' from parcle concentrtion in the water /P wi in the range of flow spees teste:

(2) F p = F w . R(u) ; where Fp = Pf./t (the parcle accumulation over tie in' cm-2 s-I), u is the along-stram flow speed (cm s-l , F w = /Plw . u (the along-stream flux of parcles in the water in cm-2 s-l) and R(u) is the empircaly measured ratio of F IF w at each flow spee Since values of

23 paricle contact did not var with along-stram flow speeds in the range of flows used in these experiments, this equation can be simplifed to:

(3) F=lpl.K'p wi - , where K (in cm s-l) is the empircaly determed value of parcle contact rate divided by parcle concentration in the flume (normalize values plotted in Fig. 7). The mean value of K from thee replicates of each plate tye in thee flow spees was 4.79 . 10-4 cm s-1 (st. dev. = 1.23 . 10-4 cm s-l). The varation in these measurments was relatively high between replicate plate types with each flow speed, but was relatively low between plate tyes and flow speeds. This result suggests that estimates of parcle concentrtion in the water colum should not be calculated from only a few replicate plates. In addition, estiates from the simplified equation (3) are vald only in flows simiar to those used in the flume (2 - 10 cm s-I), and for regions of the plate in non-varng bounda layer flow.

4.4 Interpreting Laral Settement Plates

Predcting parcle concentrtions in the water frm parcle contat rates on flat plates may be usefu in interpretig laral abundances on settement plates. Quantication of laral abundances in the open ocean can be very dicult, especialy in deep-sea habitats where traditional ship-board samplig methods (net tows) are dificult to use, and larae are relatively sparse. Measurg laral abundaces by quantig their settement onto arcial suraces suspended in the water column is an attactive alternative to net tows because laral abundances can be integrated over a period of time, and only benthc larae (and not large volumes of holoplanon and flocculent material) ar collected. The litation of ths technique is that the relationship between laral settement rates and laral abundaces is influenced by often-unmeasured factors including laral contact rate with the plate and active laral behaviors. Contact rates of sma larae with the plate ar expecte to be controlled by the same proesses controllng contact rates of small parcles, but laral settement rates wi differ because not al larae wil choose to sette on contact. This behavior can be parameterize as a "choice" term, C(larva)' representing the probabilty of laral settement on contact. Ths choice term can be integrated into the parcle contact equation to formulate a laral settement equation:

(4) Sp = Fw · R(u) . C(larva) where Sp is the laral settement rate onto the plate (cm-2 s-l), and Fw is the along-stram flux of larae past the plate (cm-2 s-\

In relatively slow flows, similar to those in many deep-water habitats, the laral choice term can be used with the simplied version of the parcle contact equation (3), resultig in a simplified laral settement equation:

(5) Sp = ILwl . K · C(larva) ;

24 where IV is the concentration of larae in the water column (cm-3).

In deep-water habitats, laral behavioral responses at settement are probably speies- specific, and may depend on a combination of surace cues and envinmenta factors, includig flow speed. Even in flow regies where laral contact rates may not be strongly flow-dependent, it is possible that laral behavioral responses are. Most of these active laral responses ar unkown, but they are not likely to var between settement plates deployed in deep-water habitats charcteried by weak grents in light, temperatu, and curent spees. Thus, relative differences in laral settement between plates in these tyes of habitats can be used to estimate relative diferences in laral abundaces. Since laal settement responses may be flow dependent, however, unexpected varabilty may be introduced into settement onto plates in different flows, even in the range of flow spees where laral contat is independent of speed.

Interpretation of laral settement rates in terms of laral abundaces should be done cautiously, but if the settement plates ar oriented parallel to the flow, and the flow rates ar measured, then relative dierences in laral settlement on the plates should reflect relative differences in laral abundaces in the water column.

5 Acknowledgements

Much of the approach, experienta design, and analysis for ths project came from many discussions with Cheryl An Butman, John Trowbridge, and Rocky Geyer. Fundig for the study was provided by the Office of Naval Research under Grant Number NOOI4-89- 1341 to L.S.M and Grant Number NOOI4-89-J-I112 to C.A. Butm and L.S.Moo The Coastal Research Center at WHOI provided funds for the flume research.

6 Literature Cited

Agrawal, Y.C. and C.H. Beltig. 1988. Laser velocimetr for benthc sedent transport. Deep-Sea Res. 35: 1047-1067.

