.u- v- r~.'. - ,mxm:q§ "Mfuc'rxv . m“? l~"‘1" ‘ 'flnx‘ ? . ~47 .. “Au {:.2.a.” bi: §"i(i!"9 ‘2'.;I:;5t ’ . , , a xyww .1... :32.- . 2 ;-t w" 5’“. 11”“ .'v:1$) { ‘2 . I 5, ‘ ,’§;:: §‘{?;1 i‘ " w 1§ig fl:::‘ ,. »“- 4&2: . y. z .. ‘ jfigisia. ?§‘3ff13f ; ;: u 3:5 .A a ._-. 1:.- RAIRES lllllllllllllllllllllllllllllllllllll 3 1293 010208 This is to certify that the thesis entitled Abundance, Structure and Biomass of Nearshore Zobplfinxton of Northeastern Lake Michigan presented by Glenn Lee Barner has been accepted towards fulfillment of the requirements for Master of Science degree inFisheries and Wildlife Major professor Date May 4, 1994 0.7639 MS U is an Affirmative Action/Equal Opportunity Institution LIBRARY Michigan State University PLACE u RETURN sex to man this man {tom your mom. TO AVOID FINES Mum on or More data duo. DATE DUE DATE DUE DATE DUE MSU ioAn Affinndivo ActiuVEqunl Opportunity Institution _ Was-m ABUNDANCE, STRUCTURE, AND BIOMASS 0F NEARSHORE ZOOPLANKTON 0F NORTHEASTERN LAKE MICHIGAN By Glenn Lee Earner 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 1993 ABSTRACT ABUNDANCE, STRUCTURE AND BIOMASS 0F NEARSHORE ZOOPLANKTON 0F NORTHEASTERN LAKE MICHIGAN BY Glenn Lee Barner Biotic changes to Lake Michigan and abiotic differences of the northeastern part of Lake Michigan define a need for basic research of these nearshore waters. Zooplankton were sampled at 10 and 30 meters, at four sites in northeastern Lake Michigan from July to November. Abundances in both individuals'2 and individuals'3 were measured. Principle component analysis was used to evaluate community structure to the observed variables, depth, month and site. Dry weight biomass was estimated using three methods; counting of instars, length-weight regressions, and volume estimates. The zooplankton community was dominated by Diaptomus sp., Cyclops sp., and Bosmina longirostris. Principle component analysis found site an insignificant variable, and although month was significant, species associations could only be split satisfactorily by the first principal component with the depth variable. Biomass was dominated by species that were the most abundant, except for Bytbotrepbes cederstroemi which averaged 8.4% of the biomass and only 0.1% of the abundance. ACKNOWLEDGMENTS I want to acknowledge my appreciation to all the people who have supported me throughout this project. In particular, to my major professor, Dr. Niles Kevern, who believed in my abilities and gave me this opportunity, to Dr. Alan Tessier who provided suggestions and guidance with the analysis, and to Dr. Darrell King for his valuable critique of the manuscript. I also want to thank Dr. John Lehman of the University of Michigan for his suggestions, particularly with the sampling methods and to his graduate student, Donn Branstrator for his valuable assistance with identification. I am also grateful to my fellow students who helped with the collection of data, Jay Hesse, and Jay Wesley and especially Robert Elliott whose own research project and ambition fostered this study. Thanks is also extended to the Holland Steelheaders, and the Ludington Charterboat Association, who assisted in the collection of data. My sincere gratitude to my parents and family who have nurtured me with love and support for a lifetime, including through the difficult times of college and this endeavor. And most importantly to my wife Clifena Yellowfox, who I am eternally grateful to for her encouragement, love, and support. This research was supported by grants in part from: the Fisheries Division of the Michigan Department of Natural Resources, Sport Fish Restoration Act, Study 472. Project F-SB-R, Michigan; the Michigan ii Agricultural Experiment Station; Wildlife Unlimited of Allegan and Ottawa Counties; the Native American Institute at Michigan State university; and the Ludington Charterboat Association. iii TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . v LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . vii INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 STUDY SITES . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Field Sampling . . . . . . . . . . . . . . . . . . . . . . . 9 Temperature Profiles . . . . . . . . . . . . . . . . . . . . 9 Zooplankton Groups . . . . . . . . . . . . . . . . . . . . . 10 Zooplankton Abundance . . . . . . . . .‘. . . . . . . . . . 12 Principal Component Analysis . . . . . . . . . . . . . . . . 13 Correlation Coefficients of PCA . . . . . . . 15 Multiple Regressions of PCs on Environmental Variables . . . 15 Verification of Zooplankton Environmental Trends on PC' 5 . . 16 Species Partitioning . . . . . . . . . . . . . . . . . 16 Zooplankton Dry Weight Methods . . . . . . . . . . . . . . . 16 Zooplankton Measurements . . . . . . . . . . . . . . . 