.5:: C 1 T \| . .- .. 3.. : ..l:....zl...iixai Elvw \w'.‘l \3 CL 7M. ‘2 Eu ,4 .h; ‘»--f D7!“ I ........l... 7.11.! . ¢ 5.: I V phi/lit”. ll“: $81 .o $3.46. It! t... F i‘ri.‘ 5: r ‘ttcr(.v(:‘ . .wul :,r.i!\?1 .rf.Ilr(tI: It}: IGrr :‘tvtkt; 5:: (71‘.- .I (gtttnfiwrn'r' , ‘V .r..!v..« , 1 tn“! . ,7,“ fi?‘1'}¢““‘ 1‘ 14:4- 0. r . I. .. Vltl" . 1.5!?) \tlfr e9721!:t , )7. . I .\ . i.\ . +> 20— 'H L m 0 l- ‘6 .- 0 l 1 4. .......1 . .......l . .......1 0.01 0.1 l 10 Mean Groin Diameter (mm) Figure 2. Mean porosity (i sd) in relation to mean grain diameter for laboratory substrates. Squares correspond to ungraded Wentworth substrates while diamonds correspond to graded substrate mixes. All porosity values were determined from frozen cores. 27 Within ungraded and graded laboratory substrates, permeability was linearly related to group mean grain diameters (figure 3). Permeability ranged between 0.02 and 200 cm min‘1 for ungraded substrates and 0.03 to 1.5 cm min‘1 for graded substrates. For a given mean grain diameter, permeability was higher for ungraded than graded substrates. 'Sediment Selection Experiments with Tests for Interference Competition Effects All substrate selection studies were conducted in twelve 100 l aquaria set in a refrigerated water bath. In each aquarium, two four-place test arenas measuring 22.2 cm by 22.2 cm by 35.6 cm (LWH) were placed at opposite ends. Tested in groups of four, different substrates in square 0.98 liter polyethylene containers were distributed among groups of four test aquaria in a preset pattern such that each test arena received a full complement of test substrates and each substrate was distributed equally among all positions in the test arenas (figure 4). Because the test arenas only permitted testing of four substrates at a time, preliminary analysis of substrate selection at both ends of the Wentworth scaled sediment distribution were conducted to identify and bracket the preferred substrate classes. After selection experiments on ' l I III'II' I I III-Ill l I UIIU‘IV' 1000 m0- a K (cm/min.) 0.01 I A A Inn-All 1 I ljljlll A A Ill-ll. 0.01 0.1 1 10 Mean Brain Diameter (mm) Figure 3. Mean permeability with respect to mean grain diameter for laboratory substrates. Squares correspond to ungraded Wentworth substrates while diamonds correspond to graded substrate mixes. All permeability values were determined from frozen cores. 29 A a A o c a a A c o p c A o o c o c 0 A a o lo a a A c a o A A Bi '7 7: I 2 .. J 5 '1 l“ }‘ Figure 4. Schematic representation of substrate distribution among test arenas and oblique view of a test arena. Letters A through D correspond to different substrate types. Each substrate tested in a given series occupies each position in a test arena twice. An air stone is positioned between each test arena in an aquarium and circulates water through the side meshes of the test arenas. 3O Wentworth scaled substrates were completed, selection experiments on graded substrates (mixes A - D) were conducted. Separate substrate selection tests were performed on small (40 to 79 mm TL) and large (120+ mm TL) ammocetes of both species. To evaluate the effect of intraspecific competition within a size class, series <3f unstaggered and staggered releases of ammocetes at different densities were conducted. The results of unstaggered releases of ammocetes at a low density of 4 per test arena and a high density of 8 per test arena were contrasted with the staggered releases of two and three groups of four per test arena. The short-term importance of intraspecific competition for space was assessed by comparison of substrate selection patterns between different release/density treatments. To evaluate interspecific competition effects on substrate selection within a given size class, the results of simultaneous releases of 4 L. appendix with 4 P. marinus were contrasted with the substrate preferences exhibited individually by both species at densities of 4 and 8 per test arena. Experimental blocks are listed in table 4 by species, size class, density, and release pattern. Table 4. 31 Blocking structure for experimental analysis of larval lamprey substrate habitat selection and evaluation of intra- and interspecific competition effects. A. Small L. appendix Large L. appendix Unstaggered Staggered Unstaggered Staggeredi Density Release Release Release Release 4 X X 8 X - X '- 12 - - - x B. Small 2. marinas Large 2. marinus Unstaggered Staggered Unstaggered Staggered Density Release Release Release Release 4 X X 8 X X X X 12 - - - x C. Large L. appendix Large 2. marinus Density Unstaggered Release Staggered Release 4:4 X - 8:8 X - 32 For a given experiment, ammocetes were released in the dark and allowed to burrow. All tests ranged in duration from 4 to 6 days and staggered groups were released at 48 hour intervals. Each experimental run was conducted at a constant temperature of 10 i 0.5 °C and the maintenance feeding schedule was sustained throughout the course of all experiments. At the end <3f a given experiment series, each test arena was dismantled and the number of ammocetes in each substrate container was recorded by sediment type and container position. Burrow depth for individual ammocetes, defined as the difference between the midpoint of the branchial basket and substrate surface, were recorded in a subset of test arenas. Food Related Habitat Selection Experiments A different set of habitat selection experiments were conducted to test the effect of food particle distribution on burrowing behavior. These tests were broken down into two categories to analyze the effect of food particle distribution in sediments versus that of food particle distribution in the water column. For the sediment/food particle distribution experiments, a constant sediment size of 0.375 mm mean 33 grain diameter (range: 0.25 to 0.50 mm) was used in all four positions of the test arenas. Two of the four positions in each test arena were injected with 20 ml of a 100 g/l yeast perfusion. The pattern of perfused versus unperfused sediments across all replicates insured equal distribution of treatments over all positions in the test arenas. Ammocetes of both specd_es were released individually in unstaggered groups of 8 and allowed to come to distributional equilibrium in accordance with the test arena protocol detailed above. To test the effect of food particle distribution in the water column as opposed to sediments, a different test apparatus was used. Shown in figure 5, the four individual two-place test chambers permitted the ammocetes a choice of burrowing where the water column contained an abundance of food particles versus an area of zero food particle concentration. Solutions of distilled and yeast perfused water were pumped into opposing sides of the test unit at a rate of 7.8 ml per minute, draining towards a common center drain. Additional tests were run with an algae perfusion (Ankistrodesmus falcatus.), a green alga common in the gut of larval lamprey (Hardisty and Potter, 1977). Concentrations of yeast in the food chambers at the 34 : - . .0'4 . 0.. I ' I ll 0‘ . 7. s: . "b ' \ a. .0 .' '.‘ ’00 ‘- ° 0 I o’ c.O. .‘.. . “' O ' o. . 0 ' e .l e ..I . . p '0' 3! 0°: 3 .0- 0. .. on .'.I a 0 go ‘ o. ... . .:o n . . ’00 I . .' fl . 0' o O I awn '- -~' ~sx ,,.,...... balsam-I Q: g . 9 . . .. 0. ..e. .5 . . I! ... 0 o O . .. o . . all. . '.- . : . . .. : c . 0' I ‘. n t e I . . O ‘ .. I e .0 o. . i. .. . C a e O o O o .0 I o ’. o ‘ g. . O. 0 b ..... U 0. ‘ ,o e a. ‘ ~." _ 0 0"», g a 0 a... . n , l \ . ' .0. ..0 ‘ 0‘ - '0 u o ‘ 0 O c . 0‘ .0 . ......0 e LAA .' . 0". ~.Io . o ‘ J- c 0" . C :SEFTIUNC P\ .— / (WSNLLED umTEP s rocx _ SOLUTION (WSWMLED “MESH Figure 5. Schematic representation of food/habitat A food particle stock solution is pumped through separate lines to the inner ends of four separate test chambers while distilled water is pumped through another set of lines to the selection test apparatus. outer end of each test chamber. for each line and controlled with a In use, opposing flows from each end to 7.8 ml min."1 Peristaltic pump. The fl ow rate was set of a test chamber run towards a center drain. 35 start of the experiments were z 0.06 g/l. for yeast and 35 to 43 ug chl a./l for Ankistrodesmus. Ammocetes were lightly sedated (MS-222: 3- aminobenzoic acid ethyl ether) and released in the dark at the center of each test chamber and allowed to make their choice upon recovery from the anesthetic. Tests were run overnight and counts of ammocetes burrowed in both sides of the test chamber were recorded. The experimental blocking structure for all food/habitat selection experiments is listed in table 5 by species, category, and food type. Statistical Analyses Frequency data from each experimental series was initially tested for departure from random substrate selection using a replicated G-test with William's correction (Sokal and Rohlf, 1981). This approach permitted separate testing of substrate selection and homogeneity of replicates. When the total number of ammocetes tested in an experimental series was less than 33, results of the G-test were compared with the likelihood ratio test, an exact probability test. 36 Table 5. Blocking structure for experimental analysis of food/habitat selection. All releases were unstaggered and sediment type remained constant across all experiments (mean grain diameter 0.375 mm, range 0.250 - 0.500 mm). L. appendix 2. marinus Food Substrate Water Substrate Water Yeast X X X X Perfusion Algae - X - - Perfusion 37 Further analyses of substrate preferences were conducted with multifactor model I ANOVA's to test for effects of position in the test arenas, differences in substrate preferences between species and size classes, and effects of intra- and interspecific interference competition. Unplanned multiple comparisons among means for a given analysis were conducted with the T-method when variances were homogenous (Sokal and Rohlf, 1981) and the Games and Howell method when variances were heterogenous (Games and Howell, 1976)(see Appendix 4). 