.-.,..w- j 4-: _,.A4 _‘. ___ :4 4:41..- . - — - ; ~37." A2,..- ._ _. 12:). fl‘ ., __" ‘7}. A . -y-— ~v~ ‘0 “.1.-.- -,‘ __ L..— ‘. '~ - ;-oon— .I-t‘ ' ..- ”2.. -.-... ‘-V?L.—;'«:V A" - - -’ ‘ I 111111 1;..1I 111111111 1111. . '1' '1 11 11111; '11; '1 1111111 111111111 11'; 1"1111111K'1'1I1" 11111111 111111111111111111111'11'1 “. 1111111111111111'1'1 '1,.1"1"-'.11111'."1 ' 11111 11111 «11111111111111;11111111111111 111111" 1111111'1'"1111"1 '1 11111111111111111'11 1 1 11"" 1 111111 11111111111111 11.111111 11111111 1 1" .111 1111111111111 1111111111 11111 11'1" 1' 111111.. 11111 '1'11 1'1111"'11111111'1"" 1'11 11' 1111111111111 1 111 "1 111 111111.11'11'111I'1 11 ,. . 11 ' 'Jx.‘ 11 11.1. 1 1111332111? 5111.? . 1 '1" 119111115 J... I 1 II. 9 '1 '1 , :' ' I 1115.1111(% L 11'J'1111111" 1 1- .11. . . ' '1" 1 ' '11] 1M? 1111' 1111111111 11111.11 11' '11 ”1111111 1111" 1111111111 1 : 11 3:. 1.1... 111151 1‘" 111'11111 1" ' "11'1111"1 '1";" “11.11111 111111111111" '1 ’ 11 111111111.‘.111111111}'111 1111111111111111111111111.1111” 1 11 III 11.1 I' .111 1 111.1 1111 11.1.11. 111111" '1' '1' 1 111111111" 111111 - , v~ . '11. 11.1111111‘11'11111'111'111111 111'111"'1'111'11'111111".11111111‘111;1;11|1 11'1'11;"11' 11111111111111.1111111111111 1 11111111111 111'1'1 111: 11111111] ' 1'131111 1L1 1' 1:11.11 11".. 111:1‘."1::1U"1i':f b} 11" 11.11.11 1 ' 11113-11111 -.:11‘1:‘1:5.1-‘11 |111 1.111111111111111 11111111111111111111111113111111'1111111111111111111111:11111111 1’11‘1" '11 325:1?” 152:1 1 'I NHL 1111.. 1111.111111'1'1"11‘1'3.l 1111111111..” 11 1‘ . 11 11. 11111'11'1111 1'1 "111"" 1'11 )1111111111" 111111111111 1'11" 11.311 '1' 11' '1'1' 1'1 11 ' 1 .111 1'1'1'11'11'111111111' 1'11" ”11'1'11 11"1'11'111'1' 71.1.. . m 1 ‘1 111111'11111 11111“ 111.1111'1 1 '11111.:111111111111111111'11111'111'11111‘1 111” (\1'. 1:77.1111‘. t! 1111 1'1'11‘11'111'1' 1'1 .11 1.1111 1.1111111 ' ' 1'1'1'1'1'1'1'11'111'11'111111 " '1'1'11'11' ' '11" '1111'1"'11""11111111111""11'I11111'111'1"11"1 1'1 '1'111'1115" 1'11: ' '1 . 1 I... '1. 11.11:; I 1"11.g1.f"11 .1 11I'1'111 1111111‘11.1111 1"1 71111;. 112.1. 1"'1.1' 111111 11111'11111111'1111111111111'111 ‘1'1'11'1111411 1.51:; 1'11111 1 ' 1 111.1 11'1'111 11 1111111111111” 11.1111111Wf‘>1'11"111'1 1 .1111. .1... I" 1‘11 1 '11 . 11;" ""333; - 1 .111. ,1 “ .1' _1 I 1 .' 1' 1 1" 1.1 111; ','2' ‘ 1.11.1": ' '1 131." :. "1115111 .111'. .1 =." .. '. :1"I11,111.1'1'1'21':-. 1 ,. 1‘! 31. . I 1.7.3 :3 .. . 1 111111.111...”- ' .I '1‘??- 1111111 1.11 1311, : 111111111 1-1 '1'11'11 1'31? 1. 1 ~ .— ‘ “.1'1'1111'1"".1'52'I'1' 1111 1111 1111 1.111111. '1. 1.. 11} 11111 111111.11, 11.1111 I 1 1.1111 1111, 111 I 11111"" “'1'11' 1"" '1'11'1'" '11 '1" '1 1'1'1'1'1' 11:11:11’" ""1" '1 1 1' 1111'.”‘ '1111'1I11'11'11'1." 1.111 '11111'11 '1 '1' 1' ' '1'1'141 10.111111 11111 11.11" "1"'11111"""1111111111 111111 ""1'11 111111 - 1"" '1" 1 1 1' '1" '1 1" 1 11.11 11.....I'11 11111111111111 .1 111111.11'11I111"111111111.11111.1111111'11‘111 I11. I F111“ 2‘ ' '1 1.1.1: ...t11 .1211 11111.1 ' r11 1"I'-111111I11111111111 ‘1'11.1.1'11111111111.1\111 Ill}nluyuzanllllulwl"will 1/ m w, 'Llnivcmizy This is to certify that the thesis entitled USE OF ARTIFICIAL INSTREAM TROUT SHELTERS BY TROUT IN THE AU SABLE RIVER, MICHIGAN presented by Andrew Joseph Nuhfer has been accepted towards fulfillment of the requirements for Master of Science degreein Fisheries and Wildlife c. t /2a47\——W ( MMrofessor Janua 7, 1980 Date ry 0-7639 79V... i came”. 1 - J OVERDUE FINES: 25¢ per day per mu RETURN!“ LIBRARY MATERIALS: Place in book ntum to move chum frou circulation records USE OF ARTIFICIAL INSTREAM TROUT SHELTERS BY TROUT IN THE AU SABLE RIVER, MICHIGAN By Andrew Joseph Nuhfer A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Fisheries and Wildlife 1979 v" // (11‘ / o / ABSTRACT USE OF ARTIFICIAL INSTREAM TROUT SHELTERS BY TROUT IN THE AU SABLE RIVER, MICHIGAN By Andrew Joseph Nuhfer Abundances of trout beneath 70 man—made coverts in a kilometer of stream were estimated by wetsuit diving and by electrofishing, and physical features of the coverts were measured to determine sources of variation in covert use by trout. Trout were more abundant in shelters having longer margins providing overhead cover parallel to flow, having deeper water adjacent to the device, and having ample interstices for concealment. Current velocity beneath or adjacent to coverts is likely to have influenced covert use, but my rough, indirect measures of velocity were insufficiently sensitive to detect a relationship. There were no significant differences between the types of arti— ficial structures tested: stream-edge log jams, log rafts submerged in open stream, and bundles of tree stumps partially submerged in open stream. More than 50% of the study area's trout that were 150mm or larger were associated with man-made shelters. The percentages of trout using man-made shelters increased as trout size increased. There was weak positive correlation between trout population density in 100m subsections of the study area and the amount of natural and man- made overhead cover per subsection. ACKNOWLEDGEMENTS I would like to thank Dr. Ray J. White, my major professor for his help and advice during this study and during preparation of my thesis. Thanks are also extended to committee members Dr. Niles R. Kevern and Dr. John A. King for guidance and for review of the manuscript. Dr. Stanley J. Zarnoch and Dr. John L. Gill provided valuable advice on statistical principles and procedures. For help with field work, I thank Kurt Fausch, Mary Whalen, Jim Gruber, Chris Bennett, and Rick Staples. William Buc and Gaylord Alexander, both of the Michigan Department of Natural Resources, and Robert Larson of the U.S. Geological Survey in Grayling, all provided valuable data and advice. Thanks are also due to Cecil L. Williamson for his advice on computer programming. Financial support for this study was provided by Project 1169 of the Agricultural Experiment Station of Michigan State University and by contributions from the George Mason, Jackson, West Michigan, Paul Young, and Challenge Chapters of Trout Unlimited, as well as by the Michigan State Council of Trout Unlimited, all were greatly appreciated. A personal contribution of funds by Mrs. Ruth Gruitch of Grayling, Michigan deserves my special thanks. Finally, I am particularly grateful to my family and friends for their moral support throughout this study. ii TABLE OF CONTENTS Page LIST OF TABLES ..................................... . .......... .. V LIST OF FIGURES ................................. .... ............ viii INTRODUCTION .. ..................... . ................. . .......... 1 DESCRIPTION OF THE STUDY AREA .. ...................... . ....... ... 7 Specific Location and Dimensions of the Study Area ......... 9 water quality ......COCOOOOOOOOOOCOO...... OOOOOOOOOOOOOOOOOO 9 DiSCharge000000000cocoo...coo.0.0000000000000000...00000000 12 Bed Materials .. ............................................ 12 Trout Cover . ............................................... 14 Fishes ..................................................... l9 Benthic Invertebrates .... .................................. 22 Instream vegetation o o o o o o o o o o o o o c o o o o o o o o 000000000000000000 22 River use 0.00.000. ooooooooooooooooooooooooooo o ooooooooooooo 22 METHODS .. ....................................................... 25 Population Measurements .................................... 25 Preparation of the Study Area ........................... 25 Snorkel Diving Observations . ............................ 25 ElectrOfishing Procedure 0 o o o o ooooooooooooooooooooooooooo 26 Population Estimates 0 o o o o o o o o c o o o o o o 000000000000000 o o o o o 28 Habitat Measurements o o o o o o o o o o o o o o o o o o oooooooooooooooooooo o 32 Trout Shelter Classification Scheme --------------------- 32 Measurement of Overhead Cover ....... ............. ....... 32 Statistical Analyses . ..... .......... ....................... 33 Analysis Of Effect Of cover Type O O C I O O O O O O C O C C C O O C O O O O O O 35 Snorkel Diving Data Analysis .................. .......... 36 Relationships of Trout Numbers and Biomass to Cover per Station 0 C O O O C O O C O O O O O O O C C O O O O C O O O O O O O 0 O O I O O O O O ..... 36 RESULTS ......COOOOOOOOOOO0........OOOOOOOOOOOOOOO0.0.0.0.0000... 37 Trout Abundance ........................ ..... ..... .......... 37 Brown Trout ........... ..... .. ..... . .............. ....... 40 Brook Trout ............................................. 43 Rainbow Trout ........................................... 43 Comparison of Brook and Brown Trout Abundance ........... 46 iii Page Comparison of Standing Crop in Various Michigan Streams ..... 57 Trout Population, Cover Relationships ....................... 57 Variables Influencing Density of Trout in Individual Trout Shelters................................................... 68 Regression Models with Polynomial and Interaction Factors ... 76 Regression Analysis of Snorkel-diving Counts and Physical Parameters of Shelters..................................... 81 Comparison of Shelter Types...... ..... . ....... .............. 84 DISCUSSION......OOOOOOC....OOOOOOI........OOOOIOOOOOOOCOOOOOOOOOO 89 Shelter Characteristics Correlated with Trout Abundance ..... 89 Linear Regression Models................................. 89 Nonlinear Regression Models.............................. 89 Regression Models Based on Snorkel Diving Counts ......... 92 Comparisons of Shelter Types................................ 95 Log Jams, Rafts and Stump Shelters....................... 95 Trout Density in Bank Shelters........................... 96 Relationships Between Trout Population and Overhead Cover ... 97 Percentage of Trout Beneath Man-made Shelters ............... 99 Population Estimates........................................101 Trout Abundance .......................................... Estimation Procedure.....................................102 Application to Stream Management and Implications for Further Research...................................... ..... 103 CONCLUSIONS .. ............. . ..................... .................105 REFERENCES CITED .00................OOOOOOOOOO......OOOOOOOOOOOOCO106 APPENDIX . ................ .... ........................ ............111 iv LIST OF TABLES Table Page 1. Water chemistry above Stephens Bridge ....................... 11 2. Species of fish captured in the 19703 in the Au Sable River X's indicate fish presence .... ..... ........ ..... .. .......... 20 3. Common benthic organisms found and identified on the Stranahan and Krlight Tracts ......OOOOOOCOOOOOOOOOOO ........ O 23 4. Common aquatic vegetation found and identified on the night Tract ......OOOOOOO..0.........OOOOCOOOOOOOOO...0.00.0 24 5. Population estimates for brook, brown and rainbow trout combined in the l-km study section (all 10 stations) of the Au Sable River, July 17 to 21, 1978 ............. . ........... 38 6. Biomass (kg) estimates for brook, brown, and rainbow trout combined in the 1-km study section (all 10 stations) of the Au sable River, July 17 to 21’ 1978 0............OOOOOOOOO... 39 7. Brown trout population estimates in the 1-km study section (all 10 stations) of the Au Sable River, July 17tx>21, 1978 . 41 8. Brown trout biomass (kg) estimates in 1-km study section (all 10 stations) of the Au Sable River, July 17 to 21, 1978 .. 42 9. Brook trout population estimates in the l-km study section (all 10 stations) of the Au Sable River, July 17th 21, 1978. Estimates by proration from estimate for all species together .................................................... 44 10. Brook trout biomass (kg) estimates in the 1—km study section (all 10 stations) of the Au Sable River, July 17 to 21, 1978 ......0.000.000.0000.............OOOOOOOOOOOOOO0.0.0.... 45 11. Brown and brook trout population estimates by 100-m stations in the Au Sable River, July 17 to 21, 1978. (8-station estimate calculated from data with stations 1 and 3 omitted) . 47 12. Brown and brook trout biomass Ugo estimates by 100-m station in the Au Sable River, July 17 to 21, 1978 .................. 48 Table 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. POpulation estimates for rainbow trout in the l-km study section (all 10 stations) of the Au Sable River, July 17 to 21, 1978. Biomass (kg) estimates shown in parenthesis ...... Brown trout population estimates by lOO-m stations in the Au Sable River, July 17 to 21, 1978. Estimates by pro- ration from 8 stations combined brook and brown trout estimates .........0.0.00.0000......0.0.0.........IOOOOOOOOOO Brook trout population estimates by lOO-m stations in the Au Sable River, July 17 to 21, 1978. Estimates by pro- ration from 8 station combined brook and brown trout es- timates ..........OOOOOOOOOOOOOOOOOO......OOOOOOOOOOOOOOOO... Brown trout biomass (kg) estimates by lOO-m station in the Au Sable River, July 17 to 21, 1978 ......................... Brook trout biomass (kg) estimates by lOO-m station in the Au Sable River, July 17 to 21, 1978 ................. ....... . Standing crops of trout in various Michigan streams ......... Length (m) and area (m2) of overhead cover by station ....... Abundance of brook trout per lOO-m station associated with man-made cover and with other stream area ............ ..... .. Number of brown trout per lOO-m station associated with man-made Cover and With Other Stream area coco-00000000000000 Percent of total brown trout in man—made cover by lOO-m Station ......0.0.9..........OOOOOOOOOOOOOOOO0.0.00.0... ..... Correlatign coefficients (r) and coefficients of determi- nation (r ) for trout population variables (Y), total length, and total area of man-made and natural overhead cover per station in the Au Sable River ............................... Correlation coefficients (r) and coefficients of determi- nation (rz) for trout papulation variables (Y), total length, and total area of man-made overhead cover only per station (n=8) in the Au Sable River ................................. Correlation coefficients (r) and coefficients of determi- nation (rz) for trout population variables (Y), total length, and total area of natural overhead cover only per station (n=8) in the Au Sable River ................................. vi 51 52 54 55 58 59 6O 62 63 65 66 67 Table Page 26. Characteristics of 57 man-made trout shelters and the number of trout beneath them estimated by electrofishing. Shelters 1-7 and 18-23 not shown due to unwadable pools in stations 1 and 3 ........................ . ............... . ............ 70 27. Multiple regression analyses (backward elimination proce- dure) of trout population variables (Y) on linear physical measurements of instream trout shelters (n=57). F for removal from regression = 2.0 .......................... ..... 72 28. Multiple regression analysis (stepwise procedure) of trout population variables (Y) on linear polynomial and inter- action factor variables involving measurements of instream trout shelters (n=57). F for entry into regression = 3.0 ... 78 29. Characteristics of 39 man-made trout shelters and the number of trout 3_150mm sighted beneath them while snorkel diving. Shelters whose identifying tags were removed by vandals are not shown ..... ........ ..... ................................. 82 30. Multiple regression analysis (backward elimination procedure) of the mean number (from three snorkel diving observation periods) of brook and brown trout 3 150mm (Y) beneath 39 man-made trout shelters on various physical parameters ...... 83 31. Estimated biomass (kg) of trout beneath 1 square meter of overhead cover for 3 types of man-made trout shelters ....... 85 32. Estimated number of trout beneath 1 square meter of overhead cover for 3 types of man-made trout shelters ................ 86 A1. Streamflow discharge and flow characteristics for the Main- stream of the Au Sable River at Stephans Bridge (site shown in Figure 2) 0.0.000...0.0....00............IOOOOOOOOOOOOOOOO 112 A2. Estimated number of brook and brown trout by 50-mm size class beneath S7 trout shelters (estimates by proration from elec- trofishing population estimates) in the Au Sable River, July 17 to 21, 1978 ..... . ................. ...... ............ 113 vii LIST OF FIGURES Figure Page 1. The Au Sable River in Crawford County Michigan ............. 8 10. 11. Mainstream of the Au Sable River in Crawford County Michigan, showing the study area and Township sections. The study area lies within sections 3 and 11 of Township 26 North, Range 2 West ..................................... 10 Approximate daily streamflow discharge at Stephans Bridge in m3/s, June through August 1978 ..................... ..... 13 A large rectangular log jam shelter (foreground) and an- other log jam shelter constructed beside a tree angling downstream to form both a deflector and overhead cover (left center). View is downstream. Mainstream of the Au Sable River, June 1979 ..................................... 16 Submerged log raft shelter positioned near midstream. Main- stream of the Au Sable River, June 1979 .................... 16 Stump shelter positioned near midstream. View is downstream. Mainstream of the Au Sable River, June 1979 ................ 18 Bank shelter in station 1. This device provides an overhang up to 4m wide adjacent to a deep pool and also serves to stabilize the bank. View is upstream. Mainstream of the Au Sable River, June 1979 .................................. 18 Comparison of number and biomass (kg) per 100-m station for brook and brown trout combined ......................... 49 Comparison of the number of brook and brown trout by 100~m station ...............0.00.0000...............OOOOOOOOOOOOO 53 Comparison of the biomass (kg) of brook and brown trout _ by100‘mstation ......OOOOOOOOOOOOOI......OOOOOOOOOOOOOOOOO 56 Comparison of percent of brook and brown trout in mandmade cover by 50mm size class ................................... 64 viii _F_iau_re_ 12. 