18 Volume to Dry Weight Estimation . . . . . . . . . . . 18 Length- Dry Weight Regression Estimation . . . . . . . . l9 Bythotrephes Dry Weight Estimation . . . . . . . . . . 19 RESULTS and DISCUSSION . . . . . . . . . . . . . . . . . . . . . . 21 Zooplankton Abundance . . . . . . . . . . . . . . . . . . . 21 Principal Component Analysis . . . . . . . . . . . . . . . . 29 Correlation Coefficients for PC 1 . . . . . . . . . . . 31 Correlation Coefficients for PC 2 . . . . . . . . . . . 31 Regressions and ANOVA of Principal Components . . . . . . . . 31 Zooplankton Community Trends - Areal . . . . . . . . . 36 Zooplankton Community Trends - cubic . . . . . . . . . 39 Zooplankton Community Biplots . . . . . . . . . 45 Verification of Zooplankton Environmental Trends . . . . . . 48 Areal Abundance Verification . . . . . . . . . . . . . 51 Cubic Abundance Verification . . . . . . . . . . . . . 60 Verification of Correlated Species . . . . . . . . . . 66 Species Groups . . . . . . . . . . . . . . . . . . . . . 72 Areal Species Groups . . . . . . . . . . . . . . . . . 72 Cubic Species Groups . . . . . . . . . . . . . . . . . 78 iv Species Associations . . . . . . . . . . . . . . 78 The Littoral Species Association . . . . . . . . . 84 Limnetic Non- -persistent Species Association . . . . . . 84 Limnetic Persistent Species Association . . . . . . . . 8S Estimations of Dry Weight Biomass . . . . . . . . . . . . . . 86 Volume Estimations of Biomass . . . . . . . . . . . . . 86 Length- Weight Regression Estimations . . . . . . . . . 88 Bythotrephes Biomass Estimations . . . . . . . . . . . 91 Total Biomass - All Locations (mg/m2) . . . . . . . . . . . . 91 Ludington Seasonal Biomass Trends . . . . . . . . . . . . . . 97 SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . 100 Zooplankton Abundance . . . . . . . . . . . . . . . . . . . . 100 Species Associations . . . . . . . . . . . . . . . . . . . . 100 Dry Weight Biomass . . . . . . . . . . . . . . . . . 101 Grand Traverse Bay Biomass . . . . . . . . . . . . . . 101 Bythotrephes cederstroemi . . . . . . . . . . . . . . . . . . 102 Copepod Dominance . . . . . . . . . . . . . . . . . . . . . . 103 APPENDIX A. Temperature Profiles . . . . . . . . . . . . . . . . . 105 APPENDIX B. B. cederstroemi: counts of instars and temperatures (°C).109 Table Table Table Table Table Table Table Table Table Table Table Table Table 10. 11. 12. 13. LIST OF TABLES Taxonomic explanation of abbreviations of species and families used in tables and figures. . . . . . . . . . . 11 Length-dry weight regressions. . . . . . . . . . . . . . 19 Zooplankton areal abundances (#/m?) at two depths (10 and 30 m) and four sites in northeastern Lake Michigan in 1991. 22 Zooplankton cubic abundances (#/m3) at two depths (10 and 30 m) and four sites in northeastern Lake Michigan in 1991. 25 Correlation coefficients of principle components 1 - 4, using areal abundance data. . . . . . . . . . . . . . . . 32 Correlation coefficients of principle components 1 - 4, using cubic abundance data. . . . . . . . . . . . . . . . 33 Regression and ANOVA of principle components 1 and 2 using transformed areal abundance data. . . . . . . . . . . . . 34 Regression and ANOVA of principle components 3 and h using transformed areal abundance data. . . . . . . . . . . . . 35 Regression and ANOVA of principle components 1 and 2 using transformed cubic abundance data. . . . . . . . . . . . . 37 Regression and ANOVA of principle components 3 and 4 using transformed cubic abundance data. . . . . . . . . . . . . 38 Unverified estimates of mean dry weight (ug), standard error, and 95% confidence intervals of Diaptomus sp., and Cyclops sp. . . . . . . . . . . . . . . . . . . . . . . . 87 Unverified estimates of mean dry weight (ug), standard error, and 95% confidence intervals of Bosmina coregoni, and Daphnia galeata. . . . . . . . . . . . . . . . . . . . . 89 Unverified Estimates of mean dry weight (ug), standard error, and 95% confidence intervals of Daphnia retrocurva, Epischura lacustris, and Limnocalanus macrurus. . . . . . 90 Table Table Table Table 14. 15. 16. 17. vi Estimated dry weight biomass (ug) of three instars, neonates, and broken spine animals of Bythotrephes cederstroemi at two depths and four sites in northeastern Lake Michigan in 1991 . . . . . . . . . . . . . . . . . . 92 Unverified estimated biomass (mg/m?) of measured species at two depths (10 and 30 m) and four sites in Northeastern Lake Michigan in 1991. . . . . . . . . . . . . . . . . . . . . 9h Counts of three instars, neonates and broken spine animals of Bythotrephes cederstroemi at two depths and four sites in northeastern Lake Michigan in 1991. . . . . . . . . . . 109 Temperatures ('0) used to estimate dry weight biomass (ug) of three instars and neonates of Bythotrephes cederstroemi based on Burkhardt's (1991) linear regression of mean dry weight on epilimnetic temperatures in Lake Michigan. . 