1987/88 Field Methods An underwater transect was used as a reference frame for fine scaled field measures of substrate grain size distribution and associated properties, food particle distribution, and temperature regime off the mouth of the Chippewa River. Using SCUBA, I lined out an initial transect extending perpendicularly from the upper edge of the alluvial fan along the bottom with 90 m of nylon cord on 21 May 1987. The transect endpoints were anchored with concrete blocks and identified at the surface with buoys. Between the endpoints, the transect line was anchored with PVC stakes every 10 m. 38 Thirty reference stations were defined along the transect at 3 m intervals with indelible ink marks on the line and twist-tied aluminum labels. The transect was further extended on 25 July 1987, with the addition of three buoyed stations at 120, 200, and 320 m from the edge of the alluvial fan along the same heading. At each station, depth was measured with an electronic depth sensor (ORCA Industries) with 25 cm accuracy and the bed slope was gauged within i 1.0° with an inclinometer. This information was used to fix the horizontal distance of each station from the edge of the alluvial fan using right triangle geometry in tandem with the known station positions along the transect line. To track the formation and downslope movement of the thermocline over the course of the summer, three Ryan J-180 bathythermographs were deployed along the transect at stations 10, 16, and 24 at depths/distances of 10/25.3, 12.2/43.1, 13.7/67.0 m respectively. The station 10 bathythermograph was anchored on the bottom with a concrete block on 24 May 1987 while the others were similarly placed on 13 June 1987. The thermal Sensors were set within 25 cm of the sediment/water 39 interface. Resultant strip charts were digitized and daily median temperature and range were calculated for each station To characterize the sediment particle size distribution along the transect, 0.98 liter sediment grabs at 1.3 m depth increments were made by hand on 24 May using SCUBA. Sand, silt, and clay fractions were separated using standard hydrometric methods (ASTM Designation D422—63) at the M.S.U. Soils Testing Laboratory. In addition, to characterize substrate porosity, permeability, and substrate food particle distribution over space and time, three 2.54 cm diameter cores exceeding 8 cm in length were taken by hand according to the schedule outlined in table 6 using SCUBA. Core samples were slowly frozen in the core tube to limit disruption of the sediment fabric (Rutledge and Fleeger, 1988). Two of the replicate cores for a given station and date were used in the determination of porosity and permeability while the remaining core was saved to analyze sediment food particle distribution and abundance. Kept in the dark, the frozen cores were processed in the laboratory within six months of collection. Lenz and Fritsche (1980) verified 40 Table 6. Summer 1987 sediment core sampling schedule. Stations 31—33 not set up until mid July. Missing values on 21 August due to unsafe diving conditions. Station Depth Distance Date (m) (m) 7/03 7/26 8/21 9/22 1 1.8 0.0 ' ' ' ' 2 3.4 2.6 ' ' ' ' 3 4.9 5.2 ' ' ' ' 4 6.4 7.8 ' ' ' ' 5 7.3 10 6 ' ' ' ' 8 8.8 19.4 ' ' ' ' 10 10.1 25.3 ' ' ' ' 12 11.0 31.2 ' ' ' ° 16 12.2 43.1 ' ' ' ' 20 13.1 55.1 ' ' ' ' 24 13.7 67.0 ' ' ' ' 29 14.9 81.9 ' ' ‘ ' 31 17.0 130 ' ' 32 19.0 200 ' ' 33 22.8 320 ' ' 41 negligible degradation chlorophyll in frozen samples stored up to six months at -20° C. Food particle distribution and abundance were monitored in the substrate and at the sediment/water interface throughout the summer. The temporal distribution and abundance of organic material and chlorophyll a in the sediments were estimated from replicate cores collected as described above. Using a vacuum sampling system (figure 6), two 1140 ml water samples were collected from the water column within 1.0 cm of the sediment/water interface at a given station. The sampling schedule by date and location is listed in table 7. A detailed discussion of field and laboratory processing of substrate and water column samples is included in appendix 3. The distribution and abundance of lentic ammocetes along the transect was assessed using a trap tray method modified from Thomas (1961). A total of 30 trap trays were built from plastic children's sandboxes (75 x 75 x 18 cm LWH). Each trap tray was fitted with a continuous aluminum flange riveted to the inner edge of the box and protruding 5 cm above the lip. A 1.0 cm hole was drilled into one corner of the tray for use as a TFM injection port. Sampling an area of 0.5625 In2 each, 42 TAIR our «\ENMPLE FLOW Figure 6. Schematic representation of diver operated water sampler. In use, the plexiglass sampling plate was slowly set over the sampling point without disrupting the substrate. A 1140 ml sample from the layer of water within 1.0 cm of the sediment/water interface (SWI) was taken by slowly opening the stopcock (V1) at the plexiglass sampling plate and allowing the partial vacuum created in the submerged flask to suck in part of the water sample. After equalization of pressure in the vacuum flask, the rest of the sample was taken by slowly opening of the flask stopcock (V2) and bleeding off residual air. The volume of water covered by the plexiglass plate was 1.5 times as great as that held in the flask to insure that the sample came from the SWI as opposed to shallower strata. 43 Table 7. Summer 1987 water sampling schedule. Stations 31—33 not set up until mid-July. Missing values for 21 August due to unsafe diving conditions. Station Depth Distance Date (m) (10) 7/05 7/27-28 8/21 9/18-19 1 1.8 0.0 ' ' ' ' 2 3.4 2.6 ' ' ' ' 4 6.4 7.8 ' ' ' ' 9 9.8 22.3 ' ' ' ' 17 12.2 46.1 ' ' ' ' 29 14.9 81.9 ' ' ' ' 31 17.0 130 ' ' 32 19.0 200 ' ' 33 22.8 320 ' ' 44 trap trays were used in groups of three. Trapping effort was extended over space and time according to the schedule outlined in table 8. Trap trays were placed on the bottom with the aluminum flange imbedded in the sediment acting both as an anchor and an impediment to lateral burrowing of ammocetes. A concentrated solution of TFM was then introduced through the injection port with a 25 ml syringe. The amount of TFM injected was scaled to provide a final concentration of 9.5 ppm active ingredient within the volume enclosed by the trap tray. Ammocete movement patterns with respect to changes in abiotic and biotic factors over space and time were assessed with a mark/recapture study. A total of 1046 larval L. appendix were collected on 20 May from East Davignon Cr., Sault Ste. Marie, Ontario. Separated into two groups, 522 received a ventral subcutaneous mark while the remaining 524 received a dorsal mark. The marks were different from marks used in previous years. The water temperature was 10 °C when the ammocetes were collected. During marking and prior to lentic release, ammocetes were held in 7 °C water. On 23 May, the ventral marked ammocetes were released at station 10 (10 m depth, 25.3 m from edge of alluvial 45 Table 8. Temporal distribution of sampling effort by 5+ station blocks and distance from edge of alluvial fan. Area Sampled (m2) Station Block Distance (m) June/July Aug./Sep. l - 5+ 0.0 - 13.5 9.00 5.63 6 - 10+ 13.5 - 28.5 6.19 10.69 11 - 15+ 28.5 - 43.1 8.44 11.25 16 - 20+ 43.1 - 57.9 8.44 7.31 21 - 25+ '57.9 - 73.0 6.75 5.06 26 - 30 73.0 - 84.7 6.75 9.56 31 130 3.38 3.38 32 200 3.38 3.38 33 320 0.00 3.38 46 fan) while the dorsal marked ammocetes were released at station 24 (13.7 m depth, 64 m from edge of alluvial fan). The temperature at both stations was 9 °C during release. Ammocetes were transported to the release points in sealed 24 liter containers using SCUBA and released from individual containers by removal of the cover with immediate inversion of the container and placement on the substrate. Inverted containers were not removed until all ammocetes burrowed (within 5 minutes). Ammocetes were released at station 24 in a five die pattern within a 1 m square while dorsal marked ammocetes were released in a four die pattern within the same amount of area. On 24 September 1987, the marked ammocete release points were sampled with 8 trap trays in a modified "X" pattern to measure the final density and the local pattern of dispersal at each point. To couple results of laboratory substrate selection experiments with probable conditions encountered by ammocetes released in 1985, six substrate cores were taken at each of the 1985 release points on 31 July 1988 using their LORAN C coordinates. Frozen at -20 °C as per 1987 methods, permeability was measured in three 47 cores from each station using the procedures outlined in appendix 2. RESULTS 1985 Field Results Daily median, minimum, and maximum temperatures at 9.1 and 21 m in 1985 are shown in figure 7. Median temperature at 9.1 m increased gradually to a maximum of 19.5 °C on 18 August while temperatures at 21 m remained relatively constant. Daily temperature variation was greater at 9.1 m than 21 m, exceeding 4 °C in conjunction with passing weather disturbances. The thermocline (10 i 1 °C) set up by 4 July at 11.6 m and progressively descended to a depth of 17 m by 31 July. The size distribution of sediment particles relative to depth, distance from the edge of the alluvial fan, and interdiction of the thermocline with the bottom is shown in figure 8. The sediment particle size distribution along the delta drop-off zone is characterized by a large sand fraction. The silt/clay content of the substrate increases with increasing depth and distance from the alluvial fan. 48 PO Ln F‘ D0 Ln C3 IUIIIII IU'UUII'I llllllllllljll U Ln I U I l l I I Temperature (deg. C) C3 1 l l l l I 05/25/95 06/25/95 07/26/95 09/26/95 09/26/95 Date Figure 7. Daily median, minimum, and maximum temperatures at 9.1 and 21 m during summer 1985. DEPTH (m) IO- I4- [8' 22- 26~ 30- 49 q THERMOCLINE —: I l 0 200 400 600 800 [000 [200 I400 I600 I800 DISTANCE FROM ALLUVIAL FAN (m) Figure 8. Composite relationship between depth, substrate particle size distribution, edge of the alluvial fan, and final depth of the thermocline. . 