13. Comparison of biomass (kg) of brown trout beneath 1 square meter of overhead cover for 3 shelter types and 5 size Classes Of trout 0.000..........OOOOOOOOOOOOOO......OOOOOOO Comparison of the number of brown trout beneath 1 square meter of overhead cover for 3 shelter types and 5 size Classes Of trout 0.0.0....0.0.00.0.0.........OCOOOOOOOOOOOO 87 88 INTRODUCTION This study was intended to examine cover used by trout in a l-km section of the Mainstream of the Au Sable River which lies within several kilometers in which the Michigan Department of Natural Resources installed instream shelters in 1975 to enhance habitat for trout. It was not known how intensively these structures were used by trout. To my knowledge no study has been made which examines how many trout use individual shelters of the types constructed in the Au Sable River. One of the primary purposes of these shelters was to increase the carrying capacity of the stream for trout, especially large trout, by increasing the amount of overhead hiding and resting cover. The shelters are readily characterized by size, type of construction, surrounding water depth and velocity, composition of streambed material beneath shelters, and position relative to the stream banks. A better understanding of the relationship of these parameters to the density of trout beneath shelters could lead to construction modifications that maximize the carrying capacity for large trout while minimizing con- struction costs. It has been fairly well documented that trout abundance in streams 'is often positively correlated. with the amount of hiding cover. Enk (1977) found that the length of overhead bank cover in Michigan's. Pigeon River accounted for 88% of the variation in July number of trout_: 150mm long and 72% of the variation in July biomass of trout 150-399mm long. Lewis (1969) examined a number of physical parameters influencing trout pOpulations in pools and found that current velocity and total cover were the most important factors. Both coho and cut- throat trout prefer sidepools offering overhanging bank cover as Opposed to those without bank cover (Bustard and Narver 1975b). Bustard and Narver (1975a) found that the cover types used most frequently by coho and age 1+ steelhead at low winter temperatures were logs and upturned tree roots. Knowledge of the association of trout with cover has led fishery managers to construct artificial cover in streams to increase carrying capacity and overwinter survival. The addition of bank-cover—deflectors in Lawrence and Big Roche-a-Cri Creeks in Wis— consin resulted in dramatic increases of both brook trout standing crops and anglers' catch (Hunt 1971; White 1975). The increases in abundance were greatest for the larger size classes. The addition of dams, deflectors and covers in a 450-yard section of Hayes Brook, Prince Edward Island, Canada resulted in a near doubling of the numbers of brook trout age I and older (Saunders and Smith 1962). Increases in trout standing cr0ps following the installation of cover were reported for a mixed rainbow, brook, and brown trout population by Boussu (1954). Stream salmonids are usually found in microhabitats associated with some type of shelter. Variables which have been examined in relation to microhabitat choice by salmonids include, water temperature, velocity, nearness to preferred velocity, turbulence, depth, turbidity, and direction of flow. Other factors are photoperiod, light intensity (incident and reflected), spatial limits, thigmotaxis, substrate type and color, visual reference points, lateral concealment, presence or or absence of competitors, overhead cover, distance to nearest overhead escape cover, size, amount, and periodicity of food items drifting past the salmonids position. The following section is a review of the literature dealing with the effect of some of the above factors on a salmonids choice of microhabitat. Baldes and Vincent (1969) observed that brown trout (average length 21.3 cm) in an experimental flume occupied resting microhabitats within a velocity range of 12.2 to 21.3 cm/sec. Vincent (1969) states that modal water velocity in a resting microhabitat was 21.3 cm/sec for brown and rainbow trout and 15.2 cm/sec for brook trout. Areas with water turbulence, lack of cover, or water velocities less than 9.1 cm/sec or greater than 30.5 cm/sec were not used as resting micro- habitat. When 25-39 cm brown trout were provided a choice of overhead coverts they preferred a range of sub-covert velocities somewhere between 12.5 and 17.5 cm/sec (Gruber 1978). Griffith (1972) found that the focal point velocities occupied by age O-III+ brook trout ranged from 7.6-9.6 cm/sec in sympatric brook and cutthroat trout populations. Average focal point velocities ranged from 8.4-10.9 for allopatric brook trout. Griffith measured average maximum velocities within 0.6m of the centers of activity of trout ages I—III+ ranging from 12.7-24.1 and 15.7-25.7 cm/sec for sympatric brook and cutthroat trout and allopatric brook trout populations respectively. Trout minimize energy expenditures by positioning themselves in microhabitats of relatively low velocity adjacent to faster which carrying more drifting food items per unit of time thereby minimizing the amount of foraging time spent in swift water. Fausch (1978) found that in a sympatric brook and brown trout population, resting brook occupied resting microhabitats with mean focal point velocities near 20 cm/sec with mean maximum water velocities at 0.6m from the focal point near 36 cm/sec. For feeding brook trout, on the other hand, both focal point velocities and velo- cities at 0.6m were slightly higher, but the velocity difference between the focal point and maximum adjacent velocity was essentially the same as that for resting brook trout. The frictional force exerted on passing water by instream trout shelters can create areas of reduced current beneath the shelter, while faster current sweeps along the edge of the device. Light is also an important activity regulating stimulus. Investi- gations of the activity of brook, brown, and rainbow trout show that all three species are photonegative (Baldes and Vincent 1969; Butler and Hawthorne 1968; Gibson and Keenleyside 1966; Gibson and Power 1975; Gruber 1978; Kwain and MacCrimmon 1969). Overhead cover provides hiding areas of low light intensity. Stewart (1970) reported that overhead cover use increased with increased structure size, decreased structure height and decreased percentages of holes in the overhead cover. The response was strongly related to the light intensity under the structures. DeVore (1975) reported that adult brown trout preferred overhead cover which was low (10cm) rather than high (15 or 20cm)' in the water column. He concluded that the response was related to the close visual proximity of the cover to the stream bed but it may have been the result of de- creased light intensity beneath low coverts. Gruber (1978) found that brown trout most often occupied coverts offering the greatest darkness within the range of 0.0100-S.0000 ft-c at current velocities ranging from 0-149mm/sec. At current velocities within the range of 150-199mm/ sec, trout randomly selected coverts with different light intensities. Bassett (1978) demonstrated that brown trout also respond to reflected light, preferring overhead cover with a dark bed beneath it. Gibson and Power (1975) found that more brook trout were found in a shaded portion of a shallow tank (24-29cm) than in unshaded portions. Con- versely, in a deep tank (43-50cm) more trout occupied unshaded areas than shaded ones. They suggested that a water depth of 50cm provided sufficient cover for trout 8-27cm. However, trout usually do not find shade as attractive as overhead cover in contact with the water (R. J. White pers. comm.). More trout 3 152mm have been found in deep water than in shallower water beneath undercut banks and overhanging vegetation (Wesche 1976). Larger fish are typically found in deeper faster water than smaller individuals (Chapman and Bjornn 1969; Everest and Chapman 1972). To my knowledge no one has examined preference for overhead shelters in different water depths while controlling for light intensity, water velocity, and other behavior directive stimuli which may change with increased depth. Tactile features of overhead cover also influence trout behavior. DeVore (1975) found that brown trout preferred overhead cover with clear plastic streamers beneath them over coverts without streamers. Most stream-dwelling salmonids are strongly territorial. As fish grow larger the size of their territory increases and its physical characteristics change (Allen 1969). The size of each territory is also influenced by such factors as current velocity, bottom irregularities, or other forms of lateral concealment (Allen 1969; Keenleyside 1962; Basset 1978). Lateral concealment beneath overhead cover permits the establishment of smaller territories and reduces agonistic behavior by visually isolating trout from each other. General objectives of this study are: (1) To define the relationship of subshelter trout density in the Au Sable River to various physical and hydrological parameters (2) To compare trout density beneath three types of artificial shelters. (3) To determine what percentage of trout in lOO-m stations are found beneath man-made shelters. Specific objectives of this study were: (1) To determine how much of the variation in subshelter trout density is accounted for by shelter size, maximum water depth adjacent to the shelter, and subshelter water velocity (as indicated by stream- bed material beneath shelters) and to define the relationship of these factors to trout density. (2) To test for differences in subshelter trout density among three shelter types. Secondary objectives of this study were: (1) To obtain trout population estimates for a 1-km section of the Au Sable River. (2) To determine if trout abundance/lOO-m station is correlated with the amount of permanent man-made, natural, or total overhead cover/lOO—m station. DESCRIPTION OF THE STUDY AREA The Main Branch of the Au Sable River lies in Otsego, Crawford, Oscoda, Alcona, and Iosco counties in Michigan's lower peninsula. The stream arises north of Frederic and flows south to Grayling, then generally eastward to its confluence with Lake Huron at Oscoda. The East Branch of the Au Sable River joins the Main Branch at Grayling. The study area (Figure 1) consisted of one kilometer of the Mainstream, lying about 21.5 km downstream from Grayling in the Knight Tract which is owned by Trout Unlimited and designated as a research area. The study area is also within a 14.5-km section of stream on which there are sportfishing regulations more restrictive than those on most other Michigan trout waters. The area of the drainage basin above the study area is 567.2 km2 (Hendrickson and Doonan 1972). Riparian woody vegetation in the study area consist primarily of spruce, balsam fir, northern white cedar, speckled alder, paper birch, and pine, as well as some hardwood trees (Hendrickson and Doonan 1974; Schmidt and Rusz 1974). The river basin in this area is characterized by sandy soils and glacial deposits. The permeability of these deposits causes a high percentage of precipitation to recharge the groundwater rather than to run off. The strong contribution of ground water to stream flow results in rather stable discharge and serves to stabilize and reduce « x 1 :23; .cmwacowz auczoo 2.5.3:.Hfl 38320 5 .832 38$ 3. 2: .5 85w: M s ”....fllllhu o , m o M... macaw oz.w»¢cu \" <2: aoapw gazrggu summer temperatures in the upper reaches of the Au Sable, making these waters thermally suitable for trout. Mean annual precipitation at Grayling is 76.2cm. The minimal infiltration rates is 30.48cm per hour (Bent 1970). Specific Location and Dimensions of the Study Area The study area lies in Crawford County, 14.5km east of the town of Grayling, Michigan. The study area is within Sections 3 and 11 of Township 26 North, Range 2 West, and consists of 100-m, as measured upstream from the south edge of the northeast quarter of Section 11 (Figure 2). The study area is 1.6km upstream kilometers above Wakeley Bridge and 21.5 stream kilometers below Grayling. Two dirt roads leading south from Wakely Bridge Road provide access to the downstream end of the study section. Dirt roads leading to the Thunderbird Club provide access to the central and upper portions of the study area. The mean stream width and mean maximum depth in the study area are 29m and 78.7cm respectively (Schmidt and Rusz 1974). The mean streambed slope in the study area is 1.39m/km. Barker Creek enters the study section from the northest about 300-m upstream from the lower edge of the study area. Water Quality Stephans Bridge and Wakely Bridge are 4.6km upstream and 1.6km downstream respectively from the study area. Hendrickson and Doonan conducted chemical analyses at a site above Stephans Bridge, NW 1/4 sec. 5 T. 26N., R. 2W (Figure 2) in 1972 (Table 1). 10 .uuoz N owes: .guuoz ow awzmcao 9 p ... 9: mo 2 new m 3338 :23, 3.3 a»; ”v u a: .N unsung u vuouamuo 5" ass: 63mm :< 0.: .«o 360.3 an! was .ucouuuoa eugenics vac nous arson oau wearers .cmmqgoqz .Aucso 4 nw . m «a . m~ . ~— 22::. I] u/ lullJl- 11’s}! _ m. o . .3 .v // . ' . .PI/ ‘. ..‘Ik _. x / . u o I I. a. . x u. . .- ¥T. I . ~ Inn-Ill “A , ..\ \ m _ J1 st. 1: ;/ gr I 77 . c ~ MW n . c m e a . \ .u . ft x A}. 1 S. s w" . ss \ . - - . .. ‘3: 3:: \ use: at... ‘ 11 Table 1. Water chemistry above Stephans Bridge.* Item Measured mg/l Calcium (Ca) 42.0 Magnesium (Mg) 7.6 Sodium (Na) 2.3 Potassium (K) 0.6 Bicarbonate (HCOB) 158.0 Carbonate (003) 0.0 Sulfate (804) 7.6 Chloride (Cl) 4.0 Fluoride (f) 0.2 Nitrate (N03) 0.2 Dissolved solids: Residue on evaporation at 180°C 150.0 Hardness, as CaCO3 140.0 Noncarbonate 6.0 Specific conductance 265.0 (micromhos at 25° C) pH 7.4 * Hendrickson and Doonan (1972). 12 Dissolved oxygen in most of the river does not drop below 6 mg/l at any time (Hendrickson and Doonan 1974). (The Michigan Water Re- sources Commission water quality standards adopted in 1968, set the minimum dissolved oxygen standard at 6 mg/l for trout). Discharge Owing to the morphometry and soils of the Au Sable River basin, streamflow is very stable. The high permeability of the basin's glacial till causes most water to be absorbed into the ground and released slowly to the channel. At Stephans Bridge (about 4.6km upstream from the study area) mean annual discharge is 5.27m3/s with 10-percent duration discharge of 6.80m%@ and 90-percent duration discharge of 4.19m3/s. The ratio of 10-percent to 90—percent duration discharge is 1.62. Additional discharge data is shown in Table A1. Approximate daily streamflow discharges at Stephans Bridge for a period of time encompassing the period of data collection are plotted in Figure 3. Bed Materials The bed materialsixxmost of the study are sand and gravel. Gravel predominates in riffle areas and provides excellent spawning habitat and substrate for aquatic invertebrates. Silt and muck predominate in shallow, low-velocity areas near the bank. Extensive silt beds are present beneath and downstream from many instream trout shelters. Most of the silt beds lying downstream from these shelters have been somewhat stabilized by rooted aquatic macrophytes. Some patches of clay are present in the upstream portion of the study area. 13 um=w=< I ~n sass cu A.aaou .nuom .wcqfizouu .mUmD combo; uuonom scum muonv .wsad unawa< :waousu wean .m\na ca oquun nauseoum um omuanomuu ao~uaaouua hadav oualwxoune< .n ousmqm All .522 33m 1.7 cu 3 ON mesh 11v o— n.c o.— c.¢ s/Em u; afizuqosra not;msazas 14 Trout Cover Cover for trout is provided by aquatic vegetation, water depth (pools), stream improvement structures, natural log jams and a few undercut banks. The most cover is provided by 70 man-made instream trout shelters. Four major types of man-made shelters were found in the study section: (1) log jams (Figure 4), (2) sunken log rafts (Figure 5), (3) stump shelters (Figure 6), and (4) overhanging bank shelters (Figure 7). Log jam shelters were built using both streamside and instream logs. In some cases natural log jams were anchored to the stream bottom to prevent them from washing away during high water. In other cases, natural log jams were widened and extended to form larger expanses of overhead cover. Streamside trees angling downstream across the current were used to deflect water against and beneath artificial log jams anchored at their downstream ends. The irregular shapes of the building materials prevented most of the log jams from presenting an artificial aspect when viewed from above and also created large amounts of lateral concealment in subshelter areas. The average amount of overhead cover available for use by trout was 8.28m2 for the 45 log jam shelters in the study section. Sunken log raft shelters were usually solid rectangular structures constructed with natural logs and anchored beneath the water surface 4-10 inches off the stream bottom parallel to the current. Some logs were anchored on the bottom beneath the structures to provide areas of reduced current velocity and to create lateral concealment, as well 15 Figure 4. A large rectangular log jam shelter (foreground) and another log jam shelter constructed beside a tree angling down- stream to form both a deflector and overhead cover (left center). View is downstream. Mainstream of the Au Sable River, June 1979. Figure 5. Submerged log raft shelter positioned near midstream. Main- stream of the Au Sable River, June 1979. Figure 4. Figure 5. 17 Figure 6. Stump shelter positioned near midstream. View is down- stream. Mainstream of the Au Sable River, June 1979. Figure 7. Bank shelter in station 1. This device provides an over- hang up to 4m wide adjacent to a deep pool and also serves to stabilize the bank. View is upstream. Mainstream of the Au Sable River, June 1979. 18 Figure 6. Figure 7. 19 as visual and thigmotactic reference points. Most raft shelters were places in fairly deep water in midstream to allow canoes to pass over them unhindered. The mean amount of overhead cover beneath the 9 rafts in the study section usable by trout was 3.7m2. Stump shelters were constructed by binding large stumps into a roughly circular shape and anchoring them off the bottom. The root structure of the stumps provided lateral concealment and caused the current to scour bowl-shaped depressions beneath the shelters. The mean amount of overhead cover beneath the 3 stump shelters in the study section potentially available to trout was 2.08m2. Two large bank shelters, 61 and 79 meters in length, were within the study section. These shelters were constructed on the outside of meander bends in stations 1 and 3. Their solid surfaces were from 1.5-4m in width and provided large areas of overhead cover. These two structures prevent the stream from eroding the bank and have caused it to scour out long, deep pools which are 1.7m deep in some areas . Fishes The fish population in the study area consists primarily of brown, brook, and rainbow trout. These species are not native to the system but were introduced between 1884 and 1891 (Richards 1973). Richards collected fish with seines from most sections of the Au Sable watershed in 1972. Table 2 is a summary of the species of fish taken in 1972 by Richards at 3 stations near the study area. 20 asmaaowwm can camacowwm x mamoaououmn mamouuoz “spasm omonxomam Amuoov N covououon mamouuoz Hosanm nanoxomam Aaaasouazv x x mauasuoo manOuuoz womanm soaaoo x x commumnoam msumcwou mauuoo dfimaaom madam x x x x saunas Susana nausea aaaasom vuauuoz Ammaammv x x Eamomuwawaao adamomoum :mauouwns venom x x x x mammacwq muuauu osamm uaouu saoum x x compumnofim wumcvuwmm oaHMm uaouu sopaamm Aaaanuuazv x x x x mHHmoaucow manaao>amm usouu xooum moan hvaum owvaum oqunm mcmcmoum omkum mamnaoum mmwommm cast aoamxmz scams as 50.5 o>onm as He.a ouaumdu mo soHumooq .oocmmoum swam oumoacfi an £94m 63mm «E on”. a.