111 Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure 10. 11. 12. vii LIST OF FIGURES Zooplankton sampling sites on Lake Michigan in 1991. . . 7 Scree diagrams showing percent of variance explained by each component for both areal and cubic abundance data, and the cutoff lines for number of principle components analyzed. 30 Category plots by month of principal components 1 (a), and 2 (b) using areal abundance estimates. . . . . . . . . . . 40 Category plots by depth of principal components 1 (a), and 2 (b) using areal abundance estimates. . . . . . . . . . . 41 Category plot by site of principal component 3 using areal abundance estimates. . . . . . . . . . . . . . . . . . . 42 Category plots by month of principal components 1 (a), 2 (b), and 4 (c) using cubic abundance estimates. . . . . . 43 Category plots by depth of principal components 1, (a) 2 (b), and 4 (c) using cubic abundance estimates. . . . . . 44 Biplot of first two principal components using areal zooplankton abundance data from Northeast Lake Michigan in 1991 (axes are unitless). . . . . . . . . . . . . . . . . 46 Biplot of first two principal components using cubic zooplankton abundance data from Northeast Lake Michigan in 1991 (axes are unitless). . . . . . . . . . . . . . . . . 47 Illustration of the significantly correlated variables, depth and month, of principle component 1, expected to influence a species position on both the areal and cubic abundance biplots. . . . . . . . . . . . . . . . . . . . 49 Illustration of the significantly correlated variables, depth and month, of principle component 2, expected to influence a species position on the areal abundance biplotSO Biplot illustrating species that are significantly correlated with principle components 1 and 2 using areal abundance data. . . . . . . . . . . . . . . . . . . . . . 52 Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. viii Log transformed (ln) areal abundances of Daphnia galeata and D. retrocurva for the months of July through November in 1991. . . . . . . . . . . . . . . . . . . . . . . . . . . 53 Log transformed (1n) areal abundances of Daphnia galeata and D. retrocurva at two depths, 10 and 30 meters. . . . . . 55 Log transformed (1n) areal abundances of Chydorus sphaericus for the months of July through November in 1991. . . . . 56 Log transformed (1n) areal abundances of Chydorus sphaericus at two depths, 10 and 30 meters. . . . . . . . . . . . . 57 Log transformed (1n) areal abundances of Daphnia galeata and Epischura lacustris for the months of July through November in 1991. . . . . . . . . . . . . . . . . . . . . . . . . 58 Log transformed (1n) areal abundances of Daphnia galeata and Epischura lacustris at two depths, 10 and 30 meters. . . 59 Log transformed (1n) areal abundances of Daphnia retrocurva and Chydorus sphaericus for the months of July through November in 1991. . . . . . . . . . . . . . . . . . . . . 61 Log transformed (1n) areal abundances of Daphnia retrocurva and Chydorus sphaericus at two depths, 10 and 30 meters. 62 Log transformed (1n) cubic abundances of Bythotrephes cederstroemi for the months of July through November in 1991. . . . . . . . . . . . . . . . . . . . . . . . . . . 63 Log transformed (1n) cubic abundances of Bythotrephes cederstroemi at two depths, 10 and 30 meters. . . . . . . 64 Illustration of the significantly correlated variables, depth and month, of principle component 2, expected to influence a species position on the cubic abundance biplot. . . . . . . . . . . . . . . . . . . . . . . . . . 65 Log transformed (1n) cubic abundances of Daphnia retrocurva and Chydorus sphaericus for the months of July through November in 1991. . . . . . . . . . . . . . . . . . . . . 67 Log transformed (1n) cubic abundances of Daphnia retrocurva and Chydorus sphaericus at two depths, 10 and 30 meters. 68 Log transformed (1n) cubic abundances of Epischura lacustris and Daphnia galeata for the months of July through November in 1991. . . . . . . . . . . . . . . . . . . . . . . . . 69 Log transformed (1n) cubic abundances of Daphnia galeata and Epischura lacustris at two depths, 10 and 30 meters. . . 7O Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. ix Biplot illustrating species that are significantly correlated with principle components 1 and 2 using cubic abundance data. . . . . . . . . . . . . . . . . . . . . . 71 Illustration of significantly correlated species of the areal abundance biplot. Verified species and their responses to hypothetical expectations related to month and depth in the verification process (circled), and unverified species (uncircled) within proposed species associations or quadrants. . . . . . . . . . . . . . . . . . . . . . . . 