Butman, C.A. 1986. Laral settement of soft-sedient invertebrates: some predctions based on analysis of near-bottom velocity profùes. In: J.C.I Nihoul (ed.). Marine Interfaces Ecohydrodynamics. Elsevier Oceanogr. Ser., VoL. 42, Elsevier. Amsterda, pp 487 - 513.

Butman, C.A. 1987. Laral settement of soft-sedient invertebrates: the spatial scales of pattern explained by active habitat selection and the emergig role of hydrynamcal processes. Oceanogr. Mar. BioI. Annu. Rev. 25: 113-165.

Butman, C.A., and R.I. Chapma. 1989. The 17-Meter Flume at the Coasta Researh Laboratory. Pan I: Descrption and user's manual. WHOI Tech. Rep. 89-10. 31 p.

25 Butman, C.A and J.P. Grassle. Submitted. Active habitat selection by Capitella sp. I larae: i. Simple two-choice experients in sti water and flume flows. (Submitt to J. Mar. Res. 1992).

Crisp, D.J. 1955. The behavior of baracle cyprids in relation to water movement over a surace. J. Exp. BioI. 32:569-590.

Duggins, D.O., J.E. Eckman, and A.T. Sewell. 1990. Ecology of understory kelp environments. n. Effects of kelps on recritment of benthc invertbrates. J. Exp. Ma. BioI. EcoI. 143: 27-45.

Eckman, J.E. 1990. A model of passive settement by planonic larae onto bottoms of differing roughness. LimoI. Oceanogr. 35(4): 887-901.

Hannan, C.A 1984. Initial settement of mare invertebrate larae: The role of passive sinkg in a near-bottom tubulent flow envinment Ph.D. thesis, Woo Hole Oceanogr. Inst./Mass. Inst. TechnoI. Joint Progr. 534 p.

Keen, S.L. 1987. Recruitment of Aurelia aurita (Cndaa: Scyphozoa) larae,is position-

dependent, and independent of con specifc density, within a settg surace. Mar. EcoI. Prog. Ser. 83: 151-160.

Kiya, M. and K. Sasak. 1983. Strcture of a turbulent separation bubble. J. Fluid Mech., 137, 83-113.

Middleton, G.V., and J.B. Southar 1984. Mechanics of sedent movement. Society Economic Paleontologists and Mineraogists, VoL. 2. Tulsa, Oklahoma.

Mullneaux, L. S. 1992. Implications of mesoscale flows for dispersal and retention of larae in deep-sea habitats. In: K. J. Eckelbarger and C. M. Young, eds. "Reproduction, Laral Biology and Recritment in the Deep-Sea Benthos". Columbia University Press (in review).

Mullneaux, L.S. and C.A Butmn, 1990. Recruitment of benthic invertebrates in bounda- layer flows: a deep water experiment on Cross Seamount Limo1. Oceanogr. 32: 409-423.

Mullneaux, L.S. and C.A. Butmn. 1991. Initial contact, exploration and attchment of baracle (Balanus amphitrte) cyprids setting in flow. Ma. BioI. 110: 93-103.

Nowell, A.R.M. 1983. The benthic bounda layer and sedment transport. Rev. Geophys. Space Phys. 21: 1181-1192.

Nowell, AR.M. and P.A Jumars. 1987. Flumes: Theoretical and experienta considerations for simulation of benthic envionments. Oceanogr. Mar. BioI. An. Rev. 25: 91-112.

26 Paola, C. 1983. Flow and skin frction over naturl rough beds. D.Sc. thesis. wOO Hole Oceanogr. Inst./ Mass. Inst. Techno!. Joint Progr. 347 p.

Ruderich, R. and H.H Fernolz. 1986. An experimenta investigation of a tuulent shear flow with separation, reverse flow, and reattchment. J. Fluid Mech., 163: 283-322.

Schlichtig, H. 1979. Bounda-layer theory, 7th ed. McGraw-Hi, New York.

Siegel, S. 1956. Nonparmetrc statistics for the behavioral sciences. McGrw-Hi, New York.