80/]5/5 76 /I5/9 46/39/15 32/55/[3 6/64/30 6 /%/ 36 0/58/42 distance from the The measured distribution of lentic ammocetes relative to all factors is delimited by the arrows. PERCENT SA ND /$lLT/CLAY Results of the handling mortality study were confounded by the loss of some individuals during loading of the cages on the bottom but of those individuals successfully transferred, 100 percent survived. As a check, cage sediments were carefully sifted for dead individuals upon retrieval. No evidence of dead individuals was found. Although problems with the suction system tied into the electrosampler array precluded collection of ammocetes with the submersible in 1985, emergent ammocetes were recorded visually during transects. A total of 69 ammocetes were located over approximately 116 102 within 50 m of the edge of the alluvial fan in an area of known distribution based on annual Bayer 73 treatments. Five additional ammocetes were located at depths of 15.2 to 19.2 m, ranging approximately 50 to 170 m from the edge of the alluvial fan. No marked or unmarked ammocetes were located with the submersible in deeper portions of the study site at or near release points C, B, or E. Sampled water was saturated with oxygen at all depths. A total of 4,216 ammocetes were recovered during the 26 July Bayer 73 treatment of 20,130 102 surface area 51 along the delta drop-off in depths ranging from 1.8 to 13 m. Of those recovered, 113 were recaptures of shallow released ammocetes. Also recaptured were 4 individuals from release point A, 4 from release point E, and 1 each from release points B and C. Three days later, an additional 10,065 m2 of surface area bracketing the above area was treated with Bayer 73 and an additional individual from release point C was recovered along with 430 unmarked ammocetes. Total bottom area treated (31,596 m2) was estimated by calculating the hypotenuse of a triangle formed by the surface widths of the treatment areas and change in depth, multiplied by the length of the treatment areas. The percent species composition of captured ammocetes was 19.5 P. marinus, 80.4 L. appendix, and 0.1 L. spp. The Bayer 73 treatment at Sand Point, approximately 3.5 km due west of the experimental area, on 30 July resulted in the recovery of another lamprey from release point A. In addition, during stream collections of ammocetes in May 1986, 1 ammocete from release point C was recovered in Sawmill Creek over 8 km from the study site and 1 each from release points C and B were recovered in Stokely Creek over 12 km from the site. &— _ —‘.._,—' ,— 52 1986 Field Results Results of the cross factor study of sediment particle size distribution versus thermal acclimation of ammocetes are shown in table 1. The greatest number of ammocetes were collected from the unmodified, unacclimated patch (40), while similar numbers (11 - 13) were collected in the other three patches. Only subsamples I and III were used in the model II ANOVA (table 9) because all samples taken in the middle of a patch were biased with respect to habitat modification by sand used as a burrowing substrate within release cages. Small sample sizes contributed to a mean square with a large error. There were no statistically significant differences. 1987/88 Laboratory Results Summarized results of substrate preference and food related habitat selection experiments are presented below with basic frequency analyses and abstracted ANOVA results. Appendix 4 contains results of tests for homogeneity of variance for both single experimental series and pooled data sets along with full ANOVA tables for all pooled analyses. Results of a stepwise linear 53 Table 9. Model II ANOVA for 1986 habitat modification/thermal acclimation experiment off the mouth of the Chippewa R. i Source of Variation df SS MS F P Habitat modification 1 3.12 3.12 0.609 0.505 Acclimation l 1.12 1.12 0.218 0.308 Interaction 1 1.13 1.13 0.220 0.309 Error 4 20.50 5.13 Total 7 25.87 54 regression analysis of mean burrowing depth with respect to substrate permeability, porosity, and ammocete weight follows the summarized results of burrowing substrate selection tests. Frequency Analysis of Substrate Preferences Table 10 provides a summary of all substrate preference/competition series with reference to tested substrates. Neither large L. appendix or P. marinus selected silt/clay or very coarse sand substrates (0.0315 and 1.5 mm mean grain diameter respectively). In each case, the data departed significantly from uniform or random substrate selection and the replicates were statistically homogenous. Consequently, all subsequent tests using the Wentworth scaled substrates were limited to very fine through coarse sands (0.0938 through 0.75 mm mean group grain diameters). With respect to statistical departure from random substrate selection and homogeneity of replicates, results of substrate preference tests for very fine through coarse sands were varied across experimental series. Generally, for L. appendix and P. marinus released allopatrically, fine and medium sands were preferred over very fine and coarse sands. This 55 Table 10. Pooled percentages of ammocetes occupying different burrowing substrates. Experimental series are identified by ammocete size, species, density (0) per test arena, release pattern, and number of replicates. Subgroups for all staggered releases are listed in sequence vertically. To indicate departure from random selection of substrates, significance levels for the likelihood ratio test (LRT; when n s 32) and G-test for pooled data (G ; Williams correction) are included. The significance level of Gh t provides an index of homogeneity among replicates within an experimental series. Experimental Series Mean Group Grain Diameter 'Signif. Level 0.0315 0.0938 0.1875 0.3750 0.7500 1.500 LRT Gwp Ghet Large 2. 222222i2. D8, unstaggered, 0.0 34.5 31.0 34.5 -- -- .0016 <.001 .9340 4 replicates Large 2. mazinufi. 08, unstaggered, 0.0 18.8 43.8 37.5 -- -- <.001 <.001 .7446 4 replicates Large 2. 22222212. D8, unstaggered, -- -- 53.1 34.4 12.5 0.0 <.001 <.001 .9129 4 replicates Large 2. 2221222. 08, unstaggered, -- -- 30.0 56.7 13.3 0.0 <.001 <.001 .3910 4 replicates small L. 22222212. D4, unstaggered, -- 15.6 21.9 31.3 31.3 -- .0033 .0029 .4480 8 replicates Small 2- 22222212. 08, unstaggered, -- 13.3 41.6 39.8 5.3 -- - <.001 .0958 16 replicates Large L- 22222212. D4, unstaggered, -- 21.7 25.0 38.3 15.0 -- - .1569 .0056 16 replicates Large 2- 22222212. 08, unstaggered, -- 23.3 23.3 37.9 15.5 -- - .0770 <.001 16 replicates Large L. 22222212. -- 12.9 29.0 45.2 12.9 -- 012, staggered, -- 25.0 28.1 34.4 12.5 -- - <.001 .0070 16 replicates -- 24.6 34.4 31.1 9.8 -- Small 2- 2221225. D4, unstaggered, -— 15.6 21.9 31.3 31.3 -- .5180 .5157 .6851 8 replicates Small 2- 2211225. 08, unstaggered, -- 19.4 41.9 19.4 19.4 -- - .0411 .4455 8 replicates Table 10 continued on next page Table 10 (cont'd). 56 Experimental Series Mean Group Grain Diameter signif. Level 0315 0.0938 0.1875 0.3750 0.7500 1.500 LRT Gwp Ghet Small 2. marinus, -- 10.0 50.0 20.0 20.0 -- .8295 .7979 .0043 D8, staggered, -- 26.7 20.0 33.3 20.0 -- 4 replicates Large 3. marinus, D4, unstaggered, -- 9 4 34 4 46 9 9 4 -- .0037 .0040 .0341 8 replicates Large 2. marinus, D8, unstaggered, -- 5.7 32 5 41.5 20 3 -- - <.001 .2841 16 replicates Large 2. marinus, -- 10. 30.8 38.5 20.5 -~ - <.001 .0220 D8, staggered, -- 11 1 36 1 47 2 5 6 -- 10 replicates Large 2. marinus, -— 2.1 22.9 58.3 16.7 -- D12, staggered, -- 6.0 30.0 56.0 8.0 -- - <.001 .0393 14 replicates —- 10.0 14.0 44.0 12.0 -- Large L. appendix: 2. marinus, D4:4, -- 9.7 38.7 41.9 9.7 -- .0081 .0071 .0924 unstaggered, -- 0.0 34.4 53.1 12.5 -- <.001 <.001 .4517 8 replicates Large L. appendix: 2. marinus, D8:8 -- 11.5 36.5 42.3 9.6 -- — <.001 .3589 unstaggered, -- o O 39.6 50 9 9 4 -- - <.001 .9452 7 replicates Experimental series -- mix A mix 8 mix C mix D -- LRT Gwp Ghet r Large L. 22222113: D8, unstaggered, -- 0.0 0.0 0.0 0.0 -- - - - 3 replicates Large 2. 22211212: D8, unstaggered, —- 0.0 0.0 0.0 0.0 -- - - - 3 replicates Experimental series -- 0.0938 0.1875 0.3750 mix D -- LRT Gwp Ghet small L. AELedLl-Xi D8, unstaggered, -- 34.4 31.3 31.3 3 l -- .0109 .0096 .6980 4 replicates Large L. appendix: D8, unstaggered, -- 25.0 34.4 25.0 15.6 .5399 .5267 .0375 4 replicates Small 2- 2%: D8, unstaggered, —- 26.7 20 0 46 7 6.7 -- .0180 .0182 .2081 4 replicates 57 preference was statistically significant for eight of twelve series. However, replicates were statistically homogenous in only five of twelve series. For large L. appendix and 2. marinus released sympatrically, both species exhibited significant departure from uniform substrate selection at densities of four and eight each per test arena. In both series, the replicates were statistically homogenous. In preference tests of graded substrate mixes A through D, none were selected by either species. Ammocetes attempted to burrow in all mixes but quickly emerged or gave up and lay prostrate on the bottom for the duration of the experiment. Based on the similarity in permeability between fine sands (0.1875 mm mean grain diameter) and substrate mix D, additional preference tests were conducted between mix D and very fine through medium sands. Small L. appendix and 2. marinus exhibited significant rejection of mix D. Replicates for both species were statistically homogenous. Large L. appendix also exhibited reduced usage of mix D, but the departure from uniform substrate selection was not significant. Replicates for large L. appendix were statistically heterogenous. 58 Multifactor Model I ANOVA's of Substrate Preference Test Blocking Factors Due to the similarity of results among experimental series within a given species and size class in table 10, frequency data were pooled by species and size for the model I ANOVA's. Proportional use data for very fine through coarse sands from all replicates were subjected to a square-root arcsin transformation prior to analysis (Sokal and Rohlf, 1981). Within given experimental series and pooled data sets, variances in frequency among substrates were generally homogenous. Exceptions were attributable to lower preferences for very fine and coarse sands resulting in lower variance at the extremes of the tested substrate distribution (appendix 4). To verify the apparent uniformity of substrate preferences across all test arena densities, ANOVA's on test series pooled across test densities by species and size class were conducted (table 11). With the exception of small P. marinus, the effect of substrate type on habitat selection was highly significant. Treatment effects of substrate position in the test arenas and density of ammocetes were nonsignificant. Only in the case of large 2. marinus pooled over all Table 11. series by species and size. 59 Results of multifactor model I ANOVA's for pooling of experimental Frequency of specific substrate selection (square root arcsin transformed) analyzed with respect to substrate type, substrate position in test arena, 5 = not significant; *, release pattern). it ***, and experimental series (eg. density [D]/ and **** represent significance at the 0.05, o. 01, 0. 001, and 0. 0001 levels respectively. 1 Significance Level Experimental Series Substrate Position Series Substrate Substrate Position only only x x x Position Series Series Sm. L. appendix, pooling of D4ns **** ns ns ns ns ns with D8ns Lg. L. a e , pooling of D4ns, **** ns ns ns ns ns D8ns, D125, D4:4ns, and D8:8ns Sm. mapinus, pooling of D4ns, ns ns ns ns ns ns D8ns, and D85 Lg. 2. marinus, pooling of D4ns, **** ns ns ns * ns D8ns, D4:4ns, D85, D125, and 08:8ns 6O densities was a significant interaction noted between substrate type and density of ammocete release. No interaction effects between substrate type and position nor position and density of ammocetes released were significant. Figure 9 displays the resultant 95 percent confidence intervals for transformed frequencies of selection with respect to substrate type. Within a species there appears to be a shift in preference from fine to medium sands between small and large ammocetes (figure 9, a to c and b to d). Within the small size class, selectivity for substrates was higher for larval L. appendix than for larval 2. marinus. Within the large size class, selectivity appears to be highest for larval 2. marinus with a marked preference for medium sands and avoidance of very fine sands in comparison to larval L. appendix. Results of an ANOVA to test for intraspecific differences in burrowing substrate selection within both species by size are shown in table 12. With respect to both species, substrate type was a highly significant factor in burrowing habitat selection. Position in the test arena was significant only for L. appendix. There was a clear interaction effect between substrate type 61 Figure 9. Ninety—five percent confidence intervals for mean values of test substrate selection frequency (square-root arcsin transformed) by species and size class. Mean grain diameters of 0.09, 0.19, 0.38, and 0.75 mm correspond to ungraded test substrates of very fine through coarse sands. 