73 Log transformed (1n) abundances of the significantly correlated quadrant I groups, Cyclops sp. and Diaptomus sp., for the months of July through November in 1991. . . . . 74 Log transformed (1n) abundances of the significantly correlated quadrant I groups, Cyclops sp. and Diaptomus sp., at two depths, 10 and 30 meters. . . . . . . . . . . . . 75 Log transformed (1n) abundances of the significantly correlated quadrant I species, Daphnia pulicaria and the quadrant 11 species, Leptodora kindti for the months of July through November in 1991. . . . . . . . . . . . . . . . . 76 Log transformed (1n) abundances of the significantly correlated quadrant I species, Daphnia pulicaria and the quadrant 11 species, Leptodora kindti at two depths, 10 and 30 meters. . . . . . . . . . . . . . . . . . . . . . . . 77 Illustration of significantly correlated species of the cubic abundance biplot. Verified species and their responses to hypothetical expectations related to month and depth in the verification process (circled), and unverified species (uncircled) within proposed species associations or quadrants. . . . . . . . . . . . . . . . . . . . . . . . 79 Log transformed (1n) cubic abundances of the significantly correlated quadrant I species, Leptodora kindti, Daphnia retrocurva, and Eubosmina coregoni for the months of July through November in 1991. . . . . . . . . . . . . . . . . 80 Log transformed (1n) cubic abundances of the significantly correlated quadrant I species, Leptodora kindti, Daphnia retrocurva, and Eubosmina coregoni at two depths, 10 and 30 meters. . . . . . . . . . . . . . . . . . . . . . . . . . 81 Log transformed (1n) cubic abundances of the significantly correlated quadrant 11 species, Daphnia galeata, Diaptomus sp. , and D. pulicaria for the months of July through November in 1991. . . . . . . . . . . . . . . . . 82 Figure Figure Figure Figure Figure Figure 38. 39. 40. 41. 42. 43. X Log transformed (1n) cubic abundances of the significantly correlated quadrant II species, Daphnia galeata, Diaptomus sp., and D. pulicaria at two depths, 10 and 30 meters. . 83 Comparison of Ludington total 200p1ankton cubic abundances (#/m3) between this study (1991) and Duffy's (1974). . . 99 Temperature (°C) profile at 30 meter depth at Ludington on August 16, 1991. . . . . . . . . . . . . . . . . . . . 105 Temperature (°C) profile at 30 meter depth at Ludington on September 13, 1991. . . . . . . . . . . . . . . . . . . 106 Temperature ('C) profile at 30 meter station (over 110 m of water) at West Grand Traverse Bay on August 26, 1991. . 107 Temperature (°C) profile at 30 meter station at Manitou Passage on August 27, 1991. . . . . . . . . . . . . . . 108 INTRODUCTION Biotic changes to Lake Michigan and abiotic differences of the northeastern part of Lake Michigan define a need for basic research of these nearshore waters. Abundance, community structure and biomass of northeastern Lake Michigan zooplankton are the components of this study. This study focuses on the zooplankton community of the nearshore waters (<30 meter depth) of northeastern Lake Michigan. Evans et al. (1980) examined relationships between zooplankton abundance (#/m3) with depth or season. From mid-spring to mid-autumn, zooplankton densities (#Vh?) were strongly related to depth. Maximum densities occurred between the 20 and 30 meter contours, and minimum densities between the 5 and 10 meter contours. I examined the zooplankton community at two depth contours, 30 meters and 10 meters, from July to November 1991, at four locations so as to assess the influences of these observed variables on zooplankton abundance, structure, and biomass. The nearshore waters of Lake Michigan tend towards higher productivity than those of the offshore waters because of local inputs that are not readily circulated into the deeper basins of the lake because of water movement differences. The nearshore (<20 m) waters are in a system of currents that run parallel to shore, while the offshore waters are essentially an open lake gyre system separated from the nearshore (Gannon 1972). Duffy (1975) studied the vertical distribution and abundance (#/m3) of nearshore zooplankton off of the Ludington Pumped Storage near Ludington, Michigan. Other Lake Michigan nearshore studies have focused on the more eutrophic waters of Lake Michigan. Gannon studied the horizontal distribution and abundance of Lake Michigan zooplankton in a cross-lake transect from Milwaukee, Wisconsin to Ludington, Michigan (1975, 1972) and found less distinct inshore and offshore differences off of Ludington (the primary sampling site of this study), than off of Milwaukee. Roth and Stewart (1973) studied zooplankton abundance (#/m3) and biomass (mg/m3) in a study near the of Cook Nuclear Power Plant in southeastern Lake Michigan. Abundances off of Cook were much greater than those near