“ mega on”. ma wouaummo swam mo mmwomam .N 0.3m“. 21 .wmmH .zaah .soauoom absum osu ca wcfiswwmouuomao an vousuamu as x mammcsfiA msfiosa Noam oxfim suonuuoz mavmwnfimmm x maufinoouoma mfiaomua HHmeDHm Amowmoawmmmv x mwuumumwu mouwamoama< mmmn xoom Awsmauuaxv x mamumcoosa mamaso xomnoaxoaum xooum Aowomwumqv x x x x Hcomuoeaou msfioumoumo umxonm ouagz Aaaanouazv x msumasomaouum msafiuoamm nsno xoouo Aaamaummv x x maasumuud mzsusofiawnm womb mmosxomflm onummsfimmm x wmfimaoum.moamnmwafim Boaaws vmmnumm mono hbdum mwwwum owkum mcmnaoum owvfium mcmnaoum mofiooam aH«« amaoxmz scams ex no.5 spasm as Ho.a musumwo mo coaumooq eases .N 33 22 Benthic Invertebrates Schmidt and Rusz's (1974) examinations of the benthos in the Stranahan tract (1.21km upstream from Stephans Bridge) and the Knight Tract revealed a diverse and abundant insect population. Table 3 lists the more common organisms they found. Instream Vegetation Abundant aquatic vegetation in the study area provides cover for trout as well as substrate and food for aquatic invertebrates. Vege- tation proliferates in the shallow, silted areas found in slowly flowing water downstream from many trout shelters. Table 4 lists common types of aquatic vegetation found on the Knight Tract. River Use The Au Sable is used primarily for recreation. The section from Grayling to Wakeley Bridge is intensively used by both fishermen and canoers. Numerous cabins and homes are found along this stream reach. This wide, shallow section of river provides easy wading and fly casting. Fishing pressure is intense, especially during insect hatches. There are special angling regulations on the 14.3km section be- tween Burton's Landing and Wakeley Bridge. At the time of this study, fishing gear was restricted to artificial flies. Three trout could be taken a day. Minimum size limits were 8 inches for brook trout and 12 inches for all other trout. There was no closed season; trout could be caught but not kept from November 1 through April 28. 23 Table 3. Common benthic organisms found and identified on the Stranahan and Knight Tracts.a Ephemeroptera (mayflies) Trichorythodes s2, Baet is _p. Ephemerella invaria Ephemerella lata Pseudocloeonigp. Trichoptera (caddisflies) Brachycentrus s2, Glossosoma._p. Heliocopsyche sp. Protoptila'_p, Neophylax‘_p, Lepidostoma _p, fiydropsyche'gp. Coleoptera (beetles) Elmidae Odonata (dragonflies, damselflies) Agrionidae Diptera (flies, midges) Simulidae (black flies) Chironomidae (midges) Hydrobaeninae Tendipedinae Tabanidae (deerflies) Hemiptera (true bugs) Gerris sp. (water striders) Megaloptera Corydalis sp. (Hellgramites) Other ghyga'sp. (snail) Sphaerium sp. (fingernail clam) (scud) Ascellus sp. (sowbugs) Gammarus sp. a From Schmidt and Rusz (1974). 24 Table 4. Common aquatic vegetation found and identified on the Knight Tract.a Common Name Scientific Name Submerged Forms Threadleaf Pondweed Whitestem Pondweed Crispedleaf Pondweed Water Buttercup Waterweed Water Milfoil Attached Algae Stonewart Freshwater Red Algae Green Algae Potamogeton filiformis Potamogetonlpraelongus Potamogeton crispus Ranunculus sp. Elodea sp. gyriophyllum sp. Chara sp. Batrachospermum sp. Cladophora sp. Emergent and Semi-aquatic forms Bulrush Spike Rush Sedge Rush Arrow Arum Scirpus sp. Eleocharis sp. Carex sp. Juncus sp. Peltandra virginica aFrom Schmidt and Rusz (1974). METHODS Population Measurements Preparation of the Study Area The study area was divided into ten lOO—m stations which were marked by tying plastic tape to streamside trees at station boundaries. Stations were measured by trailing a 30m plastic clothesline marked in meters in the thalweg. Therefore, the stations are measures of thalweg distance rather than streambank distance. Station markers were numbered 1 (downstream) through 10 (upstream). Small squares of plastic tape numbered 1 through 70 were nailed on the downstream end of each artificial instream trout shelter in the ten stations. Snorkel Diving Observations Wearing a black wetsuit consisting of hood, boots, face mask, snorkel tube and gloves, I slowly approached each shelter while sub— merged from downstream, looked beneath it and observed the number of visible trout. Trout were categorized as less than and greater than 150mm in total length. The number, size class, and species of trout observed under each shelter was told to a notetaker who waded about 30m downstream. An Ikelite C-Lite II underwater light was used to illuminate beneath objects so I could see trout not otherwise visible. Counts were made June 28, 29, and 30, 1978, from about 0830 h to 1200 h and from 1400b to 1630b. Counts were made 3 times for each of 39 shelters during this time. Only 39 of the 70 shelters were observed 25 26 at this time because vandals had removed many of the identifying tags from the shelters. Underwater visibility was measured at the beginning of each obser— vation period and about every half hour afterward. Visibility was determined by holding the yellow tip of a black snorkel tube underwater in midstream and measuring the maximum distance away which it could be seen by a submerged diver. Sky conditions for each shelter observation were classified as overcast, clear, hazy, or with intermittent clouds. These broad categories were chosen because they were believed to influence fish behavior and possible cover-seeking activity. ElectrofishingiProcedure Trout populations were inventoried using mark-and-recapture elec- trofishing. The procedure was designed to make p0pulation estimates for individual trout shelters, natural cover, man-made cover, and for individual lOO-m stream sections. The electrofishing unit consisted of a 2.1m plastic boat carrying a 250—v, 1.75-kw generator. Three spring-loaded retracting reels mounted on the bow were connected to the generator's anode and to fiber- glass handled capture electrodes. The reels permitted the crew members to be separated by as much as 16 meters. The cathode consisted of brass window screening under a Styrofoam float trailed behind the boat. A 3-man electrofishing crew moved upstream toward each trout shelter, one man towing the boat and each carrying an anode and hand net. The anodes were held out of the water until the crew had surrounded the shelter so that fish holding positions in midstream would not be 27 driven into the shelter by the electrical field (but they could flee there when frightened by approaching crew). The crew then thrust their electrodes under the cover simultaneously, creating an electrical field around the shelter. Trout drawn to the electrodes were netted and transferred to a holding tank on the boat. The electrodes were probed into all accessible interstices of the shelters and withdrawn, pulling trout out where they could be netted. When no more fish could be found, the electrodes were removed from the water. A piece of plastic tape with a number corresponding to the number of the shelter was dropped into the boat's holding tank with that group of fish. A separatory net was then placed in the tank. The crew proceeded upstream to the next shelter with electrodes out of the water and repeated the above proce- dure. After about 5 shelters were inventoried, the holding tank was transferred to the stream for processing by a Z-man team following the electrofishing crew. This procedure was continued until all individual shelters in a lOO-m station were sampled. Because the stream was about 29m wide, we sampled shelters on one side until we reached the upstream station marker. The generator was shut off, and the crew moved to the downstream end of a station, walking on the side of the river already sampled. The crew then fished the shelters on the other side of the river until the upstream.marker was reached. After all mandmade shelters had been sampled individually, the crew once again moved to the downstream end of the station and began shocking all areas of the stream, holding their electrodes in the water at all times. Beds of aquatic plants, log jams, and riffle areas were all sampled. Man-made shelters were all reshocked to capture any fish 28 driven into them from elsewhere. All fish captured on this sweep were kept separate in the notes. A Z-man processing crew anesthetized captured trout with tricaine methane sulfonate (MS-222), weighed them to the nearest gram, and measured them to the nearest millimeter. Trout captured on the first run were marked by clipping the lower tip of the caudal fin. The length and weight data were recorded and cataloged according to species, shelter number, and station number. Captured trout were held in a live- box until at least one lOO-m station was completely electrofished. Fish were then released at the downstream end of the station from which they had been captured so that the disoriented trout (which tend to swim upstream when released) could better assume their former distri- bution. The second or recapture run was done by the same electrofishing procedure described above. During the process, fish were examined for caudal fin clips. The upper tip of the caudal fin was clipped on this run so that these fish would not be counted twice if they moved upstream overnight from the point where the crew stopped fishing. Unmarked fish were weighed and measured. Fish bearing fin clips from the first run were only measured, not weighed again. The shelter and station number where each fish was captured was recorded. The first marking run was conducted July 17-18; the second or recapture run on July 20-21, 1978. Population Estimates Population estimates were calculated by the Schaefer modification of the Peterson Formula (Ricker 1975). 29 m + 1 (r + u +-1;] _ 1 (r + 1) .1 Where P a estimated population m = number of fish marked on first run r = number of marked fish recaptured during the second run, and u = number of unmarked fish captured on second run The standard error of the estimate is calculated according to the following formula from Ricker (1975). l P (u) (r+u+1) (r+2) SE = I calculated 95% confidence limits for all estimates. If the calculated lower limit was less than the sum of the marked and unmarked fish, the sum of marked and unmarked fish was recorded as the lower limit. Separate population estimates and 95% confidence limits were calculated for fish of each 50-mm size class from 50mm to 350mm. Fish from 350-500mm were grouped for population estimates because both the total number and the number of recaptured fish in individual 50-mm size classes above 350mm were small. The number of fish in the three individual 50—mm size classes over 350mm were then calculated by pro- rating the total estimate for the 150-mm size class according to the procedure described below for species and station estimates. Papulation estimates were made for the 1-km study section (all 10 stations). Estimates were also calculated for each size class for stations 2,4,5,6,7,8,9, and 10 combined and used as the basis for proration into individual trout shelters. Stations 1 and 3 were 30 excluded from this section of the analysis as they were too deep and swift for effective electrofishing. Estimates of the number of brown trout in each size class were obtained by multiplying the ratio of m + u brown trout to m + u brook and brown trout combined for a size class by the population estimate for that size class. This same ratio was multiplied by the upper and lower 95% confidence limits of each combined-species estimate of each size class to obtain confidence limits for brown trout size class estimates. These brown trout estimates were then partitioned into individual trout shelters according to the ratio of m + u fish for individual stations to m + u fish summed over all 8 stations. The same procedure was used to estimate the numbers of brown trout found in man-made shelters and in natural cover. Brook trout estimates were obtained by partitioning the combined brook and brown trout estimates for 8 stations, into individual stations on the basis of m + u as described above. Individual station estimates were prorated from combined brook and brown trout estimates for 8 stations rather than brook trout estimates for 8 stations to minimize the rounding errors that would have occurred, owing to the small number of brook trout captured I used the subdividing methods described above for a number of reasons. Cooper (1952) noted that, since large fish are captured more easily than small ones, it is necessary to make separate estimates for different sizes. He noted further that the total estimate can be subdivided into numbers of fish of each species if each species is represented in the total population in direct proportion to its 31 representation in all of the sampling. A similar breakdown may be made for different portions of the stream if capture efficiencies for dif- ferent parts of the stream are similar. Cooper states that subdividing the total estimate is believed to be more accurate than estimating the numbers of fish of each size and species in small portions of the stream separately and combining all the individual estimates for the total population. Brown trout biomass estimates for stations and trout shelters were computed as the product of the number of fish and the average weight of brown trout in each size class for the entire study section. Brook trout biomass estimates were computed using the average weight of brook trout in each size class for all stations. I obtained 95% confidence intervals for trout biomass by multi— plying the upper and lower pOpulation limits for each species, size, and location category by the average weight of the corresponding species and size category. Rainbow trout abundance estimates were caculated in the same manner described above for brook trout. Habitat Measurements Trout Shelter Classification Scheme Shelter types were classified as treatments. These coverts were categorized as log jams, sunken log rafts, stump coverts, or overhanging bank coverts. After stations 1 and 3 were removed from data analysis only 1 overhanging bank cover remained. This bank cover was believed to be functionally equivalent to a log jam and was placed in this category, leaving only 3 shelter types or treatments for analysis. Measurement of Overhead Cover High rents and other expenses coupled with a paucity of funds at the time of data collection prevented me from making detailed maps and measurements of habitat. Therefore the kinds of habitat measurements I made were based on literature and on one week of snorkel diving observations. Based on preliminary observations of the minimum dimensions of cover used by trout in the Au Sable River, the area and length of overhead cover in contact with the water, 10cm or more above the substrate and 10cm or more in width was measured for each trout shelter. Similarly, permanent natural overhead coverts such as logs and undercut banks were measured and cataloged according to station and location relative to man-made trout shelters. A diver with wetsuit and meter stick looked under each cover and carefully measured the area of overhead cover meeting the minimal criteria described above. A notetaker recorded the measurements under the appropriate shelter 32 33 code number. Lengthy overhead cover was measured with a steel tape. The diver made visual estimates of the percentages of subcovert rubble, gravel, sand and silt. Substrate size classifications were based on a table by Platts (1976), who defined substrate types according to particle diameter, where rubble is 76.1—304.7mm, gravel 4.7-76.0mm, and sand is less than 4.7mm in diameter. I defined silt as any fine organic matter. The maximum water depth found immediately adjacent to the trout shelters, the shelter type, and station was also recorded. Substrate type was recorded to obtain rough estimates of the subcovert water velocities. The large size and complexity of the trout shelters created a mosaic of current velocities, turbulence and direction beneath them and precluded measurement of any one representative or series of representative current velocities. The type of sediment present beneath the covert is an expression of the integrated effects of the mosaic of velocities. (Particles of 0-5mm diameter require a current velocity of around 20cm/sec to be eroded. Larger particles require progressively higher current velocities to be picked up and transported downstream (Morisawa 1968)). Light organic silts are much more readily eroded than inorganic particles of the same size and are found only in areas of very low current velocity. Statistical Analyses Multiple linear regression analysis was used to determine how much of the variation in trout numbers and biomass beneath individual trout shelters was accounted for by water depth, size of shelter, substrate type, and shelter type. All calculations were made with Michigan State University's CDC 6500 computer. I used the backward 34 elimination procedure to reduce the number of independent variables in the equation to those which provided the best linear regression fit. Maximum water depth proximal to the shelter, length of overhead cover, area of overhead cover, percentage of subcovert gravel and rubble, percentage of subcovert sand and percentage of subcovert silt were initially entered into the regression equation. Electrofishing esti- mates of the numbers or biomass of trout in selected Species size classes associated with individual trout shelters were used as the dependent variables. Dummy variables to account for the effect of treatments (shelter type) were also created and made available for entry into the regression equations. Residuals were plotted and examined to determine if error components were independent with a mean of zero and if they had the same variance throughout the range of Y values. To obtain better regression fits I elected to enter a number of interaction and polynonial terms in addition to the simple independent variables listed above. I used estimated numbers or biomass of trout in selected species size classes as the independent variables. These regression analyses were made using both the backward elimination method and the stepwise method. In the stepwise method independent variables were entered one-by-one in a series of regression steps if they had a F ratio greater than 3.0. Variables already in the equation were removed on subsequent steps if they had an F ratio of less than 2.8. Variables removed from the equation could be re—entered at later steps. Dummy variables were created and made available for entry into the regression equations to account for possible treatment effects. Re- siduals were plotted and examined for each regression equation. 