90) arcsin) Frequency (sqrt. 0.2 0.0 Small L. appendix Mean Grain Diameter (mm) 90) Large L. appendix arcsin) Frequency (sqrt. 0.2 0.0 0.09 19 0.38 0 75 Mean Grain Diameter (mm) 62 90) arcsin) Frequency (sqrt. 0.2 0.0 Small P. marinas 0.09 19 0 38 0.75 Mean Grain Diameter (mm) 9d) Large P. marinus arcsin) Frequency (sqrt. 1.0 0.8 0.6 0.2 0.0 D 09 0 19 0.38 0 75 Mean Grain Diameter (mm) 63 Table 12. Results of multifactor model I ANOVA's for size specific differences in substrate preferences. selection (square root arcsin transformed) analyzed with respect substrate type, substrate position in test arena, and Size class (eg. and **** represe ent Frequency of specific substrate t small vs. large). [ns = not significant; *, **, ***, significance at the 0.05, 0.01, 0.001, and 0. 0001 levels respectively. ] Significance Level Experimental Series .Substrate Position Size Substrate Substrate Position only only only x x x Position Size size Small vs. large L. appendix, pooled **** * ns ns * ns over all series Small vs. large 2. marinus, pooled **** n5 n5 * **** ns over all series 64 and ammocete size class. Interaction effects between substrate type and position in the test arena or position and ammocete size were nonsignificant. Table 13 shows the results of an ANOVA to test for differences in burrowing substrate selection between species in the same size class. With respect to treatment effects, only substrate type was significant. There was a significant interaction effect between substrate type and species. No significant interaction effects between substrate type and position in the test arena nor position and species was detected. To specifically test for the effects of intraspecific competition on burrowing substrate selection, an ANOVA was run on all staggered release experiments (table 14). With the exception of small 2. marinus, substrate type again exerted a highly significant effect on burrowing habitat selection. Treatment effects of position in the test arena and the release group number were not significant. There was no significant interaction effects between substrate type and release group with the exception of large 2. marinus at a density of 8 ammocetes per test arena. There were no significant interaction effects between substrate and position nor position and release group. 65 Table 13. Results of multifactor model I ANOVA's to identify species differences in substrate preferences within a given size class. Frequency of specific substrate selection (square root arcsin transformed) analyzed with respect to substrate type, substrate position in test arena, and species. [ns = not significant; *, **, ***, and **** represent significance at the 0.05, 0.01, 0.001, and 0.0001 levels respectively.] Significance Level Experimental Series :Substrate Position Spp. Substrate Substrate Position only only only x x x Position Spp. Spp. Sm. :2- 222229112 VS- 2. marinus, pooled **** ns ns ns ** ns over all series Lg. L. appendix vs. 2. marinus, pooled **** ns ns ns **** ns over all series 66 Table 14. Results of multifactor model I ANOVA's for testing of intraspecific competition effects on substrate selection. Frequency of specific substrate selection (square root arcsin transformed) analyzed with respect to substrate type, substrate position in test arena, and staggered release groups. [ns = not significant; *, **, ***, and **** represent significance at the 0.05, 0.01, 0.001, and 0. 0001 levels respectively. ] Significance Level Experimental Series :Substrate Pos. Rel. Grp. Substrate Substrate Position only only only x x x Position Rel. Grp. Rel. Grp. Large L. appgndix , D12, 3 staggered **** ns ns ns ns ns release groups Small 2. mapipus, D85, 2 staggered ns ns ns ns ns ns release groups Large 2. maripus, D85, 2 staggered **** ns ns * ns ns release groups Large 2. mgpinus, 0125, 3 staggered **** ns ns ns ns ns release groups 67 Table 15 contains the results of ANOVA's to test for the effect of interspecific competition on burrowing habitat selection. With respect to treatment effects, only substrate type was highly significant. With respect to interaction effects, including substrate type versus the presence or absence of the competing species, no interaction effects were significant. Stepwise Linear Regression Analysis of Burrow Depths One hundred and twenty-four observations of branchial basket depth in substrate mix D and very fine through coarse sands were subjected to back-stepped multiple regression analysis. Independent variables tested in the model were substrate porosity, permeability (natural log transform), and individual ammocete weight. At a significance threshold of 0.05, only permeability was included in the final model (a 5 0.001). With an R2 of 30.12 percent, branchial basket depth (D: cm) was related to permeability (K: cm min-1) by the following model: D = 1.4159 + 0.4655(1n K) The regression slope and intercept were significant at the a = 0.0001 level. Ninety-five percent confidence 68 Table 15. Results of multifactor model I ANOVA's for testing of interspecific competition effects on substrate selection. Frequency of specific substrate selection (square root arcsin transformed) analyzed with respect to substrate type, presence/absence of competing species. and **** represent significance at the 0.05, respectively.] 0.01, substrate position in test arena, and [ns = not significant; *, 0.001, *** **, I and 0.0001 levels significance Level Experimental Series ‘Substrate Pos. Compt. Substrate Substrate Position only only only x x x Position Comp. Comp. Large L. appendix, with and without **** ns ns ns ns ns g. marinus Large 2. r nus, with and without **** ns ns ns ns ns L. appendix 69 intervals of mean branchial basket depth are plotted against permeability in figure 10 along with the regression line. Frequency Analysis of Food/Habitat Preference Tests Table 16 provides a summary of all food/habitat selection tests with reference to selection of habitats with food versus those without. For large L. appendix and 2. marinus, the presence of food particles in the substrate did not have a significant effect on habitat selection. Replicates were statistically homogenous for both species. Results of habitat selection tests based on the presence or absence of food particles in the water column were more equivocal. For large L. appendix, the presence of yeast particles had a significantly positive effect on habitat selection while the presence of the green alga, A. falcatus, did not. The presence versus absence of yeast did not have a significant effect on habitat selection exhibited by large 2. marinus. Replicates were statistically homogenous for both species when a yeast perfusion was used but heterogenous in the case of large L. appendix and green alga perfusion. 7O I [1 I'VYjT T I VIII!!! l I 711111] -& (I in ~ llllllllllllllllllllllll C) L. 114411 1 n I‘lllll l n I 141111 1.0 IO I00 PE RMEAB/L/ TY (cm/min.) BRANCH/AL BASKET DEPTH (cm) ho .0IIIIIITIIIIIIIIIIIIIIIII Figure 10. Ninety-five percent confidence intervals for mean branchial basket depth of ammocetes with respect to substrate permeability. The regression line for corresponds to that calculated for a regression through all data points. 71 Table 16. Pooled percentages of ammocetes occupying different burrowing substrates based on presence or absence of food. Experimental series are identified by ammocete size, species, density (D) per test arena, release pattern, food type (yeast or algae perfusion), and number of replicates. To indicate departure from random selection of substrates, signifcance levels of G-tests for pooled data with Williams correction (Gwp) and heterogeneity across replicates (Ghet) are included. Experimental Series Treatment Level Signif. Level no food food Gwp Ghet Substrate based food/ habitat selection: Large L. appendix: D8, unstaggered, 50.0 50.0 0.8544 0.4690 yeast, 8 replicates Large 2. marinus: D8, unstaggered, 45.5 54.5 0.5036 0.4933 yeast, 8 replicates Water column based food/hab. selection: Large L. appendix: D8, unstaggered, 40.6 59.4 0.0464 0.4233 yeast, 8 replicates Large L. appendix: D8, unstaggered, 42.1 57.9 0.4140 0.0278 algae, 6 replicates Large 3. marinus: D8, unstaggered, 54.7 45.3 0.4815 0.4877 yeast, 8 replicates 72 Multifactor Model I ANOVA'a of Food/Habitat Preference Test Blocking Factors Table 17 contains the results of ANOVA's testing the effect of food particle distribution on habitat selection pooled by species, but separated by substrate versus water column experiments. In both sets of experiments, the treatment effect of food versus no food was nonsignificant. With respect to food perfused versus unperfused substrate tests, position in the test arena was significant. This was not the case for water column tests. There were no significant interaction effects between any of the main treatments. 1987 Field Results Daily median, minimum, and maximum temperatures at stations 10, 16, and 24 (10, 12.2, and 13.8 m depth respectively) are shown in figure lla-c. Daily temperature variation was greatest from mid-June through late August, exceeding 8 °C at all stations on several occasions. At each station, temperature increased gradually, peaking in mid to late August, and decreasing thereafter. The fine scale distribution of substrate particles relative to distance from the edge of the alluvial fan 73 Table 17. Results of multifactor model I ANOVA's to test the effect of food particle distribution on habitat selection. food perfused vs. unperfused habitats (square root arcsin transformed) analyzed with respect to treatment (food gs. no food), substrate position in test arena, species, and food type (yeast vs. algae). significant; *, **, 0.001, and 0.0001 levels respectively.] Frequency of selection for [ns = not ***, and **** represent significance at the 0.05, 0.01, Significance Level EXperimental Series Trtmnt Pos. Spp. Food Trtmnt Trtmnt Trtmnt P05. P05. only only only Type x x x x x Pos. Spp. Food Spp. Food Yeast perfused vs. unperfused substrate ns ** ns -- n5 n5 -- ns -- selection tests Food perfused vs. un- perfused water column ns ns ns ns ns n5 ns ns ns selection tests 74 Figure 11. Daily median, minimum and maximum temperatures at stations 10, 16, and 24 (10, 12.2, and 13.8 m depths) during summer 1987. 