the Ludington Pumped Storage project (Duffy 1975). Waters warmed by the Cook Power plant may be a significant factor in the higher concentrations of zooplankton and the subsequent higher biomass found in that study. Although total phosphorus concentrations have declined since the early 1970's (Scavia et.al. 1986), it is likely that southern Lake .Michigan is still more productive than the northern portions of the lake (Roth and Stewart 1973). Point source inputs of nitrogen and phosphorus from sewage treatment plants and industrial wastes, as well as non-point sources from runoff of animal wastes and storm sewers are more likely to increase phytoplankton growth in the more densely populated southern portion of Lake Michigan, than the sparsely populated northern and northeastern parts. 3 Biologically, the Great Lakes have been under constant change ecologically for over four decades. Theintroduction of exotic species has affected the food-web, and changed the species interactions at all trophic levels, including the zooplankton community. Two invading species, the alewife (Alosa pseudoharengus) a planktivorous fish and Bythotrephes cederstroemi, a large voracious predatory cladoceran have both, more than any other species, affected the zooplankton community of Lake Michigan. Alewife can have profound effects on the size-structure of zooplankton communities (Brooks & Dodson 1965), and the alewife directly affected the composition of zooplankton by selectively feeding on the largest zooplankton (Wells 1970). The alewife is a planktivore throughout its life and is most important to nearshore populations of zooplankton because here they remain primarily a zooplanktivore. In the summer the planktivorous larvae dominate the inshore region of southeastern Lake Michigan (Nash & Geffen 1991), and smaller fish (40 m)(Rasmussen 1973). a Yellow perch (Perca flavescens) were more likely to contribute to zooplanktivory in the nearshore as a result of long residence times in the nearshore waters. Yellow perch spawn in the very shallow waters (<5 m) in mid-June and the young-of-the-year remain there throughout the summer. Yellow perch inhabit a zone from 0 to 40 meters, but concentrate their activities in 0t25 m depths (Rasmussen 1973). 'Yellow perch are more 200planktivorous when young (<10.2 cm), and increasingly eat more Mysis and Pontoporeia as they get larger. As adults, yellow perch target B. cederstroemi when abundant in both Ludington (Peterson 1993) and in northern Lake Michigan (Schneeberger 1991). The other species to greatly affect Lake Michigan zooplankton is an invertebrate predator. In 1986 the spined palearctic cladoceran, Bythotrephes cederstroemi was first detected in Lake Michigan (Lehman 1987, Evans 1988). The B. cederstroemi invasion has coincided with depressed populations of Leptodora kindti, another predatory cladoceran, and a subsequent increase of Bosmina longirostris populations in the offshore waters off of Grand Haven, Michigan (Branstrator & Lehman 1991). B. cederstroemi has also altered the Daphnia assemblage by preying on the smaller Daphnia retrocurva, and has further suppressed the already declining, larger-sized, D. pulicaria (Lehman & Caceres 1993). This study had four sampling locations. Three were in northeastern Lake Michigan: Ludington, Manitou Passage (part of Sleeping Bear Dunes National Lakeshore), and West Grand Traverse Bay. The fourth location, Holland, was in southeastern Lake Michigan. 5 The northeastern waters of Lake Michigan from Ludington to Manitou Passage (Carr 1971), and the deep protected waters of West Grand Traverse Bay (Lauff 1957), are cooler, than those of southeastern Lake Michigan. Roth and Stewart (1973) found differences in concentrations of individuals and biomass between their offshore southeastern Lake Michigan study site (11.2 km offshore at 40 m depth) and Gannon's (1972) offshore station, (16 km offshore at 60 m depth) off the coast from Milwaukee. They suggested the difference was because of the less productive waters of northern Lake Michigan. Zooplankton abundances in Lake Erie have been suggested to vary most with heat content and eutrophication (Patalas 1972), and large scale differences appear to be north-to-south gradients of abundance on lakes Ontario and Erie (Patalas 1969). This study had two major objectives, first, identify zooplankton community trends with respect to the observed variables: season (month), depth (30 or 10 m), and location. Secondly, look for possible species associations to describe trends within this community. The study utilized the estimation of areal (individuals/m2) and cubic (individuals/m?) abundances of each species, principle component analysis (PCA) of the community, and total biomass (mg/m2) to evaluate both the trends and species associations of the