35 Analysis of Effect of Cover Type After examining the regression equations for various species/size classes, I chose 6 species size classes which seemed most likely to provide statistically valid comparisons between shelter types. Dif- ferences among shelter types in biomass (grams) of brown trout 3 100mm in total length per shelter, biomass of brook and brown trout_3 150mm per shelter, and biomass of brown trout greater than or equal to 150, 200, 250, and 300 mm per shelter were tested using analysis of co- variance. The three independent Variables, maximum water depth proximal to the shelter x total length of overhead cover, maximum water depth proximal to each shelter and maximum water depth proximal to each shelter squared were chosen as covariates. Dummy variables were created to account for the effect of shelter type (treatments). Factors accounting for interactions between treatments and covariates were also created. Three separate regressions were calculated for each dependent variable. One regression equation contained only covariates, one contained covariates and dummy variables, and one contained co- variates, dummy variables, and interaction factors. These regressions were first used to test for interaction or the lack of homogeneity of slopes. Since the F test for interaction was not significant (o = 0.05) for any of the dependent variables I then tested the hy- pothesis that all treatment effects were equal to each other and zero. There wererm>significant treatment effects at a = 0.05 so analysis ceased at this point. 36 Snorkel Diving Data Analysis Mlll'iILPle regression analysis was used to determine which physical parameters provided the best regression fit for the mean number of brook and brown trout (from 3 snorkel diving observation periods) sighted beneath individual trout shelters on various physical para- meters. I used the backward elimination procedure and included inter- action and polynomial factors among the independent variables. Dummy variables were available for entry into the regression equation to account for possible treatment effects (shelter types). Equations containing 2 and 3 predictor variables were derived. Residuals were plotted for the 2 variables "best fit" regression equations and examined for randomness. Relationship of Trout Numbers and Biomass to Covergper Station The relationship between trout abundance (numbers or biomass) per station and overhead cover per station was examined by both one way and multiple regression analysis. The relationship between trout abundance/100-m station and amount of overhead cover/lOO-m station was examined for natural and man-made overhead cover indi-~ vidually and in combination for stations 2, 4, 5, 6, 7, 8, 9, and 10. RESULTS Trout Abundance The Au Sable was difficult to electrofish owing to its large size. However, the efficiency of recapture (60%) for trout 200mm or more in total length provided for precise p0pulation and biomass estimates. Recapture efficiency was 31% for trout 100-199mm and 8% for trout 50-99mm. Brook, brown, and rainbow brout abundance and 95% confidence limits for all 10 stations combined are presented in terms of numbers (Table 5) and biomass (Table 6). Trout_: 150mm comprise 25% of total trout numbers and 82 percent of total trout biomass. Trout 3 300mm comprise 1.9% of total trout numbers and 22% of total biomass. These estimates for all 10 stations combined are undoubtedly low since two of the lOO-m sections contained long deep pools which could not be electrofished effectively. Snorkel-diving observations of these deep pools showed that many large trout were present but very few were captured during electrofishing. Rainbow trout were observed frequently while diving but only 12 were captured in all 10 stations on the combined electrofishing runs. This indicated that rainbow trout were able to avoid the electrical field. To compensate for this electrofishing inefficiency, estimates were calculated for brook and brown trout combined in the 8 stations which did not contain unwadable pools. These data are shown for individual lOO-m stations and 50mm size classes in terms of numbers 37 38 Table 5. P0pulation estimates for brook, brown, and rainbow trout combined in the l-km study section (all 10 stations) of the Au Sable River, July 17 to 21, 1978. Total length Lower* Point Upper** size class (mm) Estimate Estimate Estimate 0-99 3482 4910 6338 100-149 331 530 729 150-199 997 1122 1247 200-249 278 315 352 250-299 208 237 266 300-349 98 112 130 350-399 20 22 28 400-449 2 2 3 450-499 2 2 3 500-549 1 1 1 Total 5419 7253 9097 * Lower limit of 95% confidence interval ** Upper limit of 95% confidence interval 39 Table 6. Biomass (kg) estimates for brook, brown, and rainbow trout combined in the l—km study section (all 10 stations) of the Au Sable River, July 17 to 21, 1978. Total length size class (mm) Lower* Point Upper** 0-99 20.72 29.23 37.71 100-149 9.12 14.60 20.08 150-199 52.99 59.63 66.28 200-249 31.82 36.05 40.29 250-299 42.05 47.91 53.77 300-349 32.27 36.88 42.81 350-399 9.80 10.78 13.72 400—449 1.49 1.49 2.23 450-499 2.05 2.05 3.08 500-549 1.55 1.55 1.55 Total 203.86 240.17 281.52 * Lower limit of 95% confidence interval ** Upper limit of 95% confidence interval 40 (Table 11) and biomass (Table 12). Station 2 which lay between the two stations with long deep pools had the greatest numbers and biomass of 300-349mm trout. Numbers and biomass of trout per station are presented graphically in Figure 8. The trend in numbers of trout did not closely parallel biomass trends, as trout were distributed differently among size classes in different stations. Brown Trout Brown trout abundance and 95% confidence limits for all ten stations are shown by filmmzsize class in terms of numbers (Table 7) and biomass (Table 8). These estimates were obtained by proration from the brook brown, and rainbow estimates for all 10 stations. Brown trout comprises 94% of the numbers and 95% of the biomass of all trout less than 300mm. Brown trout greater than 300mm made up 97.1% of the numbers and 98.5% of the biomass of all trout in this size group. Brown trout abundance and 95% confidence limits for 8 individual stations by SO-mm size class have been calculated in terms of numbers (Table 14) and biomass (Table 16). These estimates were obtained by proration from combined brook and brown trout estimates for the 8 stations without deep, unshockable pools. Stations 4 and 10 held the least brown trout with 401 and 410 respectively. The greatest numbers of brown trout were found in station 2 with 1102 and station 7 with 1224. Biomass of brown trout decreased from 29.81kg-21.36kg from station 2—5. Biomass was greatest in station 6 with 31.23kg and lowest in station 10 with 18.89kg 41 Table 7. Brown trout p0pulation estimates in the 1-km study section (all 10 stations) of the Au Sable River, July 17 to 21, 1978. Estimate by proration from estimate for all species together. Total length size class (mm) Lower* Point Upper** 0-99 3273 4615 5957 100-149 312 499 687 150-199 937 1054 1172 200-249 265 301 336 250-299 201 229 257 300-349 97 111 129 350-399 19 21 27 400-449 2 2 3 450-499 2 2 3 500-549 1 l 1 Total 5109 6835 8572 * Lower limit of 95% confidence interval ** Upper limit of 95% confidence interval 42 Table 8. Brown trout biomass (kg) estimates in the l-km study section (all 10 stations) of the Au Sable River, July 17 to 21, 1978. Total length size class (mm) Lower* Point Upper** 0-99 19.52 27.28 35.52 100-149 8.58 13.73 18.90 150—199 49.72 55.93 62.19 200-249 30.43 34.57 38.59 250-299 40.52 46.17 51.81 300—349 31.91 36.52 42.44 350-399 9.33 10.32 13.26 400-449 1.49 1.49 2.23 450-499 2.06 2.06 3.08 500-549 1.55 1.55 1.55 Total 195.11 229.62 269.57 * Lower limit of 95% confidence interval ** Upper limit of 95% confidence interval 43 Brook Trout Brook trout abundance and 95% confidence limits for all 10 stations by 50-mm size class are presented by numbers (Table 9) and biomass (Table 10). These estimates were obtained by proration from combined brook, brown and rainbow estimates for all ten stations. Brook trout comprised 5.6% of total trout numbers and 3.2 percent of total biomass. There were 94 brook trout 100-199mm in total length. Only 14 brook trout 200—299mm in total length were present in all ten stations. No brook trout greater than 300mm were captured. Brook trout abundance and 95% confidence limits for 8 individual stations by 50—mm size class were computed in terms of numbers (Table 15) and biomass (Table 17). Station 4 contained the greatest numbers (129) and biomass (2.16kg) of brook trout. No brook trout were captured in station 6 and only 1 was captured in station 10. The mean number of brook trout greater than 99mm for the 7 stations containing brook trout was 12. Rainbow Trout Rainbow trout abundance and 95% confidence limits for all 10 stations by SO-mm size class are shown in terms of number and biomass (Table 13). It was noted earlier that although rainbow trout were sighted frequently while snorkel diving, only 12 were captured on the combined electrofishing runs. None of the 9 rainbow trout marked on the first run were recaptured. Only 1 rainbow trout was captured in stations 1 and 3 although ten or more were often sighted in each of these stations during snorkel diving observations. Because rainbow 44 Table 9. Brook trout p0pulation estimates in the 1-km study section (all 10 sections) of the Au Sable River July 17 to 21, 1978. Estimates by proration from estimate for all species together. Total length size class (mm) Lower* Point Upper** 0—99 209 295 381 100-149 19 31 42 150-199 56 63 70 200-249 11 12 13 250—299 2 2 3 Total 297 403 509 * Lower limit of 95% confidence interval ** Upper limit of 95% confidence interval 45 Table 10. Brook trout biomass (kg) estimates in the 1-km study section (all 10 stations) of the Au Sable River, July 17 to 21, 1978. Total length size class(mm) Lower* Point Upper** 0-99 ‘ 1.23 1.73 2.24 100-149 0.54 0.88 1.19 150-199 2.99 3.37 3.74 200-249 1.16 1.27 1.37 250-299 0.40 0.40 0.61 Total 6.32 7.65 9.15 * Lower limit of 95% confidence interval ** Upper limit of 95% confidence interval 46 trout were largely able to escape capture, the estimates presented in Table 13 are undoubtedly low. Comparison of Brook and Brown Trout Abundance The number of brown trout are compared graphically to brook trout numbers for 8 stations in Figure 9. Brook trout numbers/station were less than 8 percent of brown trout numbers/station for all stations except station 4 where brook trout comprised 32.0% of brown trout numbers. Biomass of brook and brown trout per station are compared in FigureliL Brook trout biomass was 5.1% and 8.9% of brown trout biomass in stations 2 and 4 respectively. Biomass of brook trout was less than 3 percent of brown trout biomass in the other 6 stations. Brown trout were much more abundant than brook trout in terms of both numbers and biomass for all stations. Brook trout in the 50-99mm and 200-249mm size classes were slightly lighter than brown trout of this size. Thus, the ratios of brook trout biomass to brown trout bio- mass were less than the ratios of brook trout numbers to brown trout numbers. Conversely, brook trout in the 100-149mm and 150-199mm size classes were slightly heavier than brown trout in the same size class. Therefore, the ratios of brook trout to brown trout biomass were slightly greater than the ratios of brook trout numbers to brown trout numbers. This relationship holds for all stations since the mean weight for a species size class was computed from trout weights in all 8 stations to maximize sample size. 47 aoauoum Housman: any * NNHw amm me 0mm Nwoa NONH Mum Nae mung INONQ ImmN Imom Imam nomm lawn loam town loom Ho Nmm NfimmlmONq mmfio NH: man was cmNH New «as omm HmHH Hmuoa mIN N N amalomq MIN N N acqlooq oNINN Nm OH a o N s a N ammuomm Healow mm om w mm m mm m NH ea memloom moalmom mum om NH MN NN mm mN mN NH mmNIomN mmNIoMN NcN mm Nm mm am as Hm Hm sq quIOON wmofiloqw omm me mad 0mm omH mqa ma omfi omH moalomfi annmmN ofim NH Nn me me Nu mo ac em madlooH awomlfiemN mNHq mNN was one Nam Hmm mHN amN mum malom Ho Nmm Aacowv :o~ m m N o m a N Aaav mamas swam Hauoe + sofiuoum sumsoa Hmuoa A.mouufiao m was Hmsofiumum sows sump aoum mommasoamo oumafiumm coaumumlwv xash .uo>Hm oaamm =< onu cw maofiumum Eloofi hm mommafiumo coaumaoaom usouu xooun was aaoum .waas .HN as as .HH manna 48 NN.¢Hm manmm s< osu aw cowumum ElooH mm mommaaumo wav mmmEOHn usouu xooun was :3oum .NH sassy usouH mo wx 0 8 6 l4 2 0 8 6 l4 2 0 8 6 I“ 2 o 3 2 7. 0. 2 2 1 1 11 1.. 1 a 4 I 4 a I 4 q a 4 q a 4 a q 49 I] Number of Trout I Biomass of Trout I I I I I IIIIII’III I I I I I I I I I I I I I I I I I I I IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII I I I I . E. 1 IIII I I III I I I I I I I I I I IIIIIIIIIIIIIIIIIIIIrIIl I I .I IIIIIIII IIIIIIIIIIHIHIHIHIIIIIII I I I I IIII IIIIIIIIIIII IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII I IIEUIIIII II vooooooooooooaaaoooaaoooaooaooooooooooooooonmooooooooooooeoo IIw’IlIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIDIIIIIII I I I III I I III I I I I I I I I IDIIIIIIIIIIIIIDIDIIIbIIIIIIIIIIIIIIIIIIIDIIIIIIIIIIII III IIIII III-I r[ _ p — I a b _ _ F — h u 0 0 0 0 O 0 O O 0 0 0 0 o 0 0 O o 0 0 0 0 0 O 0 O 0 o 3 2 .I. 0 9 8 7 6 5 I4 3 2 .I. 1 1 1 1 moose Ho uwcesz 10 Station Comparison of number and biomass (kg) per lOO-m station for brook and brown trout combined. Figure 8. 50 Table 13. Population estimates for rainbow trout in the l-km study section (all 10 stations) of the Au Sable River, July 17 to 21, 1978. Biomass (kg) estimates shown in parenthesis. Total length Lower* Point Upper** size class (mm) Estimate Estimate Estimate 4 5 5 150-199 (.269) (.386) (.336) 2 2 3 200-249 (.222) (.222) (.333) 5 6 7 250-299 (1.122) (1.346) (1.571) 1 1 1 300-349 (.360) (.360) (.360) 1 1 1 350-399 (.470) (.470) (.470) Total 13 15 17 (2.443) (2.734) (3.070) * Lower limit of 85% confidence interval ** Upper limit of 95% confidence interval. 51 coaumuo squamous any I NNsN amn omm ass osss Hess Hem Non Noam uooos -NNN loss -mms -Now unmm nsom -smN umNN No an NasNucoos sewn ass oNN NNN sNNs saw st mos Noss Nmsoa muN N u I N - u u u . assuoms NuN N u - u u n u N . assuoos smuNN Nm cs - s s N s s N mmmuomm sosuow mm as m as a ma N Na as msmuoom NasuNss sz ms as NN NN mm nN NN as mmNuomN NNNumNN amN ms Nm mm mm as IN NN ms msNuooN soos-NmN Nos ms NIH sz NNN ass ma NNI mNH ass.oms mmsuNNN mms as as so so NN no no NN ass.¢os onnnmssN mawm mNN NNs mes Nam smm Nos Nms st mauom No Nam Aaoomv :oH a N N s m s N Aaav mmmNu muNm HmuoH 4 sowumum nuwaoa Hmuoa .moumaaumo moons saoun was xooup mosfinsoo maoNuoum w soum coaumuoua he moumawuwm .NN ou Na hash .uo>fim manmm s< onu ca msoaumum aloofi hp mommawumo soauoasaoa usouu ssoum .¢~ Nanak 52 sofiuoum amouumas may * Nss I NN «N Ha I ON «NH NON IssN Is INH ION INN I IsN IsN IHN No NNN NIsIoHN NHN s NN NN NN I NN NNI NN Nmuoa III a s I I I I I I I NNNIONN NIN N I I N s I I N N NsNIooN NNINs Na I I N N I N Na NN NNNIONN NNINH NN I N N N I I s NN NssIoos NHNINss oNN I ON as oN I oN NoN «N NNIoN No NNN AaooNv cs N N N N N s N Naav mmmNo ouNm Nmooe coNumum auwamN Nancy .moumaaumo usouu szoun was xooun oosamsoo soNumum m soum :oNumuoua mp moumENumm .whma .NN ou NH mash .uo>Nm manmm s¢ onu sN msoHusum Blood Np moumswumo :oNumasmom usouu xooum .m~ manna 53 a Brown Trout I Brook Trout 1300 P 1200 1100 agony mo Homssz Station Comparison of the number of brook and brown trout by 100-m station. Figure 9. 54 :ONuMIm amouuom: any « wN.NcN MN.NN MN.mN Nm.mm mN.mm oo.Nm oo.nN «m.mN No.cm IwN.0NH ImN.nH INo.NH Imm.mN INm.mN IHN.cN IH¢.NH Imm.0N INN.cN Ho Nam wN.N¢NIwN.0N~ qo.¢CN mm.w~ oN.HN «a.NN om.mN MN.Hm om.HN mm.eN Hw.mN Houoy cm.mImN.N mN.N I I mN.N I I I I I omcIcmc mm.NINN.N NN.~ I I I I I I NN.N I maelooe ow.NHImm.ma mw.m~ sm.q I mm.N Nm.N mm.o mm.H mm.~ mm.o mmmIomm ow.NmIHH.eN NN.wN oN.m Hc.N mm.m do.N om.q do.N Nm.m NN.m mqmloom No.ocIma.Nm ON.om mo.m ow.m No.8 Nq.q HN.o mo.n wo.¢ eo.m mmNIomN ON.mmIom.oN MN.mN NN.H wo.m om.m mN.m mo.m on.m NN.m NH.m msNIOON em.mmI~m.Nc «NINc cm.m o~.o mN.o Nq.o eo.N mm.q cm.m no.0 mmanmH cm.wHImc.N o~.mH oq.o Nm.H mo.~ mN.H mm.~ HN.N HN.H mm.H mefiloofi mm.oqu~.wH mN.mN mo.N wm.m wq.m oN.N mm.m wq.~ NH.H m~.o amlom Ho Nmm Asoomv :mw. m m N o m a N Asav mamas swam Hmuoa I coaumum suwdma Hmuoa .wNmH .HN ou NH Nash .uo>Nm macaw =< was :N soNumum anoH Na mommafiumo Amxv mmMEONn unouu ssoum .ofi manna 55 GONumum amouumas may « cN.N aH.o Nm.o oo.H mm.o I om.o Ho.N Na.H I¢N.¢ ImH.o ImH.o IHN.o IoN.o I IaH.o Iwo.H INH.H Ho wa «N.NI¢N.q mm.m mH.o mN.o mm.o NN.o o mN.o eH.N Nn.H Hmuoa oH.OIaH.o NH.o oH.o I I I I I I I amNIomN om.OIcm.o mm.o I I HN.o HH.o I I Nm.o HN.o meNIOON mN.NImN.N mm.N I I mq.o ms.o I HH.o No.0 mm.o mmHIomH oH.Hqu.o mN.o I mo.o mo.o mo.o I I NH.o mm.o quIooH oN.NImm.o mm.H I «H.o oH.o «H.o I «H.o oN.o Nm.o mmIom No NNN Asoomo IoH N N N N N s N Aaav mmmHo muNm asuoe I. :oNumum sawdma Hmuoa .wNmH .HN ou NH NHsN .No>Nm oHHMm s< onu cw soNumum alooH Na mommaaumo wav mmmsoan usouu Hooum .NH IHHMH 56 :3] Brown Trout - Brook Trout usouH mo NM Station Comparison of the biomass (kg) of brook and brown trout by 100~m station. Figure 10. 57 Comparison of Standing Crop in Various Michigan Streams The standing crop (kg/km and kg/hectare) of Mainstream Au Sable trout is compared to standing crops for other Michigan stream in Table 18. The study area had a greater standing crop of trout (kg/hectare) than all streams except the upper part of the Little South Branch of the Pere Marquette River, P0plar Creek, and a section of the Au Sable Main Branch from Burtons Landing u>Wakeley Bridge which encom- passes the study area. Coopes (1974) lists the standing crop of trout for this section at 166.4kg/hectare, which is slightly more than twice the 82.8kg of trout/hectare found in my study section in 1978. Trout Population, Cover Relationships The length and area of natural, man-made and total overhead cover per station are presented in Table 19. Stations 4 and 6 contained the greatest area of natural overhead cover, while stations Zand 8 contained the greatest area of man-made cover. Station 2 had the greatest total area of cover (118.27m2) and station 9 had the least (24.7m2). Total area of overhead cover for the other 6 stations varied from 45.83— 62.60m2. There was 459.17m2 of overhead cover in the 8 stations. Thus, only about 2% of the surface area of the stream was beneath some form of overhead cover. The numbers of brook trout per station in each 50-mm size class were divided by proration into natural and mandmade cover. These data along with the percentage of brook trout found in man-made cover are shown in Table 20. Less than 45 percent of the brook trout in stations 2—9 were found in man-made shelters. The one brook trout captured in station 10 was beneath a man-made trout shelter. 