27

DOCUMNT LmRAY March 11, 1991 Ditrution Listfor Technical Report Exchange Attn: Stella Sanchez-Wade Pell Marine Science Library Documents Section University of Rhode Island Scripps Institution of Oceanography Narragansett Bay Campus Library, Mail Code C-075C Narragansett, RI 02882 La Jolla, CA 92093 Working Collection Hancock Library of Biology & Texas A&M University Oceanography Dept. of Oceanography , Alan Hancock Laboratory College Station, TX 77843 University of Southern California University Park Librar Los Angeles, CA 90089-0371 Virgina Institute of Marie Science Gloucester Point, VA 23062 Gift & Exchanges Library Fisheries-Oceanography Library Bedford Institute of Oceanography 151 Oceanography Teachig Bldg. P.O. Box 1006 University of Washigton Dartmouth, NS, B2Y 4A2, CANADA Seattle, WA 98195 Office of the International Library Ice Patrol R.S.M.A.S. c/o Coast Guard R & D Center University of Miami Avery Point 4600 Rickenbacker Causeway Groton, cr 06340 Miami, FL 33149 ' NOAA/EDIS Miami Library Center Maury Oceanographic Library 4301 Rickenbacker Causeway Naval Oceanographic Office Miami, FL 33149 Stennis Space Center. NSTL, MS 39522-5001 Library Skidaway Institute of Oceanography Marine Sciences Collecton P.O. Box 13687 Mayaguez Campus Library Savannah, GA 31416 University of Puerto Rico Mayaguez, Puerto Rico 00708 Institute of Geophysics University of Hawaii Library Library Room 252 Institute of Oceanographic Sciences 2525 Correa Road Deacon Laboratory Honolulu, HI 96822 Wormley, Godalmig Surrey GU8 SUB Marie Resources Information Center UNITED KINGDOM Building E38-320 MIT The Librarian Cambridge, MA 02139 CSIRO Marine Laboratories G.P.O. Box 1538 Library Hobart, Tasmania Lamont-Dohert Geological AUSTRALIA 7001 Observatory Columbia University Library Palisades, NY 10964 Proudman Oceanographic Laboratory Library Bidston Observatory Serials Department Birkenhead Oregon State University Merseyside L43 7 RA Corvalls, OR 97331 UNITED KINGDOM

Mac 90-32

5072-101 REPORT DOCUMENTATIONPAGE WHOI-92-26 11. REPORT NO. 2. 3. Recipint's Ac..io No. 4. Title and Subtnle 5. Repo Date June 1992 Parcle Contat on Flat Plate in Flow: A Model for Initial Laral Contact 6.

7. Aulhor(s) I. Peronnlng Orgnizatio Rept. No. Elizabeth D. Galad an Lauren S. Mulinaux WHOI 92-26 9. Perlonlng Organization Name and Addre.. 10. PrctaskIor Unit No.

The Woo Hole Ocogrphic Institution 11. Corat(C) or Grent(G) No. Woo Hole, Massahusetts 02543 (C) NOO14-89-J-1l12 (G) NOO14-89-J-1431

12. Spollorlng Organization Name and Addre.. 13. Type of Repo & Period Covered

Offce of Naval Resech Tecnica Report 14.

15. SUpplmentary Not.. This report should be cited as: Woo Hole Ocanog. Int Tech. Rept., WHOI-92-26.

16. Absract (Limit: 20 wors)

Pattern and rate of parcle contat onto flat pla in steady uniditiona flows were investigated in a laboratory flum. Plate with thee leading edge confgurtions (faied, bluff and split) were used to generate bounda-layer flows that difered in downstram patterns of plat-ward advection, tubulence and shear strss. Parcle contact onto the leadig edges of al plates was consistently low in 2,5, and 10 em s.1 along-stram flow spe. Contat was enhced under separtion eddies that fomied over bluff and split plates, but was reduce at retthment points. High contact rates appe to correspond to a combination of loc plate-ward advection, a thick bounda layer, and reduce shea strs.

Surriingly, parcle contact rates in th "non-varing" flow region furer downtream on th plate vared only slightly between plate ty and between flow spe. Contact rates did, however, var strongly with pacle abundace in the flume. These results were used to develop a predictive model of passive laal contact rate onto settlement plate in known laal concetrtions and free-strea flows. The contat model, when combined with laal behaviora observations, provides the basis for a more objectve, quantitative method of intereting laal settement pla.

17. Doument Analysis a. Derlptor

1. bounda-layer-flow 2. laral settlement 3. parcle contact- b. Idntlnerepen-£nd Tanns

c. COSATI FlelclGroup

11. Availability Statement 19. Seurit CI.. (ThIs Repo) 21. No. of Pag.. Approved for publication; disbibution unlite. UNCLASSIFD 31 20. SeurKy CIa.. (This Page) 22 Price

(Se ANSI-Z39.11) S.a/natructlone on RavanH OPTIONAL FORM 27 (4-77 (Forrl NTIS-35) Dertment 01 Commerc