75 10 meters Station 10: Ha) A A A J no hm\mN\mo ho\ow\mo hm\wN\hD hm\mm\mo hm\mm\mo Date: .mwpv uLovoLadew» 12.2 meters Station 16: Mb) A A A 25 -' AU 20 - 15 - 10 - S- 0 -. hm\wm\mo ho\om\mo hm\mw\bo hm\mm\mo hm\mN\mo Date: .mopv aLovOLomeep 13.8 meters Station 24: Ho) A A A L- 25 20 - 15 P 10 - ho\mN\mo hm\mN\mo hm\mm\ho hm\mN\mo hm\mm\mo Date: nu .mapv 0L5¢OLoeeoh 76 is shown in figure 12. As in 1985, the leading edge of the alluvial fan is characterized by a sand fraction exceeding 80 percent which generally declines with increasing distance from the alluvial fan. The fine scale measurement of particle size distribution also reveals a localized minima of sand grains with a matching maxima of silt/clays. Substrate permeability in the top 4 cm of substrate, initially high along the leading face of the alluvial fan, rapidly declines with an isolated increase at station 10 corresponding to the local minima of sand particles (figure 13). Permeability again declines with increasing depth to a local minima at a distance of 82 m from the edge of the alluvial fan. Beyond 82 m from the edge of the alluvial fan, permeability ranges between 1 to 2 cm min‘l. Substrate porosity follows a different pattern from that of permeability. In the upper 4 cm of the substrate, porosity is lowest at the upper edge of the alluvial fan, rapidly increasing to a local maxima along the face and lower edge of the alluvial fan (figure 14). This area corresponds to an active slumping zone where the slope is at its maximum angle of repose (30-35°). A local minima in porosity at station 10 corresponds to 77 100 -l 1 I I fi— 80— L “D . .. C a C3 . U" 50 - - p I g - . 0 40.- 4 L _ ‘ w . I Q' 20- - D l L 1 n A l a l a n I n a n n l A 1 1 a [-l o 100 200 300 400 Distance (m) Figure 12. Percent sand composition of substrate with respect to distance from the edge of the alluvial fan during summer 1987. In K (cm/min.) 2.0 1.5 1.0 0.5 78 1. ' I I I I l I I I I l I I I I I I I I I I I I I I ' )- h I I I I I I I I I I I I I I I I I I I I I I I I I I C3 100 Distance (m) Figure 13. Profile of mean substrate permeability (i sd) with respect to distance from the edge of the alluvial fan during summer 1987. All permeability values were determined from frozen cores. 79 .- .— .. .— 100 80 0) C3 Porosity 8 00 CD I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I D 100 200 300 400 Distance (m) Figure 14. Porosity at 4 cm depth intervals in the substrate with respect to distance from the edge of the alluvial fan during summer 1987. All porosity values were determined from frozen cores. 80 the local minima of sand particles in figure 12. After station 10, at 22 m from the edge of the alluvial fan, porosity again increases to a local maxima at 82 m, corresponding to a local minima in permeability and maxima in sand fraction. Beyond 82 m from the edge of the alluvial fan, porosity in the upper 4 cm decreases gradually with increasing depth and distance. At substrate depths exceeding 4 cm, porosity reaches a local maxima in the area between the active slump zone and station 10. From a local minima ranging between 40 to 65 m from the edge of the alluvial fan, porosity at substrate depths greater than 4 cm remains more or less constant at a value of 50 to 60 percent. Food particle distribution in the top 4 cm of the substrate based on chlorophyll a concentration is shown in figure 15. In general, the concentration of chlorophyll is lowest at the upper edge of the alluvial fan increasing to a local maxima at a distance of 15 to 25 m and tapers off thereafter. Over the course of the summer, peak chlorophyll concentrations decreased from a high of 7.5 to a low of 3.6 ug/g within 20 m from the edge of the alluvial fan. The ratio between chlorophyll a and pheophytin a generally decreases with increasing depth and distance from the edge of the alluvial fan 81 e- _ C” . 2 -Jul. 03 8 --Jul. 25 . 2 --Aug. 21 - .3 -"Sep. 22 . 3 . U) . E7 ‘ / . U \\ . . / \ . .4 Mi, \1 _____ _ 1: ‘*""— rrrr --~~~~—m- . U 1 a? . D . l I J I I 0 100 200 300 400 Distance (m) Figure 15. Chlorophyll a in the top 4 cm of the substrate with respect to time and distance from the edge of the alluvial fan during summer 1987. 82 (figure 16). A localized temporal progression of declining index values, centered at a distance of 22 m from the alluvial fan, corresponds to the local minima of sand particles in figure 12. In terms of chlorophyll a concentrations, the food particle distribution at the sediment/water interface follows the same general trend as that in the substrate (figure 17). The chlorophyll/pheophytin ratio at the sediment/water interface also exhibits a general decrease with increasing depth and distance from the edge of the alluvial fan (figure 18). In contrast to the distribution of chlorophyll a, the distribution of organic particles less than 180 um exhibits a local minima at a distance of 22 m from the edge of the alluvial fan (figure 19) except in late September. This region corresponds to the local minima of sand particles in figure 12. Trap tray effort levels pooled by 11 m distance increments and time (early versus late summer) are listed in table 8. The average density of unmarked ammocetes versus distance from the edge of the alluvial fan is plotted in figure 20. In general, density decreases with increasing distance from the edge of the alluvial fan to a local minima at 45 m distance. Beyond 83 4_.. ... ,. ... ,. ... ,. ...;fi - -Jul. 03 I 2; --Jul. 25 : - - Aug. 21 . ' -"Sep. 22 - o- _ I—I '- u a. - . L) 2- f . - - 0A, \ u- . VV \‘\\\\ . . ' . /"”“*—r:‘””‘”” ' -4 _ L //’ “‘~ _ I P” 3 -6 -| . . I I 1 . . . . l . . . I 1 . . . . 1- o 100 200 300 400 Distance (m) Figure 16. Chlorophyll/pheophytin index (CPI) values in the top 4 cm of the substrate with respect to time and distance from the edge of the alluvial fan during summer 1987. Index values are calculated as the natural log of the ratio of chlorophyll to pheophytin concentrations. An index value above zero reflects a greater proportion of chlorophyll while values below zero reflect a larger proportion of pheophytin. 84 2.5 .- .- : _1U1 gg/zg : . -- L1 _ . D 20_ --Aug. 21 i 3 I -“Sep. 18/19 2 221.5: J o / ______ : 511.0 J .c . U I g 0.5 _ _ .: : tar—MM”— 5 0'0 :1 . 1 1 I I' 0 100 200 300 400 Distance (m) Figure 17. Chlorophyll a concentrations at the sediment/water interface with respect to time and distance from the edge of the alluvial fan during summer 1987. 85 a :l ' ' I I T j ' ' I l-l.I 4:- /\/‘>K.. \\\ 3 . \\\ \‘\\ 2 : \\ \\ : H 2 _ \\\ \ _ . \. .. a. - \ . LJ 0 E \\\\‘ : : —-Jul. 05 I -2: --Jul. 28/29 l I --Aug. 21 I -4: -~Sep. 18/19 J -6 LI . n . n I n n a n I n L4 J I n a n n I.-II 0 100 200 300 400 Distance (m) Figure 18. Chlorophyll/pheophytin index (CPI) values at the sediment/water interface with respect to time and distance from the edge of the alluvial fan during summer 1987. 86 Au.nos_ . C -_- -Jun. 14 ; 370004- -- Jul. 04 - - E --Jul. 26 T g _ "‘ Aug. 21 : 0.003' ° Sep.19 '5 c: . a) _ I H I. . V 0Jm2_ ‘ 8 / ' : / I E nmml . //’ ‘ o ' /’ : ET) “,5r””//// . D 00000 -. 4 l.;. . 1 .4 l . 1 l 0 100 200 300 400 Distance (m) Figure 19. Distribution of particulate organic carbon (POM, diam. < 180 um) at the sediment/water interface with respect to time and distance during summer 1987. Sample from station 29 on July 26 was fouled with sediment during sampling and discarded. 87 2. 5 I ' fi I I l I A I E 2J1: - 5’: . \\ 1.5 ’ 1 =11: . . V b >.1.ol _ +J : I ...—q . d U) . C 0. S - - 0’ . D 0'0 l . . . . I . . . . l 1 I . . . . 1 0 100 200 300 400 Distance (m) Figure 20. Average density of ammocetes with respect to distance from the edge of the alluvial fan during summer 1987. 88 45 m from the alluvial fan, density increases to a local maxima at a distance of 78 m. Ranging from 20 to 82 m from the edge of the alluvial fan, the distribution of ammocetes corresponds to that of the sand particle distribution in figure 12. Figure 21 shows the time and pattern of marked ammocete recaptures along with the final densities and local dispersal patterns at marked ammocete release points. In contrast to initial densities of 520+ ammocetes m‘z, final densities at stations 10 and 24 were 6.7 and 13.8 m'2 respectively. Immigrants from station 10, captured over the course of the summer, were caught in deeper waters during late summer. Immigrants from station 24, captured over the course of the summer, were caught in both shallower and deeper waters in early summer. DISCUSSION Local Distribution and Abundance Based on recaptures of marked shallow released ammocetes from the Bayer 73 treatments in 1985, a total of 53,044 ammocetes resided within the treatment area with a 95 percent confidence interval of 43,469 g N g 89 Figure 21. Recovery map of marked ammocetes released at stations 10 and 24 during summer 1987. Stations 10 and 24 were sampled on 23 September with eight sample trays arranged in the illustrated pattern. Each square represents one tray (0.5625 m2) and the number of marked ammocetes recovered. Migrants from the release points are identified by release point, date, and location of recovery. Figure 2/ DISTANCE FROM EDGE ,OF ALLUVIAL FAN (meters) IO 20 30 4O 5O 60 7O 80 °1 9O -4 O Q ‘———- SIO: 9-23 3247-30 9 #1333121 ‘._——-— 524: 7- 03 91 62,619 (Chapman, 1951; Ricker, 1978). This estimate suggests an average density of 1.68 i 0.32 ammocetes per m2 bottom area of which 19.5 percent were larval sea lamprey. An average density estimate of 0.997 ammocetes per m2 obtained from the submersible transect data corrected for electrosampler efficiency yields a total estimate of 31,490 ammocetes within 50 m of the edge of the alluvial fan. From a distance of 50 to 200 m from the edge of the alluvial fan, an average density estimate of 0.108 ammocetes per m2 within and below the thermocline was obtained with the submersible. The magnitude of 1987 density estimates compares favorably with those of 1985. The density of ammocetes within 55 m of the edge of the alluvial fan ranged between 1.2 and 0.2 individuals per m2 based on 1987 trap tray results (figure 20). This area corresponds to the area sampled in 1985 with the submersible and Bayer 73. Within 30 m of the edge of the alluvial fan, the area normally subjected to annual Bayer 73 treatments, densities ranged between 0.5 and 1.2 individuals per m2. The higher densities of ammocetes (1.2 to 2.2 per m2) between 55 and 87 m from the edge of the alluvial fan were not detected in 1985. However, comparable depths sampled by the submersible in 1985 were located 300 m to the north of the 1987 transect. 