58 Table 18. Standing crops of trout in various Michigan streams. S 3 tream Trout Abundance Width in Stream and locality Kg/km Kg/hectare meters References Au Sable - Main Branch 240.2 82.8 29.0 1 Au Sable - Main Branch Upper -> Grayling to Burtons Landing 42.2 15.7 26.9 2 Lower — Burtons Landing to Wakeley Bridge 479.2 166.4 28.8 2 Au Sable - North Branch Upper - Dam 2 to Otsego County Line 114.8 53.9 21.3 2 Middle — County Line to Eamon's Landing 229.0 69.6 32.9 2 Lower - Eamon's Landing to Kelloggs Bridge 225.9 77.9 29.0 2 Pigeon River 40.2 31.9 12.6 2 Gamble River 47.7 82.3 5.8 2 Rifle River 36.9 29.3 12.6 2 Boardman River - Upper end of Brown Pond to Forks of the North and South Branch of the Boardman 71.9 55.7 12.9 2 Pere Marquette River - Little South Branch Upper 95.9 97.9 9.8 2 Lower 46.3 40.6 11.4 2 Poplar Creek 42.1 87.8 4.8 2 l - This study, July biomass estimates. 2 - Coopes 1974, Fall biomass estimates 3 - Stream widths from Gaylord Alexander pers. comm. 59 Table 19. Length (m) and area (m2) of overhead cover by station. Natural Cover Man-made Cover Total Station Length Area Length Area Length Area 2 17.9 3.87 96.5 114.40 114.0 118.27 4 54.0 11.30 97.0 47.25 151.0 58.55 5 43.5 5.50 58.0 46.85 101.5 52.35 6 66.7 10.85 43.5 35.00 110.2 45.85 7 16.0 2.53 76.0 43.30 92.0 45.83 8 10.0 3.30 47.0 59.30 57.0 62.60 9 20.0 2.45 37.0 22.25 57.0 24.70 10 59.5 7.00 51.0 44.02 110.5 51.02 Total 287.20 46.80 506.0 412.37 793.2 459.17 Table 20. 60 Abundance of brook trout per 100-m station associated with man-made cover and with other stream area. Total length Station size class (umo 2 4 5 6 7 8 9 10* Total Number of Brook Trout in Man—made Cover 50-99 14 - - - - - 7 - 21 100-149 9 - - - - 3 3 - 15 150-199 2 3 - - 2 3 - - 10 200-249 — — - - - - - — - 250—299 - - - - - - - - 1 Total 25 3 - - 2 6 10 1 47 Number of Brook Trout Elsewhere 50-99 41 102 20 0 20 14 14 - 211 100-149 3 6 - - 3 - — - 12 150-199 9 15 2 6 5 - - 37 200-249 2 3 - - 1 2 - - 8 250-299 - - - - - - — - - Total 55 126 22 - 30 21 14 - 268 % of Total Brook Trout in Man-made Cover 50-99 25 - - - - - 33 - 9 100-149 75 - - - - 100 100 - 56 150-199 18 17 - - 25 37 - — 21 200-249 - - - — _ - - _ 0 250-299 - - - - - - - 100 100 Total 31 2 0 - 6 22 42 100 15 * The upstream section 61 Numbers of brown trout associated with man-made shelters and trout found elsewhere are displayed in Table 21 by station and 50—mm size class. The percentage of brown trout in a given station and 50-mm size class are shown in Table 22. Fifty percent or more of trout 150- -449mm were associated with man-made shelters. Although man-made shelter covered only about 1.8 percent of the stream surface 71.5% of brown trout 3_350mm were in man-made trout shelters in the 8 stations. The percentage of brown trout in man-made shelters increased progressively by 50mm size class from 26% for 50-99mm trout to 100% for 400—449mm trout. The percentages of brook and brown trout in man-made shelters are compared graphically by 50—mm size class in Figure 11. The two trout in the 450-499mm class were not captured in man-made shelters. The percentages of brown trout per station in man-made shelters ranged from 16-58%. The lower values result from the large numbers of small trout which were not captured in man-made shelter. When per-station trout abundance by number of biomass were regressed on per-station amounts of overhead cover—-natural, man-made, and com- bined (Tables 23-25)--no correlation coefficients between trout abundance and Egggl overhead cover or mandmade overhead cover were significant (p=0.05), although the highest correlations were for numbers of brown trout 3 300mm. For natural overhead cover considerd separately, however, the simple correlations between numbers and biomass of brown trout 34250mm and area of overhead cover were both significant (p 8 0.05). The multiple correlation of kg of brook and brown trout 3_150mm 62 Table 21. Number of brown trout per 100-m station associated with man-made cover and with other stream area. Total length Station size class (mm) 2 4 5 6 7 8 9 10* Total Number of Brown Trout in Man-made Cover 50—99 313 27 102 102 75 88 143 163 1013 100-149 36 18 33 42 9 24 39 9 210 150—199 75 38 43 98 63 64 38 35 454 200—249 32 13 19 29 26 20 13 11 163 250-299 15 14 19 20 16 14 12 9 119 300-349 14 5 3 11 6 9 8 4 60 350-399 2 2 2 2 6 2 - 10 26 400-449 - 2 - - - - - - 2 450-499 - - - - - - - - - Total 487 119 221 304 201 221 253 241 2047 Number of Brown Trout Elsewhere 50-99 511 129 95 429 892 374 334 116 2880 100-149 36 45 3o 30 57 36 30 9 273 150—199 50 73 50 46 6O 64 76 28 447 200-249 13 15 11 15 7 11 19 5 96 250—299 2 9 7 13 5 9 7 7 59 300-349 2 7 4 4 2 2 - 6 27 350-399 - 2 2 - - 2 ~ - ' 6 400-449 - - - - - - - - 0 450-499 - - - - - 2 - - 2 Total 614 280 199 537 1023 500 466 171 3790 * The upstream station Table. 22. station. 63 Percent of total brown trout in man-made cover by lOO-m Total length Station size class (mm) 2 4 5 6 7 8 9 10* Total 50-99 38 17 52 19 8 19 30 58 26 100-149 50 29 48 58 14 40 57 50 43 150-199 60 34 46 68 51 50 33 56 50 200-249 71 46 63 66 79 65 41 69 63 250-299 88 61 73 61 76 61 63 56 67 300—349 87 42 43 73 75 82 100 40 69 350-399 100 50 50 100 100 50 - 100 81 400-449 - 100 - - - - - - 100 450-499 - - - - - - - - 0 Total 44 30 53 36 16 31 35 58 35 * The upstream station 64 L y 1 n O = n t t ..u u 0 O r r T. T n k W O O O r. r. B B B I r P p p b . b u b — b 0 O O 0 0 O 0 0 0 0 0 O 9 8 7 6 5 I... 3 2 1.. 1... um>oo oumalsmz :N usoNH mo Honssz Hmuoh mo scooumm 350- 400- 450- 449 499 399 300- 349 299 Total Length (mm) 150- 200- 250- 199 249 100- 149 50- 99 Comparison of percent of brook and brown trout in man-made cover by 50mm size class. Figure 11. 65 .Ho>oH Nm was um unmoamaswao mums muaoaowmmoou oz * HNN.o HoN.o ssN.o NNN.o ONH.o oNs.o as ooN A ooouo oaoom HHN.o ooo.o Nso.o HNN.o ooN.o «Na.o aaomN_M.o=ooo osoom HNH.o NNs.o sss.o ONN.o oHN.o HNN.o EEOON M.ooooo oaoom ONH.o st.o NNN.o NNN.o oNo.o NNN.o aaoNH M.ooooo oaoos ooo sooom NNs.o NsN.o ass.o NNN.o NNo.o NNs.o aaoos M.ooooo oaoom :oHuMum nomImmoaon NNs.o omo.o sNN.o NNN.o NNN.o NNN.o aaooN.M ooooo oaoom oas.o oNs.o NHo.o ONH.o oms.o NNs.o aaomN M.uooos oaoom ONH.o ooN.o NNN.o NsN.o oso.o Nms.o aaooN M.ooouo osoom oes.o NNN.o ooH.o NIN.o Noo.o NNo.o seems M.o:ooo oaouo ooo sooom Noo.o oNN.o sso.o NoN.o Noo.o Noo.o asses M.o=oou osoom :oNuoum mom muonssz Nu u Nu u N» u va mammauo> uooosoaou mo mama moss Hmuou was nuwsoa Hmuoe mono Hmuoe :uwsoa Hmuoe uo>ou oomaIcmz mam Nonnumz mo unsos< «.uo>Hm oHnmm s< onu aw cowumum you uo>oo omozuo>o Housman mam oomBIcoE mo oops Houou was .nuwaoH Hmuou .va moanmauo> soHumasaoa usouu How ANNV cowumcasuouou mo musoNonmooo mam Auv mucoNonmoou coaumaouuoo .MN oNan 66 .Ho>sH N0 saw us uasonNasz onus suasaoammmoo oz a «NH.0 Nmm.0 NOH.0 mHm.0 HOH.0 me.0 ssoom A uaouu azoum mm0.0 HwH.0 0H0.0 «NH.0 Nm0.0 mNH.0 ssomN_M.usouu aaoum Nea.o NNN.o oNo.o osN.o NNo.o NNN.o aeoON M.ooooo osoum mNH.0 00m.0 «HH.0 0mm.0 N00.0 HHm.0 ssomH M.uaouu asoun was xooum H00.0 mwN.0 0N0.0 HmN.0 N00.0 0HN.0 ss00H M.uaouu asoum aoNusus Hon mssaoam NNN.0 0Ns.0 HHN.0 000.0 mmH.o Hmm.0 ssoom M.uaouu aboum 000.0 000.0 H00.0 mm0.0 «00.0 000.0 ssomN M.uaouu aaoum NNo.o NON.o oNo.o OON.o NNo.o NNa.o asooN M.ooouo osoum 000.0 mom.0 000.0 NmN.0 0N0.0 m0N.0 asonH M.uaouu aaoup was xooum oso.o OON.o oso.o NNH.o HNo.o NsH.o asooN M.ooouo oaous aoNusus usQImusnaaz Nu a Nu a Na u va oHnsNus> maswaoaow mo oaNH smus Hsuou was Suwaoa asuoH sous Hsuoe sumasa HsuoH .NHao us>oo swsalasz wo uaaoa< 4.oo>NN oNasN as was oN Haney aoNusus Nos mHao as>oo wsoaas>o swssIass mo sous Hsuou was .aumasH asuou .va sanans> aoHusHaaoa uaouu you ANNV aoNusaNSNsusw 0o suaofloammooo was Auv suasNonmsoo aoNusHouuoo .qN ansH 67 .Ho>sH N0 sag us soasoamaawfis sousoawaH « ssN.o Nos.o sog.o sss.o NNo.o ssN.o aEOON A noonn ozons sss.o sNo.o Nos.o INNo.o Nss.o NNs.o aaosN.M.noonn ozons soN.o NNN.o NNN.o ass.o NNo.o oNs.o aaooN M.noonn ozons «No.o oNN.o Non.o Nos.o Noo.o sso.o aaoss M.n=onn ozons coo goons NNN.o sNN.o an.o NNN.o oooo.o NNo.o aaoos M.noonn ozons aoNusus umm.sss80Nm NNN.o NNN.o NNN.o Nos.o ssg.o osN.o aeoON M.noonn neons NoN.o ONN.o Ngs.o «ssN.o NNN.o NNN.o asosN M.noonn moons NsN.o sNN.o sss.o NNN.o Ngo.o sog.o aENON M.noonn ozons sos.o gNN.o NNo.o NNs.o sso.o NON.o seems M.noonn o3ons sow goons NNN.o NNo.o Noo.o sNo.o NNo.o ssN.o aaoos M.n=onn nsons aONusus numlsusnaaz Nu a Na a Nu m va ansHus> uaswaoasw mo saNH sous Hsuou was auwasH Hsuoe sous HsuoH nuwamH asuoe um>oo asaausz mo uaaoa< uo>NM sansm a< mam aH Awnav aoNusus ass Naao us>oo wssaus>o Hsaausa mo sans asuou was .auwasa Hsuou .va moansaus> aoNusHaaoa uaouu How ANHV aoNusaNEasusw 0o muaoNonwsoo was Any muaoHonmooo aoNusHsuuoo .nN sansa 68 with length and area of natural cover yielded the highest correlation (r=0.796). Most correlation coefficients were less than 0.5. Variables Influencing Density of Trout in Individual Trout Shelters The numbers of brook and brown trout 3_150mm beneath individual shelters and the physical parameters used as independent variables for multiple regression analysis are presented in Table 26. Shelters 1-7 and 18-23 are not shown as they were in stations 1 and 3 which were excluded from analysis because they contained long unwadable pools. The best fit multiple regression models obtained for various trout population variables using the backward elimination procedure are dis- played in Table 27. Maximum water depth proximal to the shelter and total length of overhead cover were the most important of the features I measured in accounting for variation in the per-shelter biomass of brown trout in all cummulative size groups. The coefficients of multiple determination (R2) for these 6 models ranged from 0.15 for the model for biomass of brown trout 3 350mm to 0.38 for biomass of brown trout 3_150mm and bio- mass of brown trout_: 200mm. Trout biomass always was positively related to water depth and length of overhead cover per shelter. According to the model, per shelter biomass of brown trout 1 150mm should increase 17 grams for each 1cm increase in water depth and 44 grams for each 1m increase in shelter length. Water depth and length of cover remained the most important variable influencing biomass of brook and brown trout combined. However, it should be observed that there were only 12 brook trout beneath the 57 69 shelters used to derive these models so one would not expect the models for brook and brown trout combined to differ significantly from the models derived from brown trout alone. For the same reason models derived for brook trout population variables are probably not useful. Water depth, length of overhead cover and percent subcovert sand were the most important independent variables when numbers of brown trout 3 150mm and 3_200mm were the dependent variables. Only water depth and length of cover were important in explaining variation in the numbers of brown trout_: 250, 300, and 350mm/shelter. The addition of brook trout numbers caused only a slight upward shift in the magnitude of the partial re- gression coefficient. When considered within individual SO-mm size classes, water depth and length of cover were generally the most important independent vari— ables, but the models were quite variable. Coefficients of multiple determination CR2) for these models were generally much lower than those for models derived using cummulative numbers and biomass size classes as fewer fish were found in the smaller size intervals. The regression equation for per shelter numbers of brown trout 3_250mm/shelter is as follows. Y = - 0.558 + 0.0447x 4 1 2 Where Y = Number of brown trout 3_250mm/shelter ,+- 0.00109x 4 x1 - Maximum water depth (cm) proximal to the shelter x = Length (cm) of overhead cover/shelter 2 Thus, shelters with 4.5 meters of overhead cover positions such that water depth is 70cm would be expected to hold about 3 brown trout 3_250mm on the average. According to this model, if water depth was held 70 Table 26. Characteristics of 57 Ian-Iade trout shelters and the nuaber of trout beneath thaa estimated by electrofishing. Shelters 1-7 and 18-23 not shown due to unwadable pools in stations 1 and 3. Per shelter Nunbfrlg:-:rout =::::um E:::::a:fcover Shelter - Subcovert substrate 52) depth ength sz' No. type Brown Brook Silt Sand Gravel Rubble (cm) (I) (I ) log jam 5 - 15 - 85 - 55 3.5 2.7 9 log jam 16 - - 10 90 - 50 14.0 7.6 10 log jan 33 — - 90 10 - 75 8.0 16.0 11 raft 9 - - 30 70 - 65 3.5 2.8 12 log jam 12 - - 10 90 - 60 9.0 6.4 13 log jam 12 - 55 30 15 75 17.5 17.3 14 log jam 5 - 50 30 20 - 35 13.0 13.0 15 raft 23 - - 55 45 - 90 11.0 5.2 16 log jam 20 2 40 50 10 - 90 13.0 40.9 17 log jam 3 - - 100 - - 25 4.0 2.4 24 log jam 10 - 30 30 40 - 60 6.0 1.8 25 log jam 23 3 30 23 47 - 75 59.0 17.0 26 raft 5 - 50 20 30 - 65 4.5 4.5 27 log jam 15 - 20 10 70 - 70 9.5 5.3 28 log jam 9 - - 30 40 30 60 8.0 12.0 29 log jan 14 - 21 42 30 7 65 7.0 7.0 30 log jam 20 - 20 70 10 - 70 6.0 6.0 31 log jam 17 - 32 36 32 - 65 20.0 8.6 32 stump mass 7 - - 30 10 60 70 1.5 2.2 33 log jam 12 - 35 25 40 - 55 10.0 7.0 34 log jam 15 - 25 45 30 - 70 13.5 16.0 35 log jam 3 - 100 - - - 40 4.0 3.2 36 log jam 7 - 100 - - - 40 4.0 3.2 37 log Jan 19 - 3O 40 30 - 65 7.0 10.0 38 log jam 21 - 20 50 30 - 80 6.0 4.4 39 log gal 21 2 3O 40 30 - 60 7.0 7.0 40 log jam l6 - 50 50 - - 80 2.5 1.8 41 raft 22 - 100 - - - 50 3.0 0.6 42 raft 27 - 100 - - - 80 1.0 0.3 43 log jII 7 - 20 40 40 - 28 3.0 3.0 44 log jII 19 - 20 40 40 - 70 8.0 7.7 45 raft 14 - 10 80 10 - 90 4.0 4.0 46 log jam 5 - 100 - - - 40 7.0 7.0 47 log jII 23 - 30 30 40 - 50 3.0 6.0 48 log jII 7 - 45 25 30 - 170 11.0 16.0 49 log jII 17 - 15 25 60 - 75 14.0 21.0 50 log jII 23 - 30 40 30 - 80 7.0 6.2 51 stqu Iaas 5 - 50 40 10 - 60 2.0 3.0 52 log jII 16 - 40 40 20 - 85 4.0 4.0 53 raft 17 - 50 50 - - 60 9.0 6.3 54 log jII 26 2 40 40 20 - 75 3.0 3.0 55 log jII 12 - 50 40 10 - 65 4.0 4.0 71 Table 26. (cont‘d.) Per shelter Nuaber of trout :::::u' ::::::a2fcover Shelter '1 150'. Subcover substrate (22 depth Length 1:33 No type Brown Brook Silt Sand Gravel Rubble (cm) (I) (I ) 56 log jam 5 - 20 30 50 40 2.0 0.6 57 log jam 10 - 30 30 40 100 9.0 9.0 58 log jam 5 - 20 30 50 75 7.0 10.5 59 stqu Iass 9 - 4O 40 20 75 1.0 1.0 60 log jam 9 - 100 - - 45 4.0 4.0 61 log jam 11 - 9O - 10 70 3.0 3.0 62 log jam 18 - 90 10 - 80 6.0 6.0 63 log jam 12 ~ 90 - 10 40 12.0 1.4 64 raft 12 - 90 - 10 55 4.0 1.2 65 log jam 13 - 100 - - 50 13.0 1.1 66 log jam 3 - 100 — - 30 4.0 3.1 67 log jam 15 1 100 - - 80 19.0 21.3 68 log jam 34 2 60 15 25 100 44.0 11.4 69 raft 16 - 19 50 40 40 13.0 8.5 70 log jam 11 - 40 3O 30 60 10.0 6.0 72 Table 27. Multiple regression analyses (backward elimination procedure) of trout population variables (Y) on linear physical IaasuraIInts of instreaa trout shelters (n-57). F for removal from regression - 2.0 X - variables entered op first rggression step. 1 - Maximum water depth proximal to the shelter (cm). N I than 10cm. - I subshelter sand. 2 subshelter silt. (D '4 0 VI 5‘ w 9 - 2 of subshelter gravel + subshelter rubble. - Dummy variable - 1 if shelter type is a log jam. - Dummy variable - 1 if shelter type is a sunken log raft. Total length (cm) of overhead cover/shelter more than liusvide and in water deeper - Istal area of overhead cover/shelter more than Hts wide and in water deeper than 10cm. Trout size Variables class remaining 2 (mm) in equation Regression model R P - Brown trout biomass (3!) per shelter Cumulative size groups 3 100 1. 2 Y1 257 + 16.9!1 + .453X2 .37 .001 3 150 1. 2 TI 16.47 + 1711 + .44112 .38 .001 3 200 1. 2 Y1 -87.9 + 15.111l + .39x2 .38 .001 3 250 l. 2 Y, -249 + 13.41l + .34212 .34 .001 .1 300 1. 2 Y1 -243 + 8.811 + .2312 .22 .001 3 350 1. 2 Y1 -216.6 + 4.511 + .15912 .15 .012 We! 100-149 4 Y1 20.94 + .553!“ .06 .066 150-199 2. 4 Y, 410.44 + .06212 - 1.782‘ .08 .098 200-249 2. 5 T, 163.9 + .06712 + 3.2815 .15 .013 250-299 1. 2 Y1 —5.88 + 4.611 + .11212 .25 .001 300-349 1 T, -7.13 + 4.911 .09 .025 350-399 1, 2 T, -228.5 + 4.3!l + .15982 .17 .006 - Brook trout biases (gm) per shelter Cumulative s rou s 0-300 2 12 -1.22 + .26912 .31 .001 .1 100 2. 3 12 -9.68 + .022212 + .0001413 .34 .001 3_150 2, 3, 4 Y2 -6.5 + .0222!2 + .00016!3 - .3161‘ .45 .001 W 100-149 4 Y2 -4.12 + .358!‘ .16 .002 150-199 2 Y -9.0 + .02382 .40 .001 73 Table 27. (cont'd.) Trout size Variables class rushing 2 (Is) in equation Regression model 8 P ‘13 - Brook and brown trout bio-ass (gm) per shelter gig-glam“ size gm 3100 1, 2 Y3 - 238 + 17.2X14 .4781!2 .38 .001 3150 1, 2 Y3 - 145.7 + 17.2!l + .466x2 .38 .001 3 200 1, 2 Y3 - -92 + 15.2Xl + .3981!2 .38 .001 3 250 1. 2 Y3 - -253.34 4» 13.5111 + .345112 .34 .001 1" - Number of brown trout per shelter Cumulative size groups 3100 1, 2, 4, 5 Y4 - 9.47 + .08111 + .0033X2 - .08211‘ + .072115 .26 .004 3150 1, 2, 5 Y‘ - 4.38 + .08111 + .0026)!2 + .064115 .26 .001 3 200 1, 2, 5 Y4 - .189 + .051114 .0016x2 + .o34x5 .39 .001 3 250 1, 2 Y“ - -.558 + .0447111 4» .00109112 .38 .001 3 300 1, 2 Y4 - —.53 + .02111 + .0005112 .23 .001 3 350 1, 2 Y“ - -.448 + .0091!l + .0003x2 .17 .007 SO-un interval groups 100—149 4 Y‘ - 4.72 - .0382X‘ .06 .066 150-199 2, 4 Y‘ - 7.74 + .0012X2 - .034X‘ .08 .098 ZOO—249 2, 5 Y,‘ - 1.43 + .0006112 + .029115 .15 .013 250-299 1, 2 '1. - -.028 + .02114- .0006x2 .25 .001 300-349 1 Y‘ - -.022 - .0211 .09 .025 350—400 1, 2 Y1. - -.462 + .009X1+ .0003112 .17 .006 Y - Number of brook trout per shelter Cumulative size groups 3100 2, 4 Y5 - -.215 + .0004!2 + .00922‘ .26 .001 3 150 2, 3 ’5 - -.249 + .000th 4» .0000113 .44 .001 50. interval grogps 150-199 2 Y - -.182 + .0004!2 .38 .001 74 Table 27. (cont'd.) Trout size class Variables 2 (-1 entered legression model I P ‘6 - lumber of brook and brown trout per shelter Cumulative size groups 3 100 1, 2, 4, 5 '6 - 8.99 + .0911 + .0037!2 - .07291‘ + .0704115 .27 .002 3 150 1. 2. 