zooplankton community of northeastern Lake Michigan. Vertebrate planktivore populations were not assessed, but B. cederstroemi abundance was estimated and its contribution to the zooplankton structure and biomass is described. STUDY SITES Four nearshore locations were selected. Three are in northeastern Lake Michigan: Ludington, Manitou Passage, and West Grand Traverse Bay; and one is in southeastern Lake Michigan located off of Holland (Figure 1). Ludington is located in the northwestern part of the State of Michigan. Holland is approximately 140 km south of Ludington and North Manitou Island is about 145 km north of Ludington. The West Grand Traverse Bay location was about 40 km southeast of the Manitou Passage location, over the Leelanau Peninsula. The study design was to sample at the 10 meter and 30 meter contour at each location. However, because of the steep morphometry of the West Grand Traverse Bay basin, the deeper contour was taken over 110 m of water to allow for a reasonable spatial difference. The Ludington (LU) 30 m contour (43°:53':71"N longitude, 86’:31':09"W latitude) was approximately 10 km southwest of Ludington and the 10 m contour (43°:55':09"N, 86‘:27'W) site was approximately 1.25 km off shore near Butterfield Park. The Ludington sites were the primary sampling sites and were sampled eight times between July 17, and November 14, 1991 (7/17; 7/30 & 8/1; 8/16; 8/29; 9/13; 10/3; 10/18; and 11/14). The West Grand Traverse Bay (TC) 30 m site (44°:52'N, 85°:36') was about 1.75 km west off of Marion (Power) Island on the east shore and the 10 m (44‘:57'N, 85‘:36'W) was located about 4 km NNE off of Lee's Point on the west shore. iN Q 259 ”KW .... fl! Traversecity .KWM «'00 4300' MU mm 821:? scale: Kiiom eten Figure 1. Zooplankton sampling sites on Lake Michigan in 1991. 8 Both the Manitou Passage (MP) 30 m (45°:04':50"N, 85°:58':39"W) and 10 m (45':04':50'N, 85':58':61"W) locations were approximately 25 km NW of Leland. And both were within 1.25 km of the NE side of North Manitou Island. Each of the West Grand Traverse Bay, and Manitou Passage sites were sampled on two dates in August (8/6 and 8/26) and (8/9 & 8/10 and 8/27) respectively. The Holland (H0) 30 m (42'46'N,86'l4'30') was approximately 2 km offshore of Holland and the 10 m (42‘46'N,86'13'20”) was about 1 km from shore. Both were sampled only once (8/28). METHODS We Vertical tows were taken using a one meter mouth diameter, 153 um mesh Puget Sound closing net (Research Nets, Bothell WA) with a 5:1 length to mouth ratio. The polyethylene cod end had approximately a 5 x 8 cm window of the same mesh. The net has a semi-permeable collar slightly greater than one meter, designed to increase efficiencies by reducing backflow. The towing mechanism was a hydraulically powered drum mounted on the boat deck of the RV Smolt. The net was lowered to the desired depth, just above the sediments (about 1 m). At least a five second delay was allowed (and if necessary, the boat would be repositioned directly above the net) before the net was towed vertically at approximately 0.5 m/sec. After each of the three replicate tows, samples were put in sample jars, preserved with formalin (5-10% by volume), and labeled. Sampling was mostly during daylight hours, except for July 17, 1991, when sampling took place in the evening before midnight. Patalas (1969) found the capture of zooplankton in Lake Ontario to be unaffected by time of day of sampling. One possible exception was Limnocalanus macrurus. which may be reduced in number when sampled in the daylight hours (Balcer et al. 1984). W In August, water temperature profiles (see Appendix A) were taken at each location (except Holland). Temperatures were taken at the surface and at one foot intervals using a thermistor to a depth of about 10 15 meters. Profiles were again taken at Ludington in September, but to a depth of 30.5 meters. Surface temperatures were taken at almost every sampling location throughout the study using a hull mounted thermistor. Surface temperatures detected no upwellings in northeastern Lake Michigan in 1991. Upwellings are caused by sustained winds either from the north or south (Carr 1971). Upwellings cause the colder hypolimnetic waters to surface. An upwelling causes summer epilimnetic temperatures to drop. In 1954, a large upwelling covering several hundred miles of eastern Lake Michigan, dropped surface temperatures from 22°C to only S‘C in less than 3 days (Carr 1971). Such changes in temperature will affect zooplankton abundance, growth, and biomass. WW Fifteen species/groups were counted and converted to numerical abundances. A taxonomic explanation of the abbreviations used in the tables and figures is in Table 1. The rarest species were not estimated and include: Holopedium gibberum, Polyphemus pediculus, Pontoporeia hoyi, Latona setifera, and a harpactacoid copepod, probably Ganthocamptus sp. The original intent of this study allowed for the grouping of the many species of both Diaptomus sp. and Cyclops sp. Because of these groupings, the dominance by these species groups for the duration of the season may conceal the dominance of one species. The groupings may also conceal a dominance by several species of each grouping, each at different times of the season. The Diaptomus genus, has been split into two genera Leptodiaptomus and Skistodiaptomus (Balcer et a1. 