92 Analysis of Habitat Selection Determinants 1985/86 Preliminary Analysis The distribution of ammocetes off the mouth of the Chippewa River appears to be limited to a zone parallel to the leading edge of the alluvial fan based on the Bayer 73 treatment records and the submersible work in 1985. As per working hypothesis 1, this is where one would expect to find them if they simply drift out of the river during seasonal flooding and select the first substrate they encounter (figures 8, 12). This area is an active deposition zone for large particles, containing an abundance of leaves, sticks, and occasional branches. The tendency for ammocetes to settle out in these areas has given rise to the argument within sea lamprey control circles, that lentic ammocetes do not actively select habitats. However, the absence of marked and unmarked ammocetes in the deep portions of the study site with the subsequent occupation of shallow, high gradient habitats by marked ammocetes released in deep water suggests otherwise. The dramatic distances (from 1 to 12+ km) that deep released ammocetes traversed to reach such habitats underscores this conclusion. 93 With respect to environmental correlates of lentic ammocete distribution and abundance, marked and unmarked ammocetes selected areas characterized by sandy substrates and temperatures greater than 10 °C in concordance with working hypotheses 2 and 3. As water was in equilibrium with atmospheric oxygen concentration at all depths, rejection of deepwater habitats based on low oxygen tensions (working hypothesis 4) can be dismissed. Deep portions of the study site are characterized by sand fractions less than 10 percent while areas where ammocetes were found had sand fractions greater than 35 percent with correspondingly low clay fractions (figure 8). Malmqvist (1980) noted that sediments with a relatively larger fraction of fine sand particles (0.125 - 0.25 mm diameter) contained significantly higher densities of larval brook lamprey (Lampetra planeri) in a Swedish stream. The same general preference has been qualitatively noted for all lamprey species in the Great Lakes basin (Applegate, 1950; Manion and McLain, 1971). Consequently, rejection of deepwater habitats may have been mediated by substrate type and/or unidentified substrate property. 94 The general distribution of ammocetes along the slope also suggests that thermal preferences played a role in rejection of deepwater habitats. The deep released marked ammocetes were not acclimated to the thermal environment of the release points, which may have affected their survival and behavior. However, the results of the handling mortality study do not support the hypothesis of increased mortality below the thermocline. Consequently, marked ammocetes may have rejected deepwater release points on the basis of non— acclimation or potentially, on the basis of thermal habitat selection driven in part by recent thermal history. The cross factor study of substrate particle size Idistribution versus thermal acclimation in 1986 was designed to test the relative importance and interaction of both factors. Based on the results, neither factor appears to play a significant role in terms of habitat selection. While care was taken to insure that the experimental patches were located below the thermocline, they were inadvertently located in an area characterized by sand fractions exceeding 32 percent. As a result, the quality of the substrate treatment portion of the experiment is suspect. 95 A decline in density within each patch is to be expected given the high initial release densities (1000 individuals 111'2 initially, 55 individuals 111"2 after cage removal) but the high densities used here to insure sampling success may have led to over-dispersal upon release. Adverse density-dependent effects could have promoted excessive emigration directly through competitive interactions or indirectly through a build up of metabolic products. Although not measured in 1985/86, the distribution and quality of chlorophyll in surface sediments and the water column, which in turn reflects the distribution and abundance of autochonous particles within the size range filtered by larval lamprey, has been shown to be inversely related to depth and distance from the shore (Stevenson and Stoermer, 1981; M011 and Brahce, 1986; Nalepa and Quigley, 1987). This relation suggests that larval lamprey distribution in lentic areas could also be related to local patterns of food abundance and quality. Furthermore, the strong differential in water density over a short distance resulting from the interdiction of the thermocline with the bed slope suggests that food particles could be concentrated close to the bottom and above the thermocline in depths 96 ranging from 9 to 17 m depending upon the depth of the thermocline. 1985/86 Working Conclusions The results of the cross factor study while not conclusive due to small sample sizes (potentially related to over—dispersal) and a reduced treatment differential for substrates, suggests that other factors may influence lentic habitat selection. These include substrate properties unidentified and/or uncontrolled in the 1986 experimental design, density-dependence in burrowing substrate selection, local food particle distribution, and thermal habitat selection driven in part by trends in the photo-period or recent thermal history. Taking each as working hypotheses, laboratory and field research in 1987/88 was designed to clarify the importance of each factor. 1987/88 Analysis In the laboratory, ammocetes exhibited a highly variable response to substrate type as witnessed by the significance of Ghet for experimental series listed in tables 10 and 16. Based on the model I ANOVA's, the variability among replicates in a series was not related 97 to treatment or interaction effects. The same variability was shown in the regression model of branchial basket depth versus permeability, where the R2 value was only 30.12 percent. Consequently, though selective preferences were expressed on average, it is clear that "suitability" of a burrowing substrate is broadly defined on an individual basis . The laboratory results indicate that both mean grain diameter and permeability set limits to burrowing substrate suitability. Mean grain diameter set an upper limit to substrate suitability while permeability defined the lower limit (figure 22). Very coarse sands (1.0 - 2.0 mm diam.) were rejected by all test subjects while permeability set the lower limit through its relationship to burrow depth, presumably in relation to resistance to respiratory currents. There is no evidence that mean grain diameter pg; §g sets a lower limit to substrate suitability based on 1987 field measures of permeability with respect to mean grain diameter (figure 22, small squares). Even though silt/clays and sand/silt/clay mixes were rejected in the laboratory, substrates in the field with mean grain diameters and similar particle size distributions were occupied by ammocetes. It should be noted however, that rV'V.‘ ‘ 7:3'7:. 98 1000 I . . ....n, . . ....n, T7. .... - I | a 100 - I3| ' A l C l .H 10- D I ' E l E XX 0 u '0 - I x '2,- D o \9 1" x _: '- x l 0.1r o : ' D 0 0'01-1 . . ......I . . ......I . . ......l- 0.01 0.1 l 10 Mean Grain Diameter (mm) Figure 22. Permeability of field and laboratory substrates with respect to mean grain diameter and threshold values of permeability and maximum mean grain diameter. Large squares and diamonds correspond to Wentworth scaled and mixed laboratory substrates respectively. Small squares correspond to mean values of substrates along the summer 1987 while the "x" symbols correspond to mean values measured for substrates at 1985 deepwater release points. 99 deepwater release points with even smaller mean particle diameters were rejected in 1985, though permeabilities measured in 1988 exceeded the threshold value (figure 22, x symbols). Thus substrates with mean particle diameters below 0.06 mm may be rejected though permeabilities surpass the threshold value. However, larval B. marinus readily accepted diatomaceous earth with a mean grain diameter less than 0.015 mm (Mallatt 1982). These results illustrate the importance of the substrate fabric or lattice to properties such as permeability. Webb (1975) clearly defined the effect of substrate consolidation on permeability and related this to substrate selection exhibited by lancelets, Branchiostoma lanceolatum. Webb and Theodor (1972) noted that the growth of organic films on substrate particles could increase permeability by 70 percent. Laboratory substrates in this study were well consolidated in comparison to field substrates and devoid of any organic films. Consequently, the relationship between mean grain diameter and permeability in the laboratory substrates differs from that exhibited by substrates in the field. In support of the qualitative conclusions reached by Mallatt (1983) and Norman (1987), laboratory analysis 100 of substrate selection did not provide evidence for direct density-dependent interactions between ammocetes. While crowding may affect the suitability of a substrate over time as per Mallatt's hypothesis, there is no short-term effect on burrowing substrate selection. The results of laboratory analyses of habitat selection based on local food particle distribution in the water column and substrate, indicate that the latter is also not a determinant of ammocete distribution and abundance. Comparison of the 1987 distribution of lentic ammocetes with the distribution and quality of food particles in the substrate supports this conjecture. Concentrations of food particles in the substrate were highest throughout the summer within 60 m of the alluvial fan while quality was highest within 30 m (figures 15, 16). In contrast, the average density of ammocetes increased with increasing distance from the edge of the alluvial fan to a local maxima at 80 m (figure 20). Results of the laboratory tests of food particle distribution in the water column on habitat selection were more equivocal. While larval E. marinus did not exhibit selection for yeast perfused waters in the frequency analysis, larval L. appendix did (a = 0.0464, 101 table 16). However, larval L. appendix did not exhibit selection for waters perfused with A. falcatus. When pooled together across species and food types in multifactor model I ANOVA's, there was no indication of significant habitat selection with respect to treatment or interaction effects (table 17). Comparisons of the 1987 distribution of lentic ammocetes with that of food particles at the sediment/water interface are equally equivocal. Concentrations of food particles in terms of chlorophyll a are generally higher within 60 m of the alluvial fan, decreasing with increasing distance thereafter with the exception of late September (figure 17). This pattern is similar to that