5 Y6 - 4.06 + .08!l + .003x2 + .0645:5 .29 .001 3 200 1, 2, 5 Y6 - .187 + .051!l + .0017x2 + .0328)!5 .39 .001 3 250 1, 2 Y6 - -.579 + .045111 + .0011112 .38 .001 75 to 70cm, length of overhead cover must be increased 8.6 meters if the cover is to hold 1 more brown trout 250mm or larger. The mean water depth proximal to the 57 shelters was 65.4cm with a standard deviation of 22.6cm. Mean length of overhead cover per shelter was 8.88m with standard deviation of 9.49m. The per-shelter number of brown trout 3 250mm varied from 0-10 with mean 3.34 and standard deviation of 2.59. Although there was no systematic deviation of the numbers of brown trout 3 250mm per shelter from the numbers predicted from the model, an examination of outlying observations is instructive. The above model predicts that shelter number 25 should hold 9.2 brown trout-3 250mm, whereas only 7 were present (in the ensuing examination of individual shelters I will refer to brown trout 3_250mm simply as trout). This log jam shelter consisted primarily of logs 2-5m long which were mostly less than 30cm wide. The logs were not close together and were not constructed to form a solid, light-attenuating surface. Shelter number 27 held 6 trout versus the predicted 3.6. This log jam shelter extended about 7 meters perpendicular to the stream bank and had one solid block of overhead cover 3.5 by l.Onn It was largely surrounded by silt and macrophytes, but water flow beneath the shelter had scoured away most fine sediments, as 70% of the subshelter bed was gravel. Although the deepest water proximal to the shelter was 70cm, most of the shelter was no more than 30cm above the stream bed. Shelter number 29 held 6 trout versus the predicted 3.1 This shelter also consisted of solid blocks of overhead cover about 1m wide underlain by fairly swift waters as evidenced by the presence of 37% subshelter gravel and rubble. Most of this structure was less than 76 35cm above the stream bed with a proximal water depth of 65cm. Shelter number 34 held almost 4 more trout than expected. This log jam shelter extended diagonally downstream from its point of attach— ment to the bank and acted as a wing deflector as well as overhead cover. The current was swift around the midstream side of the device and slower beneath some portions of the solid log jam which was 0.70-2.0m wide. Shelter number 40 held 7 trout versus a predicted 4.3. This log jam shelter was triangular in shape with the apex directed upstream into the current. The water was shallow on the south side of the de- vice but swift 75cm deep water swept along the north side where most of the trout were captured. The bulk of the device was solid and mostly less than 40cm above the stream bed. In general, shelters with deep water sweeping past at least one side held the most fish. One notable exception was shelter 48. However, the water adjacent to this shelter was too deep and swift to electrofish effectively and was also frequented by a 596mm northern pike. Regression Models with Polynomial and Interaction Factors Upon examination of plots of trout pOpulation variables on poly- nomial and interaction variables, I determined that they might provide better fits of the data than the ones above. Maximum water depth pro— ximal to each shelter, length and area of overhead cover/shelter, and subshelter substrate percentages were all squared for use as potential elements in the new models. In addition, maximum water depth proximal to each shelter was multiplied by length and area of cover/shelter to create variables to account for interaction factor effects on the 77 dependent variables. The product of length and area of overhead cover/ shelter and the linear factors used in the models previously examined were also available for entry into the new models. Trout population variables, independent variables, regression equations, coefficients of multiple determination, and significance levels for the new series of models are shown in Table 28. Both the backward elimination and forward stepwise regression methods result3 in identical models for almost all trout population variables. Best- fit regression equations derived for size classes of trout which con- tained large numbers of fish usually contained the independent variables of maximum water depth proximal to the shelter times length of over- head cover, maximum water depth and maximum water depth squared. The R2's for the new models were higher for trout variables involving biomass of brown trout 3 250mm (R2 = 0.43). The highest R2's (R2 = 0.65) were obtained for the two models which described abundance of brook trout of 150-199mm as functions of length of overhead cover/shelter, squared, area of overhead cover/ shelter, squared, and the product of length and area of cover/shelter. These two models were based on only 11 brook trout distributed among 5 of the 57 shelters. Usually, however, models for trout size classes with small numbers of fish containing polynomial and interaction factors did not provide better fits of the data than models with simple linear independent variables. In summary, the most important independent variables were (1) the product of maximum.water depth proximal to each shelter and length 78 Table 28. Multiple regression analysis (stepwise procedure) of trout population variables (‘1) on linear. polynomial and interaction factor variables involving meteor-eats of instream trout shelters (n-57). P for entry into regression - 3.0. x - variables available for est into r ression e ations Maximum water depth (cm) proximal to each shelter x total length (cm) of cover/shelter more than lOcmvide and in water deeper than 10cm. 2 - Maximum water depth proximal to each shelter (cm). 3 - Maximum water depth squared (cmz). 4 - Tatsl length of overhead cover (cm)/shelter more than 1030 wide and in water deeper than 13cm. 5 - T0tal ares (cmz) of overhead cover/shelter more tbsnlfihm wide and in water deeper than 10cm. , 6 — Total length of overhead cover squared (cI‘)/shelter. 7 - Total area of overhead cover squared (cm‘)/shelter. 8 - Maximum water depth proximal to the covert (cm) a total area of overhead cover (cm2)/ shelter. 9 - Total length of overhead cover (cm)/shelter x total area (cm') of overhead cover/shelter. 10 - Percent subshelter silt. 11 - Percent subshelter sand 12 - Percent subshelter gravel + percent subshelter rubble. 13 - Percent subshelter gravel + percent subshelter rubble, squared. 14 - (Percent silt)2 15 - (Percent sand)2 16 - Percent subshelter gravel. 17 - Dummy variable - 1 if shelter type is a log jam. 18 - bully variable - 1 if shelter type is a sunken log raft. Trout size class Variables 2 (mm) entered Regression Iodel I P Tl - Brown trout biomass (gm) per shelter Cumulative size ran a 3_100 1. 2. 3 r, - -1547 + .516 x ‘°-2‘1 + 71.77:2 - .35723 .53 < .001 3 150 1, 2. 3 r1 - -1642 + .501 x 10'2r1 + 71.7622- .35623 .53 < .001 _>_ 200 1. 2. 3 r, - -1506 + .449 x 10"211 + 511.721:2 - .2115:3 .50 < .001 i 250 I. 2, 3 r, - -1328 + .385 x ”.251 + 47.09112 - .22sz .43 < .001 3 300 1 t, - 324 + .338 x 10":1 .16 .002 3 350 1 r1 - 76 + .232 x 10"":1 .14 .005 50-Im interval Igggps 100-149 13 r, - 120 - .015:13 .08 .034 150-199 1 rl - 369 + .727 x 10”:1 .06 .061 zoo-249 11. 1 r, - 173 + 3.22111 + .83 II 10":1 .16 .009 250-299 1, 6 ’1 - 225 + .354 x Io‘zxl - .305 x 104:6 .25 < .001 300-349 2 r, - -7.14 + 4.92112 .09 .025 350-399 1 r, - 47 + .23 x 10'“:l .16 .002 150-349 1, 11. 2, r, - -973 + .332 II 10":1 + 6.19111 + 46.6712 .45 . .001 3 - .242: 3 Table 28. (cont'd.) 79 Trout size class Variables 2 (mm) entered Regression model .I P Y, - Brown trout biomass (gm) per shelter W 200-349 1. 2, 3 rl - -814 + .256 x 10":1 + 39.29:2 - .165x3 .41 .001 250-349 2, 4, 3. r1 - -692 + 26.26112 + .479x‘ - .12:3 - .603 .39 .001 6 x 10“'1I,5 - Brook trout biomass (gm) per shelter Cumulative size grogpg _>_ 100 9 22 - I. + .168 x 10":, .39 .001 3 150 9, 10 r, - -13 + .179 x 10":, + .2251:10 .53 .001 50-mm interval groups 100-149 13 r, - -1 + .502 II 10”:13 .20 .001 150-199 6, 7. 9 r2 - 4 + .1097 x 104:6 + .133 x 10'81.’ -.262 .65 .001 x 10'“:9 - Brook and brown trout biomass (gm) per shelter Cumulative sizeggroups ; 100 1, 2, 3 r3 - -1565 + .548 x 10":1 + 72.63112 - .363113 .54 .001 3, 150 1. 2, 3 r, - -1673 + .534 x 104:, + 72.53x2 - .3Is1x3 .54 .001 > 200 1, 2. 3 r3 - -1518 + .453 x 10":l + 59.05:2 - .26713 .50 .001 > 250 1, 2. 3 r3 - -1340 + .39 x 10":l 4» 47.4212 - .22423 .43 .001 - lumber of brown trout per shelter Cumulative sire gEgups -4 “2 _>_ 100 1, 13 r‘ - 16.5+ 1126 x 10 x, - .1096 x 10 In .20 .002 g 150 11. 1 r, - 9.3 + .362 I. 10"‘x1 + .0749:11 .23 .001 3, 200 1. 11, 2. r, - -5.6 + .169 x 10"x1 + .0275:u + 2:16:2 .53 .001 3 - .119 x 10”:3 ; 250 1. 2. 3 r, - -3.9 + .124 x 10"‘xl + .15112 - .0007:3 .47 .001 t 500 1. 2. 3 r‘ - -2.2 + .563 I. 10"!1 + .0743:z - .341: 10"313 .27 .001 g 350 1 r, - .13 + .467 x 10":, .15 .003 Table 28. (cont'd.) 80 Trout size class Variables (In) entered Regression Iodel 32 P Y“ - Number of brown trout per shelter SO-In interval grougs 100-149 13 14 - 4.4 - .544 x 10’31113 .08 .34 150-199 1 Y, - 6.96 + .137 x 1o"):l .06 .061 200-249 11, 1 Y4 - 1.5 + .026xll + .723 x 10'511l .16 .009 250.299 1, 6 1‘ - 1.11 + .174 x 10"‘xl - .15 x 10’6x6 .25 < .001 300-349 2 Y4 - -.022 + .015x2 .09 .025 350.400 1 Y4 - .095 - .465 x 10'5xl 16 .002 Y5 - Nuuber of brook trout per shelter Eggulative size grougg : 100 1, 13 15 - -.11 + .53 x 10’5111 + .138 x 10’311l3 .30 < .001 1 150 9 75 - -.05 + .276 x 10'8119 .53 < .001 SO—un interval grougp 100.149 13 15 - -.04 + .173 x 10'3x13 .20 < .001 150-199 6, 7, 9 75 - .07 + .205 x 10'6116 + .246 x 10'1°x7 .65 < .001 - .49 x 10'8x9 Y6 - Number of brook and brown trout per shelter Cu-ulative sizeggroups ..l. 3 100 1 16 - 15 + .474 x 10 x1 .17 .001 3 150 1, 11 Y6 - 9.2 - .434 x 10"“xl + .0756:11 .26 < .001 3,200 1, 11, 2, 76 - -5.7 + .191 x 10“xl 4 .0261:11 + .2361:2 .53 < .001 3 - omlzxa > 250 1 2 3 7 - .4 + 126 x 10": + 152 - 71 x 10"3 47 < 001 — U D 6 C l 0 x: 0 X3 0 o 81 of overhead cover/shelter, (2) maximum water depth proximal to each shelter, and (3) maximum water depth proximal to each shelter squared. These variables were present respectively in 30, 18, and 16 of the 43 regression equations presented in Table 28. Regression Analysis of Snorkel-diving Counts and Physical Parameters of Shelters The numbers (from three observation periods) of brook and brown trout : 150mm sighted by snorkel divers beneath 39 trout shelters are shown along with a number of physical parameters in Table 29. Although all observations were made sometime between 0830h and 1630b there was considerable variation among the numbers of fish sighted/shelter on the 3 observation periods. The correlation between mean snorkel diving counts and electrofishing estimates for the 28 shelters where the data overlaps was only 0.365. The mean numbers of trout/shelter : 150mm was 12.04 for electrofishing estimates and 2.65 for snorkel diving ob- servations for these 28 shelters. I used the snorkel diving observation trout counts to derive a fitted multiple regression equation. Numbers of brook and brown trout .3 150mm was used as the dependent variable. The independent variables available for inclusion in the regression equation, coefficients of mul— tiple correlation, probability levels, and best 2 and 3 variable regres- sion equations are presented in Table 30. The equation with 3 indepen— dent variables included the variables of percent subshelter gravel, area of overhead cover/shelter and the product of maximum water depth proximal to each shelter and area of overhead cover. These three variables 532 Table 29. Characteristics of 39 man-made trout shelters and the number of trout 1 1501-1 sighted beneath them while snorkel diving. Shelters whose identifying tags were removed by vandals are not shown. Number of brook and p3, .helt.r brown trout 1 150mm “‘31.". amount of Snorkel observation Hater overhead cover Shelter period Subcovert substrate S22 depth Length Area No. type 1 2 3 511: Sand Gravel Rubble (cm) (m) (.3) 1 log jam 2 l l - 90 10 - 45 5.0 2.7 2 log jam 1 2 1 50 50 - - 90 8.0 4.3 3 log jam 2 3 1 5 50 45 - 55 4.5 1.7 4 raft l 2 1 - 100 - - 90 4.0 1.6 6 log jam - - - 30 70 - - 40 6.0 2.1 7 log jam 3 2 2 5 80 15 - 55 17.6 9.9 8 log jam - 1 1 15 - 85 - 55 3.5 2.7 9 log jam 5 1 6 - 10 9O - 50 14.0 7.6 10 log jam 2 5 - - 90 10 - 75 8.0 16.0 11 raft 1 - 1 - 30 7O — 65 3.5 2.8 12 log jam 4 2 2 - 10 90 - 60 9 0 6 4 13 log jam 5 4 3 - 55 30 15 75 17.5 17.3 14 log jam - 1 5 50 30 20 - 35 13.0 13.0 16 log jam 9 3 - 40 50 10 - 90 13.0 40.9 17 log jam - - - - 100 - - 25 4.0 2.4 18 raft - - - 100 - - - 60 1.0 0.1 19 raft 1 2 2 33 34 33 - 7O 1 5 0 6 20 log 3am - 1 - 100 - - - 55 4.0 1.2 22 log jam 5 3 5 5 4O 55 - 110 11.0 12.9 23 log jam 2 3 5 - 50 50 - 20 4.0 3.2 26 raft - - - 50 20 3O - 65 4.5 4.5 27 log jam 4 3 2 20 10 70 - 70 9.5 5.3 28 log jam 3 6 2 - 3O 4O 30 60 8.0 12.0 32 stump mass 1 2 1 - 30 10 60 70 1 5 2 2 34 log jam - - - 25 45 3O - 70 13.5 16.0 35 log jam 1 2 1 100 - - - 40 4.0 3.2 36 log jam 2 1 2 100 - - - 40 4.0 3.2 37 log jam 6 4 6 30 40 30 - 65 7.0 10.0 38 log jam 5 3 2 20 50 3O - 80 6.0 4.4 39 log jam 4 3 4 3O 40 30 - 60 7.0 7.0 40 log jam 4 1 2 50 50 - - 80 2.5 1.8 41 raft - 2 1 100 - - - 50 3.0 0.6 43 log jam 3 - - 20 40 40 - 28 3.0 3.0 44 . log jam 5 7 9 20 40 40 - 70 8.0 7.7 45 raft l 3 4 10 80 10 - 90 4.0 4.0 46 log jam 2 1 2 100 - - - 40 7.0 '7.0 47 log jam 4 5 4 30 30 4O - 50 3.0 6.0 48 log jam 4 5 2 45 25 30 - 170 11.0 16.0 69 log jam 1 2 - 40 30 .30 - 40 13.0 8.5 Table 30. I.) I Ln I Percent Percent 83 Multiple regression analysis (backward elimination procedure) of the mean number (from three snorkel diving observation periods) of brook and brown trout 3_150mm (Y) beneath 39 man-made trout shelters on various physical parameters. 3 - variables entered on first regression step subcovert sand, gravel. and rubble. gravel and rubble. , Length squared (cm). of overhead cover per shelter more than Ukm.wide and in water deeper than 10cm. Area squared (cm)* of overhead cover per shelter more than 10cm wide and in water deeper than 10 C31. flaxinun water depth (cm) found proximal to each shelter a length (cm) of overhead cover/ shelter Maximum water depth (cm) found proximal to each shelter x area (cm)2 of overhead cover. Length (cm) of overhead cover x area (cm)2 of overhead cover/shelter. Maximum Percent Percent Percent Percent Maximum Length water depth squared (cm)2 proximal to each shelter. subshelter silt. subshelter sand. subshelter gravel. subshelter rubble. water depth (cm) proximal to each shelter. (cm) of overhead cover per shelter. Area (cm)2 of overhead cover per shelter. Dummy variable - 1 if shelter type is a log jam. Number of trout Variables > 150:; remaining 2 (Y) in equation Regression Equation I P Y 6, ll. 15 Y - 0.73 - 0.132 x 10.716 + 0.189111 0.32 0.003 -4 + 0.271 x 10 215 r 11, 15 r - 1.04 + 0.0194:ll + 0.912 x 10": 0.26 0.003 15 84 accounted for 32 percent of the variation in observed trout numbers/ shelter. The best fit equation with 2 independent variables contained the variables of percent subshelter gravel and area of cover/shelter. These two variables accounted for 28 percent of the variation in ob- served trout numbers/shelter. Trout numbers/shelter increased when the magnitude of these two variables increased. Comparison of Shelter Types The abundance of trout/m2 of overhead cover/shelter for 3 shelter types and 5 trout size classes is presented in terms of biomass (Table 31) and numbers (Table 32). Abundance of brown trout/m2 of overhead cover per shelter is compared graphically for the 5 size classes in terms of biomass (Figure 12) and numbers (Figure 13). Sunken log raft shelters held more trout of all size classes/m2 of overhead- cover than the other 2 shelter types. Submerged log rafts held from 19-50% more trout/m2 than stump shelters and 133-200% more trout/m2 than log jam shelters for the 5 size classes. Rafts held 19% more brown trout_: 250mm/m2 than stump shelters and 192% more trout of this size/m2 than log jam shelters. To test for differences in abundance (biomass) of trout beneath the 3 shelter types, treatment means were adjusted for three covari- ates, maximum water depth proximal to the shelter x total length of overhead cover, maximum water depth proximal to the shelter and max- imum water depth squared. Interaction effects between covariates and treatments (shelter type) were not significant (0 = 0.05). There were no significant treatment effects (a = 0.05). 85 NENHN.mumNHm some a u a o Nacwo.Nnomam dome m u a u Namm~.wumufim come no u a n Nammm.nnoNHm omma mm u m m mm ems om mam.o amm.o mo~.o Nm~.o one M.u=ouu 656“; van xooum we can ow acn.o «on.o moo.o «no.o oom.M nacho cache mm was sen oom.o am~.o Nun.o m~n.o on~.m naouu nsoum as own “a mam.o naa.o mmn.o mun.o oo~.M nacho :zoum mm mmn aw mam.o nwm.o ~o~.o om~.o omn.m nacho usage on can Na 6mm.o smm.o onu.o qu.o ooH.M nacho sauna A3 A3 A3 A3 18 3V .1: A55 830 03m fiwcfl «o N mm mo N no mo N on uaosm muoofiocm muamnm momma Houou was mmfioonm Amvnxqv Amv1A6v Amounmv sums won macaw am“ panamam Siam won H2 waxy nouaonm .mumuamsm uoouu dome Isms mo moahu m How um>oo omozuo>o mo momma unmovm m sundown unouu mo wav mmMEown omumaaumm .Hm oHan 86 NENHN.muoNNm some m n a m Nacwo.~9mmam some n n a o Namw~.wnuufim anus no a n ~5mm~.m unam some mm a n aw «ma ma sm.s om.m ~N.n ca.n one M.u=ouu each; man xooum om oo~ col wq.o Nm.o on.o an.o com M.u=ouu cause as use can an.n om.o am.o oa.o on~.m nacho cause we was as om.n m~.n ma.o ma.o oo~.M uaouu canon aw fine as sm.q om.m ao.n ma.n omn.m nacho sauna mm man as oo.m 6m.m NH.N m6.~ con.m nacho sauna Ame ANV Ame lav aims paws misc mmmnu «new nausea mo N on mo N no mo N no mHWuHonm muoumocm muouamsm momma HouOu mom moduunm AnvuAev Amvnaqv AmvufimV some mom cesnm as” umummsm cmxcsm won flea maze Houaocm .muouaonm unouu mansions mo monzu m new uo>oo unocuo>o mo Houme ouwscm a sundown usouu mo human: woumawumm .Nm «Haas 87 0.6 E Submerged Log Raft Shelters (n=9) I Stump Shelters (n=3) ‘ES Log Jam Shelters (n=45) 0.5 mo>ou ommsum>o mo N\\\\\\.\\\\. x \ 1. . 1 . \ . \ ..x\ .. \ N .V.w!.‘\V\VV VVV‘VV‘ \\.\\V\\1\1\\\ \ x . 4 .. . ..4..\.. t, . . 4 1 .. _ \\w~.\\\ \ ..txmm ., \. w!‘l1\.\300 >250 >200 >150 >100 Total Length (mm) Cumulative Size Groups Comparison of biomass (kg) of brown trout beneath 1 square meter of overhead cover for 3 shelter types and 5 size classes of trout. Figure 12. 88 E: Submerged Log Raft Shelters (n=9) I Stump Shelters (n=2) Log Jam Shelters (n=45) v.4 .. 431). . - -11.. 4.44.4.3. 4 4.4.4 . ..4 £4 1.4.x 4.4\\ s “4 ...44\\4\\\\ . 1 \ \.4 x 4 ,. \,, \ .\ . , .4 4 .4 44\,,, 4 \.\ \\ 44 4 . . .. 4 x . 4 .