1984), but will be reported here by 11 Table 1 Taxonomic explanation of abbreviations of species and families used in tables and figures. CLADOCERANS Abbreviation Species Family BYT ' Bythotrephes cederstroemi BOS Bosmina iongirostris GAL Daphnia galeata mendotae RET D. retrocurva PUL D. pulicaria EUB Eubosmina corego ni LEP Leptodora kindti APH Diaphanosoma s p. SPH Chydorus sphaericus CHY Chydoridae (excl udlng C. sphaericus) CYCLOPOID OOPEPODS 0Y0 Cyclops sp. . Cyclopidae OALANOID OOPEPODS EPI Epischura lacustris LIM Limnocalanus macrurus EUR Eurytemora afiinls DIA Diaptomus sp. Diaptomldae 12 the conventional names used by other Great Lakes researchers. In Lake Michigan the genus Diaptomus includes up to five possible species: D. ashlandl, D. minutus, D. sicilis, D. siciloides, and D. oreganensis. The cyclops sp. group may include the species: Diacyclops thamasl, Acanthacyelops vernalis, Mesacyclaps edax, and Tropocyclaps prasinus mexicanus. Except for one species of the Chydoridae family, Chydorus sphaericus, this family was rare and grouped. The remaining Chydaridae family was dominated by members of the Alana genus, and included the i Alonella and Acraperus genera as well. Species identified included: 3 Alana rectangula, Alana intermedia, Alana guttata, Alana costata, Alanapsis aureala, and Acroperus harpae.The remaining groups in Table l are all species and not groups of species. Numbers of zooplankton are reported both in individuals/m3, and individuals/m3. Density abundances are used to facilitate comparisons of total population sizes at sites of differing water depth. Areal abundances are reported because it is assumed that the distribution of zooplankton are primarily in the metalimnion and epilimnion after the establishment of a thermocline. Thermoclines did develope at both Ludington and West Grand Traverse Bay by mid-August. Mean values estimated by dividing these areal abundances by depth will severely underestimate peak densities, and may report a density that does not exist (Lehman 1991). Wang; Abundances as both individuals'2 (#/m2) and individuals'3 (#/m3) were measured for use in both the principle component analysis and the biomass estimates. 13 Previous to subsampling, B. cederstroemi were counted and removed from samples for exact abundances and biomass estimates. In cases where zooplankton density was high, a Folsom plankton splitter was used. Further sub-sampling was done by first bringing the sample up to a known volume, then drawing off a smaller sub-sample using a transfer macro-pipette (1-5 ml). Each numerous group of animals (in most cases, Diaptomus sp., Cyclops sp., and B. longirostris) were counted until at least 100 animals were enumerated. Additional subsamples were taken so that approximately 800-1000 animals were examined, to count the rare taxa. Samples were enumerated using a plexiglas counting tray, to allow the manipulation of the animals to identify them without dissection. A binocular microscope with a zoom lens (magnification 1-7x), with 15x ocular lenses was used throughout the study. Identification of zooplankton was made primarily according to Balcer et al. (1984). Brooks (1957) was referenced in the identification of Daphnia to species. Brooks (1959) was used for the identification to species of Chydoridae and other rare zooplankton. Chydoridae were mounted on slides in glycerin and identified to species using a compound microscope. W Description of the zooplankton community was be done by the multivariate statistical technique, principal component analysis (PCA). Pielou (1984), describes PCA as revealing the "real pattern“, of the joint responses of groups of species to persistent features of the environment, separating out the "capricious”, unrelated responses of a few individual members of a few species to environmental accidents of 14 the sort that occur sporadically (such as upwellings in the Great Lakes), and have only local and temporary effects. The purpose of performing a PCA is to reduce the dimensionality of the data matrix, and find the structure of the matrix, without the sacrifice of eliminating data from the matrix. Structure is here defined as, any systematic pattern that would indicate that species tended to occur together, or that the sampling