for substrate chlorophyll concentrations. However, concentrations of food particles in terms of particulate organic matter (POM) peak at the base of the alluvial fan, show an average local minima at a distance of 25 m, and then increase with increasing distance thereafter, again with the exception of late September (figure 19). This pattern roughly matches that of the average distribution and abundance of ammocetes in figure 20. If the overall results of the food/habitat selection experiments in the laboratory are taken at 102 face value, the rough correlation between average ammocete distribution and POM is either coincidental or evidence of a physical process affecting both in a similar fashion. The best single explanation revolves around local current patterns. The simplest model depends only upon the interaction of wind driven surface currents with river discharge. The substrate particle size distribution in figure 12 indicates the presence of a hydrodynamic wall at the confluence of bay waters with the outfall of the Chippewa River. Southwest and westerly winds predominate in the spring and summer, setting up wind driven surface currents orthogonal to the face of the alluvial fan (figure 23b). This process leads to a strong deposition zone for fine organic and inorganic particles at a depth of 10 m and distance of 25 m from the upper edge of the alluvial fan. On the bay side of the hydrodynamic wall, finer organic and inorganic particles are then picked up by wind-driven vertical circulation cells and swept out to secondary deposition zones, 55 and 82'm from the edge of the alluvial fan. Empirical support for this model is found in the relationship between daily temperature variation measured at stations 10, 16, and 24 with local 103 Figure 23. Schematic representation of current regimes at the edge of the alluvial fan off the mouth of the Chippewa River in spring, summer, and fall. DEPTH 0n) [2 I2 I8 I8 104 I —-.-.--.9 230) ..‘s’..... . ‘ o O 23b) (:::—}::9 <:; SMMMER 23c) IIIIIJIIJJIIIIIIIIJI 20 4o 60 80 I00 (”STANCE On) 105 meteorological conditions at Sault Ste. Marie, Ontario, 35 km south of the study site (Meteorological Summaries, Environment Canada). To test this supposition, average wind speed and direction along with degrees of wind shift over 24 and 48 hour periods were used as blocking factors in a series of one-way ANOVA's of daily temperature range at depth. The significance levels for each analysis are listed in table 18. Individual significance levels for average wind speed, direction, and degrees of wind shift over 24 hours increase with increasing depth and distance from the edge of the alluvial fan. In contrast, the significance level for degrees of wind shift over 48 hours decreases with increasing depth and distance. As shown in figures 24 - 26, maximum daily temperature ranges at station 24 were correlated to winds southwest to westerly, high, and steady. Maximum daily temperature ranges at station 10 were correlated with variable winds over a 48 hour period (figure 27). Station 24 is the farthest temperature station from the hydrodynamic wall and thus most likely to reflect the effects of vertical circulation cells alone. Station 10, being the shallowest and closest temperature station 106 Table 18. Significance levels for one-way ANOVA's of daily temperature range at stations 10, 16, and 24 with respect to average wind direction, speed, and wind shifts. ANOVA Significance Level (0) Blocking Factor Sta. 10 Sta. 16 Sta. 24 Average Wind Direction 0.893 0.223 0.011 Average Wind Speed 0.663 0.150 0.007 Wind Shift over 24 hrs. 0.750 0.174 0.058 Wind Shift over 48 hrs. 0.053 0.225 0.674 107 A L) . 10 pl 1 l l I l 1 l .. CD ' ‘ w : T I B a - .. OI . . m '- - 0 q C 6 - .. F .. U " q 0: - ._ . ' 4 -' -.. ‘1 d. u a. , J . E _ .. ’ -. . 01 _ . '— 2 L- 4- “ - sf ' "' ‘ 9’ I I 0' D -1 I l l l 1 l I d '1" m u: w 3 3 3 3 U) z I a) m m z I Lu U) U) '3 I 3 Z I I I 3 2 I.“ m U) 3 Z Z 3 U) 3 Ave. Wind Direction Figure 24. Ninety-five percent confidence intervals of daily temperature range at station 24 with respect to average wind direction during summer 1987. 108 A L) 12”I I I I I I I I_ U? I -- I w . d “U 10- .. v . .. . 8’7 83- .. -'- g I 1 0: 5:. _P " .: EL 4- " l w I I -. I H- _ -- . -¢ 23' i N - . 0' D l 1 l l L l l 1- 4..) m ‘7’ $ $ I $ $ :1 (D an 00 V? C3 vfl F! (v on Ave. Winds (km/h) Figure 25. Ninety-five percent confidence intervals of daily temperature range at station 24 with respect to average wind speed during summer 1987. 109 LlJllllllllLlLllllllllllllllllllIll. A L) . S :I I l I I m . CD : w 4- 'o : " __ V’ : 3r w . u g7 L’ o 2? __ CK ; .. 1 r 01 t E L w a: F' : ‘<-'lf " “1 : . -2 L l l ‘ l 'l 3 0-45 46-135 136-315 U) Wind Shift (degs. 24 h) Figure 26. Ninety-five percent confidence intervals of daily temperature range at station 24 during summer 1987 with respect to degrees of wind Shlft over 24 hours. 110 A L) . 53‘ ' ‘ ' d. m - . U7 . q o . q :9 - . S - -- d w j ' 07 _ - 8 - ‘ m 4 " .. a ' -_—. J 0'. _ -- 3 E - u -- - w 3- _ p_. . .. . c: I -‘ ' ... _ —- j 2 . l l L l l 0 En.) 0-45 46-135 136-315 Wind Shift (degs. 48 h) Figure 27. Ninety-five percent confidence intervals of daily temperature range at station 10 during summer 1987 with respect to degrees of wind shift over 48 hours. 111 to the hydrodynamic wall, is most likely to reflect temperature variations due to turbulent mixing of bay and river waters and the effects of shifting weather patterns. Based on empirical relationships between fetch, wave height, wave length, and depth, sufficient vertical oscillation of the water mass to transport unconsolidated silts, clay, and fine organic particles can be generated to a depth of 15 m at the study site. With an average fetch of 5 km from the west, the theoretical maximum wave height at the study site is 0.75 m (Hutchinson, 1957). Wave heights at the edge of the alluvial fan approached this value many times in association with westerly winds (pers. obs.). Empirically, the ratio between wave height and length ranges from 1:10 to 1:100 (Wetzel, 1975). Local minima of sand fraction are spaced an average of 32 m apart (figure 12). Assuming they represent the spacing of vertical circulation cells as per figure 23b, the resultant wave height:length ratio of 1:42 falls within the middle of the range. If the amplitude of vertical oscillation cells is halved for each depth increase of the local depth divided by nine (Wetzel, 1975), then the vertical oscillation at depth is approximated by the following equation: 112 _ e[(-6.2383d)/(hL)] Where: 0d = vertical oscillation at depth d (m) d = local depth (m) h = wave height (m) L = wave length (m) At station 24 with a depth of 13.8 m and assuming a wave height of 0.75 m with a length of 32 m, the vertical oscillation of the water column at the substrate/water interface should be approximately 2.8 cm. The rough correlation of substrate sand fraction between 25 and 82 m from the edge of the alluvial fan in figure 12 with that of average ammocete distribution in figure 20 may be explained with the model in figure 23. As the vertical circulation cells sweep across the bottom, they irrigate the substrate. Those areas characterized by an increasing sand fraction with increasing distance from the edge of the alluvial fan match locations subjected to sweeping from vertical circulation cell currents. These areas also match local maxima of ammocete density. Webb (1975) noted the importance of similar currents driven by surface pressure waves in crest and ripple substrate formations and related them to the 113 local distribution of amphioxus. Station 10 corresponds to a slack current region based on figure 23, while station 24 corresponds to an active current region. Of the marked animals released at stations 10 and 24, twice as many remained at station 24 through the summer. The model outlined in figure 23 also provides an explanation for the shift in POM distribution from summer to fall. In September, prevailing winds shift from the northwest to northeast and the temperature of the river drops significantly below that of the bay. This leads to the creation of a shallow current of river water flowing over the bottom (figure 23c). In the field, this current was quite strong, moving not only fine particles across the bottom but also small twigs and leaves within 40 m of the alluvial fan (pers. obs.). Beyond 70 m, the current was not as noticeable. Evidence for temporal shifts in ammocete distribution and abundance in response to the local thermal habitat was also obtained from summer 1987 data. Early versus late summer distribution and abundance of ammocetes is shown in figure 28. In early summer (June/July), ammocete density within 60 m of the edge of the alluvial decreased with increasing distance. Ammocetes within 40 m of the edge of the alluvial fan 114 I I 1 I ' fi 1 V I T f ‘— I U I I ' j 1 U 2.5:- ... E an: -Aug./Sepq g E 3 \ 1.5: 1 at .. . v _ . . 4 >\1.U- - +3 Z + ...—Q . . (I) .. . C 0.5- .. (II P- . D : 3: O'DLJ LI I l l . I 1 I l 4 Lid 0 40 80 120 150 200 Distance (m) Figure 28. Early gs. late summer distribution and abundance of ammocetes with respect to distance from the edge of the alluvial fan during summer 1987. 115 moved to deeper waters further offshore between early and late summer. An analysis of covariance for differences among slopes was significant at the a = 0.005 level. Ammocetes located beyond 60 m from the edge of the alluvial fan did not demonstrate temporal shifts in distribution and abundance. To track average changes in temperature with respect to depth and distance from the edge of the alluvial fan, daily median temperatures at stations 10, 16, and 24 were subjected to third degree polynomial smoothing and plotted together (figure 29). Smoothed temperatures from 9.1 and 21 m depths in 1985 were included for reference. From this a series of temperature isoclines, including that of the thermocline and the median thermal preferendum, were derived (figure 30). Based on figure 30 and diving observations, the thermocline set up by 14 June at an average depth of 12.2 m and progressively descended to a depth of 18 m by 31 July. By early August, median temperatures had reached 17 °C at a depth of 12.2 m and distance of 43 m from the edge of the alluvial fan. The 14 °C isocline tracks the downslope movement of the thermal preferendum measured by Reynolds and Casterlin (1978). By early 116 25 L— I I I I A _ : L) n a . 20 ' .. CD : / "" \ \ : C” r / / . 'o . / . V 15 - .: 8 = "’ i 3 NE” 2 ,,- ——————— ,, € C3 r . a : = o. 5 f '2 E . . a] . . F- - 1 D I-J i I . I I '7 05/25/87 06/25/87 07/25/87 08/ 28/ 87 09/ 25/ 87 Date Figure 29. Polynomially smoothed daily median temperatures at stations 10, 16, and 24 during summer 1987 (solid lines) and at 9.1 and 21 m depths from summer 1985 (dashed lines). DEPTH 117 6 -.I I I I I 8— _. 