\..44,\ 4x. . a , , \ _ h h P 5 14 3 2 .I. 0 mm>oo amonum>o mo NE\u=ouH :Bomm mo monasz >300 >250 >200 >150 >100 Total Length (mm) Cumulative Size Groups Comparison of the number of brown trout beneath 1 Figure 13. square meter of overhead cover for 3 shelter types and 5 size classes of trout. DISCUSSION Shelter Characteristics Correlated with Trout Abundance Linear Regression Models Two covert characteristics closely correlated with per-shelter trout abundance were length of overhead cover and depth of water in which the cover existed (Table 27). Length of overhead cover was more important than area of overhead perhaps because trout tend to position themselves near the edges of coverts rather than distributing themselves evenly throughout the subshelter area (Gibson and Keenleyside 1966). Deep water adjacent to cover might provide security for trout moving away from cover to feed. Trout positioned in deep water near shelters may have moved into cover as the electrofishing crew approached. If deeper water held more fish than shallow areas, this might partially explain why more trout were captured under shelters near deep water. Few trout were captured in deep water when no overhead cover was nearby. This could be interpreted as meaning that trout did not find deep water as attractive when overhead cover was not nearby. One the other hand, it could mean that trout, even if preferring deep water, fled from the electrofishing crew until they reached cover whether it was near or distant. Nonlinear Regression Models Models for per-shelter trout abundance with interaction and poly- nomial terms (Table 28) provided better fits of the data than the models with 89 90 . simple terms (Table 27) because the relationship between trout abun— dance and water depth was quadratic rather than linear and because the product of length of cover per shelter and water depth was a better predictor of abundance than length alone. The importance of the product of length and maximum water depth in explaining variation in trout abundance is probably related to the association of trout with the edges of overhead cover adjacent to deep water as discussed above. Although water depth was only measured at the deepest point beside each shelter, coverts with deeper water at one place usually had deeper water at other points. Most polynomial models for cummulative size groups were of the form Y = a + b x + b x - b3(x2)2 where x is the product of length 1 1 2 2 1 of cover per shelter and maximum water depth, and x2 is maximum water depth adjacent to each shelter. The magnitude of b was always much 3 smaller than that of b2. Hence, for water depths such as those en- countered in this study, which were mostly 25-100cm, the relationship between trout abundance and water depth remains strongly positive. According to these models, per-shelter trout abundance increases more slowly as water depth is increased. The models are probably not valid for very deep water since they imply that the net effect of water depth on trout abundance eventually becomes negative. It should be stressed that the models cannot be applied to data outside the range of values used to derive the models. The negative y-intercepts in some of the equations in Tables 27 and 28 imply that greater water depths and shelter lengths are re- quired for large fish than for small ones. Allen (1969) noted that as 91 fish grow, the size of their territory increases and its physical characteristics change. Chapman and Bjornn (1969) reported that chi- nook salmon and steelhead trout shift to faster deeper water as they grow. The relatively low R2 values to the models in Tables 27 and 28 indicate that the predictor variables examined did not explain much of the variation in subshelter trout density. The lack of direct data on subshelter current velocities may partly explain the weakeness of the models. As stated in the methods section, substrate types beneath shelters were measured to give indication of current velocities. Coarser substrate usually indicates that current velocity is enough to remove finer sediments. Organic silts are light and easily washed away. I now realize, however, that the current which laid down or eroded the sediments may not have been the current which prevailed at the time of measurement. One—way correlations between subshelter silt percentage and trout abundance were negative. Baldes and Vicent (1969) state that velocity in resting microhabitat must be sufficient for the fish to maintain orientation and must not be too high for fish to maintain position economically. Shelters with much silt beneath them may have had current velocities which were too slow for trout to maintain orientation or slower than the trout preferred in some other respect. Data on one-way correlations was not presented in the results because I believe them to be misleading. Some variables in predictor equations often become meaningless when the effects of other variables are accounted for. The multiple regression used in this study avoids 92 such errorstw'removing variables from the equation when they are highly correlated with other variables in the equation. Percent silt beneath coverts was not important in explaining variation in subshelter trout abundance after the effects of other variables had been accounted for in the models. The high probability levels associated with the equations in Tables 27 and 29 (most P's were around 0.001) might seem to indicate that the alpha for entry into the regression equations was unrealistically small. 0n the contrary, it was around 0.1. Coefficients of multiple determi- nation for equations containing all the independent variables were not appreciably higher than those obtained for the best fit equations. After the effects of water depth and length of cover had been accounted for, the addition of other predictor variables did not explain significant amounts of variation in trout abundance. Regression Models Based on Snorkel Divinngounts The 3-variable equation derived from snorkel diving counts (Table 30) contained the product of water depth and length of cover per shelter as did models derived using electrofishing data, but it also contained the variables of percent subshelter gravel and area of over- head cover per shelter. This contrasts with the models based on electro- fishing data, derived by the backward elimination method, in which area of overhead cover was usually one of the first variables to be removed by the regression procedure. In the 2-variable equation derived from snorkel diving data, only the effects of gravel and area were accounted for. The results are probably not as dependable as those derived from electrofishing data for a number of reasons. First, the snorkel diving may have disturbed many fish and chased them away from shelters before 93 they were sighted. A major reason for this was low underwater visability. Visibility was usually only about 4.5 meters as measured by the maximum distance the yellow tip of a snorkel tube could be seen underwater. Camouflaged trout would have had to be considerably closer for a diver to see them. Trout this close to a diver were probably disturbed by the noise made while crawling upstream over gravel against the swift current. Furthermore, the water pouring around a diver's body sounded like water running over and around a large stone and would have sounded unusual to fish accustomed to the quiet waters of the Au Sable. Fish could have run into or out of shelters beyond the perimeter of visi- bility. Visibility beneath shelters was lower than in midstream where visibility measurements were taken so trout may have been frightened away before they could be seen and counted. The large amounts of lateral concealment beneath many shelters undoubtedly hid some fish from view. A number of investigators have found that salmonids are not unduly disturbed by divers approaching from downstream (Ellis 1961; Keenleyside 1962; Fausch 1978). In this study, low visibility and high current velo- city reduced the effectiveness of diving techniques. The underwater light used to illuminate beneath large shelters probably also frightened some fish away. The small intershelter and relatively large intrashelter variability of trout density among the 3 observation periods (Table 29) suggests that the variance in numbers of trout counted beneath shelters is larger than desirable and thus snorkel diving counts are not a good basis for the derivation of models. The criticism of snorkel diving results above do not necessarily make all conclusions invalid. The equations in Table 30 indicate that 94 percent gravel was positively correlated with trout abundance. The presence of gravel beneath shelters usually indicated the presence of moderate current velocities. It was noted in the results that shelters with deep swift water on at least one side tended to hold more trout. Trout :_250mm preferred shelters in deep swift water more than smaller trout, as evidenced by the high proportion of these trout to all trout under many wing-deflector type log jam shelters. Gruber (1978) reported that 250-300mm brown trout preferred a range of sub-cover water velocities between 12.5 and 17.5 cm/sec. More areas with water velocities in this range may have been present beneath deflector type shelters than beneath shelters whichwere largely silted or those in slow water. Such velocities seemed to occur adjacent only to the downstream end of many log shelters of the deflector type, while current velocity adjacent to near-bank portions of the shelters was nearly zero. Deflec- tors of this type provided ample hiding space for frightened fish. Perhaps a combination of small interstices near the banks and more open area beneath cover adjacent to deeper and swifter water is preferred by trout. The preference for deflector-type log jam devices seems especially strong for trout :_300mm. Perhaps this is because larger fish select microhabitats with faster current velocities (Everest and Chapman 1972; Chapman and Bjornn 1969). It may also be related to principle lines of drift and feeding behavior (Jenkins 1969). Swift current carries more food items per unit time than slower water. 95 Comparisons of Shelter Types LongamsJ Rafts and Stump Shelters Whether the shelter was a log jam, a log raft or a stump cover did not significantly influence per-shelter trout abundance. When such shelter types were entered into models (dummy variables), they had negli- gible effect, and were eliminated in the regression procedure. The comparisons of shelter types were not significant owing to great variability of trout abundance among shelters of the same type. The finding that submerged rafts consistently held more trout per square meter of overhead cover than other shelter types (Tables 31 and 32) warrants a closer examination. Not all submerged rafts were in near— midstream positions. The 3 rafts holding the most fish were close to the banks near abundant natural cover. Many of the trout taken in these shelters were undoubtedly chased there from the natural cover. Ac- cording to my electrofishing estimation procedure, submerged raft shelter number 41 and number 42 held 37and190 trout :_150mm per square meter of overhead cover respectively. Clearly this is impossible. This result is based on what I arbitrarily measured as cover (see Methods). Actually shelter 41 was 1 by 3m and shelter 42 was 1.5m by 4m and owing to darkness beneath these devices I may not have measured all the cover that the trout were actually using. Even so, 23 trout 3_150mm were actually caught from beneath shelter 41 and 30 trout :_150mm beneath shelter 42. The mean density of trout Z 150mm/m2 for the other 55 shelters was 3.1 1:2.7. There were no extreme outlying values for density among the remaining 55 shelters. I conclude that fish were drawn or herded 96 from natural cover or other stream areas to be captured under rafts 41 and 42 to a greater extent than for other shelters. If these two shelters are omitted from analysis, the trout density beneath sub- merged rafts falls well within the density ranges found beneath log jam and stump shelters. The mean density of trout 3_150mm for the remaining log rafts is actually less than the mean density beneath the 3 stump shelters. Therefore, trout did not show preference be— tween any of the shelter types examined. Type of shelter -- log jam, submerged log raft, or stump shelter -- did not influence trout abun- dance as much as other factors. Trout Density in Bank Shelters Snorkel diving observations indicated that trout were abundant in the deep pools (up to 1.7m deep) adjacent to the 2 large overhanging bank shelters in stations 1 and 3. However, it was not possible to accurately quantify these trout numbers. These deep pools may have an important influence on overwinter carrying capacity. Bjornn (1971) observed that fish in sections without rock were primarily beneath undercut banks in winter. Hunt (1971) reported dramatic increases in standing crops of wild brook trout following the installation of overhead bank cover, and attributed the increase to improved over— winter survival after habitat development. Bustard and Narver (1975a) found that when water temperature decreased to 2 C, coho and steel- head moved closer to cover, making use of logs and upturned tree roots. All four shelter types in the study area undoubtably enhance overwinter survival in the Au Sable. 97 Relationships Between Trout Population and Overhead Cover All per-station correlations between trout abundance and amount of all overhead cover were positive and low (Tables 23-25). Most correlations were not significant at (a = 0.20). Higher correlation coefficients and significance levels were obtained for trout abundance and natural overhead cover than for either man-made or total overhead cover. The only correlations significant at a = 0.05 were for natural overhead cover. There can be several interpretations of this result. One is that natural overhead cover is superior to man-made cover and has more influence on the abundance of trout in a stream section. In view of the small amount of natural overhead cover in the study section (Table 19) and the small numbers of trout captured by electrofishing or observed while snorkel diving beneath natural overhead cover, this interpretation does not appear valid. A more likely interpretation is that natural cover was correlated with some other factors which influence trout abundance such as stream surface area, channel shape, the amount of water in a station deep enough to provide sufficient cover, or some other factors. Additional evidence that natural cover is not superior to man-made cover is that more brown trout (by far the most abundant species) 1 150mm were located beneath man-made cover than in all other parts of the stream including natural overhead cover (Table 21). It was also observed that most permanent natural overhead cover was in areas of shallow water close to the banks away from principle lines of current. 98 The amount of overhead cover (area) was more highly correlated with trout abundance than was length of cover per station (Tables 23-25). This suggests that in large streams, area of overhead cover is a slightly better index of habitat suitability than length. However, the small amount of variation in per-station trout abundance explained by overhead cover in this study indicates that overhead cover is not a good index of habitat suitability in large streams in summer. Evidence of a strong positive correlation between the amount of overhead bank cover in stream sections and trout abundance has been presented by a number of investigators (White 1975, Enk 1977; Hunt 1971; Wesche 1976). These investigators examined relatively small streams with small surface areas. My study section in the Au Sable River had a mean stream width of 29m and mean maximum depth of 78.7cm. The stream was shallow along most of the banks and contained virtually no natural overhanging bank cover. Therefore, I measured the length and area of permanent overhead cover of all types in all parts of the stream which were in contact with the water, 10cm or more above the bottom and 10cm or more in width. These dimensions were chosen after making snorkel diving observations and after examining the work of Wesche (1976) who found few trout in overhead bank cover less than 9.1cm in width or in water depths less than 15.2cm. DeVore and White (1978) found that brown trout 25-30cm in length preferred over- head cover 10cm rather than 15 or 20cm above the streambed. Snorkel diving observations in the Au Sable showed that trout did utilize overhead cover which about 10cm above the stream bottom when deeper 99 water was flowing over or around the cover. Basset (1978) states that distance between overhead cover and the stream bed should just exceed the body depth of the largest trout likely to use the cover. Since stations 1 and 3 contained unwadable and unshockable pools they were excluded from the analysis of trout abundance and cover per station. There was little variation in the amount of overhead cover in the reamining 8 stations (Table 19). This contrasts with Enk (1977) and Hunt (1971) where the amount of overhead cover varied more between stream sections. Trout abundance varied little in the 8 stations used in the analysis (Tables 11, 12, 14-17). Most correlation coef- ficients were not significant, owing to low variability in trout abun- dance and overhead cover among stations and the small number of stations available for correlation analysis. Overhead shelter covered only about 2% of the surface area of the stream. This may be too small a fraction of the suitable living space to have greatly influenced per- station trout abundance. In summary, the amount of per-station trout abundance was positively correlated with overhead cover, although the relationship was not nearly as strong in this large stream as the correlations usually obtained for permanent overhead bank cover and trout abundance in small streams. Trout abundance was slightly more correlated with area of overhead cover than with length of cover per station. Percentage of Trout Beneath Man-Made Shelters More than 50% of the total number of brown trout 3 150mm.were beneath man-made cover (Table 21) which covered only about 1.8% of the 100 stream surface area. The percentage of trout beneath man-made shelters increased progressively with trout size. Basset (1978) observed that brown trout in an experimental stream spent most of their time from sunrise to sunset under cover. Since electrofishing was done during the day it is not surprising that most fish were beneath cover. Although special care was taken to approach man-made shelters (see Methods) with electrodes out of the water some trout could have been chased into shelter from other stream areas as the electrofishing crew approached. Even if many fish were driven into cover, the results demonstrate that man-made shelters provide areas of refuge for dis- turbed fish. After all man-made shelters in a section were electro- fished individually the entire stream section was sampled. Most trout caught on this second sweep were also captured beneath man—made shelters. This indicates that these fish were not driven into man- made shelters until they were frightened there by electricity. I did not believe it was feasible to place a diver near shelters to count trout chased into them by the approaching crew because of poor under- water visibility and the potentially disruptive influence of the diver. Trout are probably not as frightened by waders in large stream as in small ones. The heavy recreational use of the Au Sable by fishermen and canoeists may serve to acclimate the trout to disturbances. ,In summary, most larger trout (3 150mm) were beneath man—made overhead cover during the day. This result is consistent with the findings of other investigators on the photonegative and cover seeking behavior of trout (Gruber 1978; Basset 1978; Butler and Hawthorne 1968; Baldes and Vincent 1969). Population Estimates Trout Abundance Trout were more abundant in the study section (kg/hectare) than in most other Michigan streams (Table 18). This may be due to a combination of factors including greater mean water depth in the Au Sable River than in some other streams, restrictive angling regulations, or food abundance. Shetter and Alexander (1966) found that special regulations on the North Branch of the Au Sable caused a temporary stockpiling of trout during a given fishing season, but nonangling mortality between seasons lowered the gains to normal levels. Latta (1973) concluded that flies-only regulations alone, which reduce hooking mortality, do not lead to an increase in the standing crOp of fish. The high population in this area of the Au Sable may be derived from high size limits and reduced creel limits. An examination of Table 18 shows that the trout standing crap from Burtons Landing to Wakeley Bridge (an area encompassing the study section) given by COOpes (1974) is more than twice (kg/hectare) the standing crop in the study section in 1978. This may reflect lower habitat suitability in the study section than in other parts of the stream, a reduction in invertebrate production since the previous estimates were made, increased angler harvest, or other factors. The Grayling sewage disposal plant ceased discharging into the Au Sable in 1971 (Hendrickson and Doonan 1974). The nutrient rich discharge may have served to stimulate the production of invertebrates used by trout and forage fish. Warren et a1 (1964) reported that trout 101 102 production and biomass increased 700 percent in sucrose enriched sections of Berry Creek. Estimation Procedure Population estimates made using Schaefer's modification of the Peterson method are biased less than 2% when the product of the marking and census runs is approximately equal to 4 times the size of the population (Robson and Regier 1964). This criterion was met and exceeded for the combined species population estimates made for this study. Another assumption on which mark and recapture estimates are based is that the capture method does not measurably affect the re- capture of fish taken on the census run (Cooper 1952). Cooper believed that changes in catchability following an electrofishing marking run were minimal. Bouck and Ball (1966) observed fish following electro- shocking and found that they fasted for about 2 weeks and were ex- tremely excitable, swimming vigorously whenever their holding tank was approached whereas seined fish exhibited normal activity. This could lower capturability in a large stream such as the Au Sable where the large expanse of water presents many escape avenues. Brook and rainbow trout abundance estimates were obtained by proration from combined brown, brook, and rainbow estimates. This method presumes that these three species were equally susceptible to capture by electrofishing which was shown by Cooper (1952). This assumption was probably valid for brook trout but not rainbow trout in this study. The relative numbers of brook and brown trout captured while electrofishing were consistent with the relative numbers observed 103 while snorkel diving in the study area. Rainbown trout on the other hand were observed by diving to be abundant in several sections of the study area, but only 9 were captured on the marking run and 3 on the census run. Most rainbow trout sighted while snorkeling were in deep pools which could not be electrofished effectively. Capturability for all trout species is low in such areas. Therefore, the rainbow trout estimates in Table 13 are undoubtedly low. By the same token, all other species abundance estimates are lpw for stations 1 and 3 because of the long deep unwadable pools there. Snorkel diving ob- servations indicated that large brown trout were also abundant in these pools but that small numbers of brook trout were present there. If the fish in deep pools of stations 1 and 3 had been adequately sampled, numbers and especially biomass estimates for the study section would be considerably higher. Application to Stream Management and Implications for Further Research This study indicated that trout preferred shelters with deep water flowing along at least one side. Shelters which serve as wing- deflector-overhead cover seem most likely to provide this kind of cover without silting in badly. Future research into the physical factors affecting per-shelter trout abundance should use a more sensitive measure of water velocity than subshelter bed material composition as used in this study. Con- siderably better explanations of per-shelter trout abundance might be possible if shelters could be accurately characterized according to velocities beneath and around them. 104 Further study could also be undertaken to determine if trout abundance in large streams such as the Au Sable increases substantially following the addition of trout shelters. Such increases are common on small streams but may be less dramatic in large rivers where the water is deep throughout most of the channel. Population data on sections of large streams before and after the installation of trout cover would provide valuable insight into the effectiveness of habitat improvement in streams such as the Au Sable River. Snorkel diving was not an effective method of determining trout abundance beneath complex log jams. However, diving observations can be used to some extent to determine the physical characteristics of cover used by trout. A diver can also measure the amount of overhead cover in a stream more accurately than someone above the surface. Snorkel diving may be an effective way to count trout in pools which are too deep to electrofish. Diving counts in pools could be compared to the numbers of trout taken by explosives or some other capture method. CONCLUSIONS 1. Water depth and length of overhead cover were the features of the study area's man-made trout shelters which were most important to trout. Trout preferred shelters with deep water flowing along at least one side with ample interstices for hiding cover. 2. More than 50% of trout 3_150mm were associated with man-made shelters. The percentages of trout associated with man-made shelters increased progressively as trout size increases. 3. There were no significant differences in the abundance of trout beneath log jam, submerged log raft, or stump type shelters. 4. Trout abundance in 100m stations was positively correlated with the amount of overhead cover per station, but the relationship was weak. The correlations of area of overhead cover per station and trout abundance usually were slightly higher than correlations of length of cover and abundance. 5. Brown trout were the most abundant trout species in the study section, comprising 95.6% of total trout biomass. Brook trout made up 3.2% of the biomass and rainbow trout accounted for 1.2%. Trout abundance estimates for the 1km section are low, especially for rain— bow trout and large brown trout, because of unwadable pools in 2 of the 100m sections. Rainbow trout were concentrated in these pools and could not be captured effectively. 105 REFERENCES CITED REFERENCES CITED Allen, K. R. 1969. Limitations on production in salmonid populations in streams. Pages 3-18 in_T. G. Northcote, ed. Symposium on Salmon and Trout in Streams, H. R. MacMillan Lectures in Fisheries, Univ. Brit. Col., Vancouver, B.C. Baldes, R. J., and R. E. Vincent. 1969. Physical parameters of micro- habitats occupied by brown trout in an experimental flume. Trans. Am. Fish. Soc. 98: 230-238. Bassett, C. E. 1978. Effect of cover and social structure on position choice by brown trout (Salmo trutta) in a stream. M. S. Thesis, Mich. State Univ., East Lansing, 181 pp. Bent, P. C. 1970. A proposed streamflow data program for Michigan. Open-file Rep., U.S. Geol. Surv., Wat. Res. Div., Lansing, Mich. Bjornn, T. C. 1971. Trout and salmon movements in two Idaho streams as related to temperature, food, streamflow, cover and pOpulation density. Trans. Am.Fish. Soc. 100: 423—438. Bouck, G. R., and R. C. Ball. 1966. Influence of capture methods on blood characteristics and mortality in the rainbow trout (Salmo gairdneri). Trans. Am. Fish. Soc. 95(2): 170-176. Boussu, M. F. 1954. Relationships between trout pOpulations and cover on a small stream. J. Wildl. Mgt. 18: 229-239. Bustard, D. R., and D. W. Narner. 1975a. Aspects of the winter ecology of juvenile coho salmon (Oncorhynchus kisutch) and steelhead trout (Salmo ggirdneri). J. Fish. Res. Bd. Can. 32:667-680. Bustard, D. R., and D. W. Narner. 1975b. Preferences of juvenile coho salmon (Oncorhynchus kisutch) and cutthroat trout (Salmo clarki) relative to simulated alterations of winter habitat. J. Fish. Res. Bd. Can. 32: 681-687. Butler. R. L., and V. M. Hawthorne. 1968. The reactions of dominant trout to changes in overhead artificial cover. Trans. Am. Fish. Soc. 97: 37-41. 107 108 Chapman, D. W., and T. C. Bjornn. 1969. Distribution of salmonids in streams with special reference to food and feeding. Pages 153- 176 i§_T. G. Northcote, ed. Symposium on Salmon and Trout in Streams. H. R. MacMillan Lectures in Fisheries, Univ. Brit. Col., Vancouver, B.C. Cooper, E. L. 1952. Rate of exploitation of wild eastern brook trout and brown trout populations in the Pigeon River, Otsego County, Michigan. Trans. Am. Fish. Soc. 81: 224-234. Coopes, G. F. 1974. Au Sable River watershed project biological report (1971-1973). Fisheries Management Report No. 7. Michi- gan Dept. Nat. Res., Lansing. 296 pp. DeVore, P. W. 1975. Daytime behavioral responses of adult brown trout (Salmo trutta) to cover stimuli in stream channels. M.S. Thesis Mich. State Univ., East Lansing. 38 pp. DeVore, P.1L, and R. J. White. 1978. Daytime responses of brown trout (Salmo trutta) to cover stimuli in stream channels. Trans. Am. Fish. Soc. 107(6): 763—771. Ellis, D. V. 1961. Diving and photographic techniques for observing and recording salmon activities. J. Fish. Res. Bd. Can. 18(6): 1159-1166. Enk, M. D. 1977. Instream overhead bank cover and trout abundance in two Michigan streams. M. S. Thesis, Mich. State Univ., East Lansing. 127 pp. Everest, F. E., and D. W. Chapman. 1972. Habitat selection and spatial interaction by juvenile chinook salmon and steelhead trout in two Idaho streams. J. Fish. Res. Bd. Can. 29(1): 91-100. Fausch, K. D. 1978. Competition between brook and brown trout for resting positions in a stream. M. S. Thesis Mich. State Univ., East Lansing. 100 pp. Gibson, R. J., and M. H. Keenleyside. 1966. Responses to light of young Atlantic salmon (Salmo salar) and brook trout (Salvelinus fontinalis). J. Fish. Res. Bd. Can. 23: 1007-1024. Gibson, R. J., and G. Power. 1975. Selection by brook trout (Salvelinus fontinalis) and juvenile Atlantic salmon (Salmo salar) of shade related to water depth. J. Fish. Res. Bd. Can. 32: 1652-1656. Griffith, J. S. 1972. Comparative behavior and habitat utilization of brook trout and cutthroat trout in small streams in northern Idaho. J. Fish. Res. Bd. Can 29: 265-273. Gruber, J. C. 1978. Response of adult wild brown trout (Salmo trutta) to light and velocity under simulated bank cover in stream channels. M.S. Thesis. Mich. State Univ., East Lansing 29 pp. 109 Hendrickson, G. E., and C. J. Doonan. 1972. Hydrology and recreation on the cold-water rivers of Michigan's southern peninsula. U.S. Geological Surv. Water Information Series Rept. 3. 82 pp. Hendrickson, G. E., and C. J. Doonan. 1974. Reconnaissance of the upper Au Sable River, a cold-water river in the north-central part of Michigan's southern peninsula. U.S. Geological Surv. Hydrologic Investigations Atlas HA-527. Hunt, R. L. 1971. Responses of brook trout pOpulations to habitat deve10pment in Lawrence Creek. Tech.. Bull. 48. Wis. Dept. Nat. Res., Madison. 35 pp. Jenkins, T. M. Jr. 1969. Social structure, position choice and micro- distribution of two trout species (Salmo trutta and Salmo gairdneri) resident in mountain streams. Anim. Behav. Monogr. 2(2): 55-123. Keenleyside, M. H. A. 1962. Skindiving observations of Atlantic salmon and brook trout in the Miramichi River, New Brunswick. J. Fish. Res. Bd. Can. 19(4): 625-634. Kwain, W. H., and H. R. MacCrimmon. 1969. Further observations on the response of rainbow trout, Salmo gairdneri, to overhead light. J. Fish. Res. Bd. Can. 26(12): 3233-3236. Latta, W. C. 1973. The effects of a flies-only fishing regulation upon trout in the Pigeon River, Otsego County, Michigan. Fisheries Research Report No. 1807. Mich. Dept. Nat. Res., Lansing. 28 pp. Lewis, S. L. 1969. Physical factors influencing fish populations in pools of a trout stream. Trans. Am. Fish. Soc. 98: 14-19. Morisawa, M. 1968. Streams, their dynamics and morphology. McGraw- Hill. New York. 175 pp. Neter J., and W. Wasserman. 1974. Applied linear statistical models. Richard D. Irwin, Inc. Homewood, Illinois. 842 pp. Platts, W. S. 1976. Validity of methodologies to document stream environments for evaluating fishery conditions. Pages 267-284 in_J. F. Orsborn and C. H. Allman, eds. Proceedings of the Symposium and Speciality Conference on Instream Flow Needs: Volume II. Am. Fish. Soc., Bethesda, MD. Richards, J. S. 1973. Changes in fish species composition in the Au Sable River, Michigan, from the 1920's to 1972. M. S. Thesis. University of Michigan, Ann Arbor. 102 pp. Ricker, W. E. 1975. Computation and interpretation of biological statistics of fish populations. Bull. Fish. Res. Bd. Can. No. 191. 382 pp. 110 Robson, D. S., and H. A. Regier. 1964. Sample size in Peterson mark- recapture experiments. Trans. Am. Fish. Soc. 93: 215-226. Saunders, J. W., and M. W. Smith. 1962. Physical alteration of stream habitat to improve brook trout production. Trans. Am. Fish. Soc. 91: 185-188. Schmidt, W. A., and P. J. Rusz. 1974. Ecological survey of the Stranahan and Knight tracts on the Au Sable River, Michigan. Mich. State Univ. Dept. of Fisheries and Wildlife. 193 pp. Shetter, D. S., and G. R. Alexander. 1966. Angling and trout popu- lations on the North Branch of the Au Sable River, Crawford and Otsego counties Michigan, under special and normal regu- lations, 1958-63. Trans. Am. Fish. Soc. 95(1): 85-91. Stewart, P. A. 1970. Physical factors influencing trout density in a small stream. Ph.D. Dissertation. Colorado State Univ. Fort Collins. 88 pp. In_Dissertation Abstracts International 31(7): 4401-B. Abstract no. 71-2432. Vincent, R. E. 1969. The tolerance of water velocity by trout as a basis for enhancement of the stream fishery. Proc. West. Assoc. State Game Fish Comm. no. 49: 188-190. Warren, C. E., J. H. Wales, G. E. Davis, and P. Doudoroff. 1964. Trout production in an experimental stream enriched with sucrose. J. Wildl. Mgt. 28(4): 617-660. Wesche, T. A. 1976. Development and application of a trout cover rating system for IFN determinations. Pages 224-234 in_J. F. Orsborn and C. H. Allman, eds. Proceedings of the Symposium and Speciality Conference on Instream Flow Needs: Volume II. American Fisheries Society, Bethesda, MD. White, R. J. 1975. Trout papulation responses to streamflow fluctu- ation and habitat management in Big Roche-a—Cri Creek, Wisconsin. Verh. Internat. Verein. Limnol. 19: 2469-2477. APPENDIX 112 Table A1. Streamflow discharge and flow characteristics for the Main- stream of the Au Sable River at Stephans Bridge. (site shown in Figure 2). Date Discharge (m3/sec) 1978 Monthly Means October 5.30 November 5.13 December 5.30 January 4.67 February 4.45 March 4.53 April 6.91 May 5.63 June 4.67 July 4.11 August 4.11 September 4.90 Mean Annual Discharges June 5.52 July 4.93 August 4.56 September 4.76 Peak Discharges 2-year recurrence 10.76 5-year recurrence 12.74 lO-year recurrence 13.88 25-year recurrence 14.44 50-year recurrence 15.00 lOO-year recurrence 16.42 7-day Low Flows 10-year recurrence 3.54 20-year recurrence 3.40 * Data from Robert Larson, USGS, Grayling, Michigan, pers. comm. 113 N m c w Nm H o o «H om m o mm w e m m . an H H N w mH N mm H H m «H Nm H m e m 0 NH Hm N 0 NH c «H on a N o m an aN H H N m wN N N H m N NN N H N m «H 0N m N m m HH m mN N N o m N «N H N m mN NH N N a o w c we 0H m e N e HH «H nH n a nN «H N q o o N mH H H N m m N NH H N o «H HH N N w HN m an 0H m e H N m ON a m H o m mmN qu mmH qu can mmN qu mmH qu mm mooaoz IomN ICON nomH IooH noon IomN ICON IomH nooH Ion uuuaonm mommmHo muwm seem mononao «mam anon ca uooua ozoum mo nooaoz cm ozone xooum mo Hooaoz .wNmH .HN on NH NHoH .uo>Hm uHomm =< onu cm AmoomsNumo oONHMHonon wcfinmamouuooao sown oOHumuoun an mommawumov muouHosm ozone Nm sundown mmmHo mean salon mo ooouu cacao one xoono Ho Hanan: moumawuwm .N< manna 114 N N O O N ON NO N N H HH NH OH NO N H H N N O N HO O N O NN ON H N H N N N ON H H N NN N N N N N N N NN H H N N ON ON H H N O NH ON NN N N N O NH NH N ON N N O N N ON NN O N O N N NN N N HN N N N NH ON N N N H N ON NO H N H N N NO N O NH N N NO H H N ON NO N N H O NO N NH O HO OO H H N O NO NO N N N NH O NO H O NH O ON HO N H NH OH OO N N N OH N N ON N H O O NH N OH ON NON NON ONH ONH NO NOO NON NON NON NON OOH OOH ON Huaasz -ONN -OON -ONH -OOH -ON -OOO -ONN -OON -ONN -OON -ONH -OOH .ON HNOHNHN mmmmmHU QNfim 350m mmwmmHU ONHm 580m 6H USOHH. .G30Hm MO ”awn—552 .GH UQOHH. XOOHM MO HOQHBJZ H.O.HOOOO .NO NHNON 115 N O N ON O O N O N NO N N H N O NH N N NO H O N N H O OH NO H N N OO H H HH NO NO N N O NN OO NON NON NNH OOH NN NOO NON NON NON NON ONH NOH NN umaaaz uONN -OON uONH -OOH ION -OOO -ONN -OON IONN -OON uONH uOOH uON NOOHOHN mommmHo oNHm 850m mommmao oNNm seem ca unoufi caoum mo umoaoz ow unoua Hooum mo monaoz H.O.uaouv .N< «HOON “‘111111111111111