10° /4° 17° [OI—D D o .... 12 .. \\\\\~ \\\\\\h~\\\\\\“\\ _ °\ o\ 0 l4 ._ a D ‘— \\\ \\ \ ’6 r- \\\ — \ \\ I8 - \ - 20 I I I I I 05/27 06/20 07/14 08/07 08/3! Ch4715 Figure 30. Derived temperature isoclines with respect to depth and date during summer 1987. The 10 °C temperature isocline tracks the downslope movement of the thermocline after 14 June. 118 August, median temperatures of 14 °C intersected the bottom at a depth of 14 m and a distance of 70+ m from the edge of the alluvial fan while temperatures within 20 m of the edge of the alluvial fan exceeded 19 °C. The dramatic shift in ammocete abundance within 20 m of the edge of the alluvial fan corresponds to median temperatures exceeding 19 °C. This matches the upper limit of voluntarily selected temperatures noted by Reynolds and Casterlin. The general pattern of ammocete movement within 40 m of the edge of the alluvial fan corresponds to that expected if ammocetes were roughly tracking the upper end of the thermal preferendum. The same pattern was observed in the marked ammocetes released at station 10 where the pattern of dispersal at the release point was biased towards deeper waters and migrants were caught at increasing depths over the course of the summer. Conclusions Acting as delimiters of substrate "suitability", mean grain diameter and permeability are both determinants of larval lamprey habitat selection. The failure of the 1986 cross-factor study to reflect a substrate treatment effect is attributable to both 119 modified and unmodified patches meeting the "suitability" limits. Whether mean grain diameter sets a lower limit to substrate suitability remains unclear but the ready acceptance of diatomaceous earth as a burrowing substrate noted by Mallatt (1982) argues against the proposition. While mean grain diameter and permeability set limits to substrate suitability, within these limits, the individual variation in substrate selection is high. The local thermal regime also drives larval lamprey ihabitat selection based in part on the recent thermal history of the individual. Demonstrated in the field by the shift to cooler waters with temperatures exceeding the upper end of the thermal preferendum, the same mechanism may also explain the rejection of 1985 deepwater release points by marked ammocetes. Though cold-acclimated ammocetes in Reynolds and Casterlin's work initially avoided temperatures of 10 - 12 °C, they gravitated towards the 14 °C preference point within a week once aware of the availability of warmer temperatures. The vertical circulation cell/current model may also explain the selection against deepwater release points in 1985 and 1986. At depths of 21 m, the total 120 vertical oscillation available to irrigate the substrate under the assumption of wave heights of 0.75 m with a length of 32 m is only 4.3 mm. Consequently, the exchange rate for interstitial waters may be too low to adequately refresh interstitial water and/or remove metabolic wastes. Given that the substrates off the mouth of the Chippewa River meet the suitability limits set by mean grain diameter and permeability, the pattern of ammocete distribution and abundance can be explained by the interaction of local temperature changes, inferred hydrodynamics and annual Bayer 73 treatments. As represented in figure 23a, warmer river water discharge flows over the cooler bay waters. In the process, velocity drops and particulates settle out according to density. Ammocetes moving into the lentic area at this time also settle out with decreasing stream velocity. The average and early summer distribution of ammocetes within 60 m of the edge of the alluvial fan in figures 20 and 28 supports this conjecture. Prevailing westerly winds in late spring through mid summer as shown in figure 23b lead to the establishment of a hydrodynamic wall and resultant vertical circulation cells. With the establishment of 121 zones of preferential substrate irrigation, local ammocete distribution shifts in favor of well irrigated substrates. When annual Bayer 73 treatments within 30 m of the edge of the alluvial fan are conducted before local temperatures prompt movement further offshore as shown in figure 28, most ammocetes are killed. However, if treatments are conducted in late August or September (as is frequently the case), many ammocetes move out of the treatment area in response to temperature. This process leads to the locally high density of ammocetes between 60 and 87 m from the edge of the alluvial fan. The implications for sea lamprey control in lentic areas are clear. There is a time window for Bayer 73 treatments that must be met to insure successful treatment. Treated too early, water temperatures may not be high enough for efficient use of Bayer 73. Treated too late, when water temperatures have risen to the point where ammocetes migrate to deeper waters, the results will be equally unsatisfactory. While this research has defined proximate determinants of habitat selection (eg. those that act over short time periods), ultimate determinants such as the individual assessment of energetic return over long 122 time spans have not been addressed. Before strong statements about ultimate determinants of habitat selection can be made, a better understanding of the energetic requirements for growth of ammocetes is required. This information in tandem with the understanding of the proximate determinants of habitat selection provided here is needed to create a null model of habitat selection against which ultimate determinants of habitat selection can be addressed. APPENDICES APPENDIX 1 Laboratory Measurement of Porosity Assuming complete saturation of sediments, and a density of 1.00 g/ml for water, substrate porosity was calculated as the difference between wet and dry weight of the core segment divided by the volume of core segment. Cores were extruded by hand from the core tubes while frozen and cut into 4 cm segments with a nichrome wire heated with an electrical current. Slow freezing of sediment cores causes some deformation of the substrate lattice as the outer edges of the core freeze at a slightly faster rate. This acts to force the central portions of the core upwards (figure A1-1). Consequently, estimation of the actual volume of the substrate segment was required to correct for "excess" water surrounding the sediment core. As shown below, this correction was achieved in two steps. First, the actual volume of the substrate and interstitial water was roughly estimated by a first order model based on the geometry of a conic segment. The difference between this volume and the cylindrical volume of the core segment yielded a corrected total 123 SEGMENT . . . .. . O I . I }' '.-‘ 4" ‘ _ . ’1‘, ‘, .7. . 7... ,:°- no: 0 s .0 I .. I 3, .,._3a '9. - C. . ' Dz. . 't ‘ <0 ‘ ..‘t ,.‘ . . ... ' l . O - \~.-. 1 .- l a --_._ -. 2'- . z-v..: ; - . .4. .J ‘ -........ O - . . .. --_ j. . - . . . 124 PROFILE OF FROZEN SED/MENT CORES Figure A1—1. Schematic profile of frozen sediment cores. 125 weight for the core when subtracted from the total wet weight of the core. The difference between the dry weight of the core and the corrected wet weight provided a volume for the voids. The total volume of the core segment was then calculated by dividing the dry weight of the core segment by the density of quartz sands (2.65 g/ml; Holtz and Kovacs, 1981) and adding the substrate volume to voids volume. P = (Vv)/(Vt) ' when Vv = [Wt - (Vt - Vc)] - Wd and Vt' = (Wd/2.65) + Vv where Vv = estimated volume of voids Vt = cylindrical volume of total core segment VC = conic segment volume estimate Vt' = corrected total volume of core segment Wt = wet weight of cylindrical core segment Wd = dry weight of core segment Considering the deformation of the sediment lattice during freezing, the nature of the relationship between porosity calculated from frozen core segments versus unfrozen cores needs to be ascertained. Because unfrozen cores could not be sectioned, comparisons were 126 made between whole frozen and unfrozen cores of laboratory substrates. Shown in figure A1—2, the relationship between mean porosity for frozen and unfrozen cores is described by the following equation: Pf = 4.702 + 0.823Puf where Pf = porosity of frozen cores Puf = porosity of unfrozen cores with an R2 of 83.0 percent. The ANOVA for the regression model is listed in table Al-l. 127 .4 T '3 1 .- -I 5 U I: J _ J A - d (I) _ . w 50 - - L _ . 0 _ . o - q 5 40 _- I T t N . -—E—— . O _ - cg; _ . x; 30 r — .. ¢ ‘ CL _ . 20 I. I . . . . 1 . . . . I . . . . I . . . . l 20 30 40 50 50 P (unfrozen cores) Figure A1—2. Linear regression through mean porosity values of frozen and unfrozen cores. Error bars for each mean are included. 128 Table Al—l. ANOVA of linear regression model through mean porosity values from frozen and unfrozen laboratory substrates. Source SS df MS F—ratio Signif. Model 130.894 1 130.894 29.229 0.0017 Error 26.869 6 4.478 Total 157.763 7 APPENDIX 2 Laboratory Measurement of Permeability Permeability of the top four centimeters of laboratory and field substrates was measured in a 12- place, low head permeameter (figure A2-l) at 25.0 i 0.50 C. A constant head of 5 cm was maintained by pumping water to the permeameter from the storage reservoir and allowing the excess to drain back to the reservoir. Permeability (K) was calculated using Darcy's law for flow through a porous media (Holtz and Kovacs, 1981): K = (QL)/(hAt) where K = the coefficient of permeability (cm/min.). Q = volume of water (cm3) drained through core over time t L = length of sediment core (cm). h = pressure head (cm). A = cross sectional area of sediment core (cm ). t = time (min.) While still frozen, the sediment/water interface was delineated in the core tube with a commercial 129 130 FHSRWHEAAflETEV? \ o . . a ’ v —- —, . ...- '_-.- -—-..- I»III-LII-----I-I--I—-Iffi--I5 SAMPLE CONTAIAERS \l/ PUMP CON TROL LE R I. I\‘ rm?- r ALSTORAGE RESERVOIR >LE Figure A2-1. Schematic representation of the constant head permeameter. Constant head is maintained With a low velocity pump and overflow stand-pipe set to the desired head. 131 ultrasonic sensor. The core tube, still containing the frozen sample, was then cut with a band saw using a 24 teeth per inch blade. The core tube and sample was then sectioned again 4 cm down from the sediment/water interface with the band saw. The 4 cm segment (still frozen) and sandwiched by two 2.54 cm female slip/slip PVC fittings was then fitted into the permeameter. The frozen segment were then allowed to thaw in the permeameter and come to thermal equilibrium with the permeameter for 12 to 16 hours before measurement of permeability. The core segments were supported in the permeameter on 2.64 cm diameter aluminum screens (18 mesh) and loss of substrate was prevented by a single layer of tissue cut from KimWipes