units, when appropriately arranged, would exhibit a gradual, continuous trend in their species compositions. The chief consequence of a PCA, is the first principle axis is so oriented to make the variance of the first principle component scores as great as possible (and the second PC, second greatest)(Pielou 1984). Analysis was carried out using the statistical package, SYSTAT 5.02 (Wilkinson 1990). An introductory description of this technique can be found in Sprules (1977), Pielou (1984), and Ludwig & Reynolds (1988). The matrix of covariances between species was calculated, thus standardizing the normalized abundances by subtraction of species transformed means from the transformed abundances. This preserves the variability of individual species. PCA summarizes the information in the covariance matrix in terms of new components, a small number of which account for most of the variation. From this simplification one can formulate hypotheses about the causes of variation in the system. This reduces the species-quadrat matrix to a species-component matrix (Sprules 1977). Each component, a linear compound of the transformed proportionate abundances, has an associated eigenvalue giving the amount of variation in species accounted for by each component of the matrix, and eigenvectors (not reported) of component coefficients giving the 15 weighting of each species in the linear compound (Sprules 1977). This study had 76 separate samples (quadrats) with abundances (#/m?) of 15 species, each with the three observed environmental variables of date (month), location, and depth (30m or 10m). Each individual sample is one of the n data points (76) each with s (15 species) coordinates. Each coordinate of each point is a function of the species abundance (#/m?) in the quadrat represented by that point. This produces a 15- dimensional plot, or ”swarm“ of data points (Pielou 1984). Plotted on the 15 coordinates, the PCA aligns and centers the first principle component (PC) axis where the largest variation occurs. Before analysis the fifteen species abundances (#/m@ and #/m9) were first transformed by ln(#/mn+1). This transformation was used to decrease the dependence of taxa variance on taxa abundance. The transformation allowed rare and abundant taxa to be potentially equal in importance in the analysis while still retaining numerical differences in abundances. Use of the variance-covariance matrix was possible because all abundances had the same measurement units (Pielou 1984). Go rel on o c en 3 a PC Correlations between principle components (1-4) and the transformed species abundances, species (1-15) were calculated to assist with the interpretation of the PCs. A species was positively correlated with a principle component if the coefficient was >.230, or negatively correlated if the coefficient was < -.230 (df-70, alpha - 0.05). To determine any environmental interpretation of observed variables, multiple linear regressions of P68 1 through 4 on month, depth, and site, were performed to determine if there was a hypothetical 16 trend with these variables and the zooplankton community along the PC ordination axes (Ludwig & Reynolds 1988). WWW Transformed abundances of species positively and negatively correlated to both PC 1 and PC 2 were plotted by each significant variable to verify if the species abundance responds in the manner as hypothesized by the regressions and correlations. We There are numerous methods for classifying ecological data. The type used here is classified as an ordination-space partitioning method, and is adequate for most, if not all ecological applications (Pielou 1984). Lefkovitch's partitioning method is the most straightforward method for divisively splitting the PCA ordination (Pielou 1984). The data are first ordinated using principle component analysis, then split divisively into positive and negative groups on the first then the second principle component (Pielou 1984). The result of using only the first two PC's, is a two-dimensional partitioning of species into each of the four quadrants of the two axis plot, also called a biplot. The initial grouping of species into associations was done only for species that were found to be significantly correlated with both components, either positively or negatively on the areal, or cubic biplot. BMW It is emphasized that the biomass estimates of this study were not verified. The animals were not weighed, and the accuracy of the estimates are unknown. Therefore the following dry weight estimates are only relative estimates within this study, and can not be applied to 17 other zooplankton populations. Aliquots of replicates were mixed and a small subsample was taken for measurement. Measurements of taxon were performed for each month, depth, and location. To reduce the number of measurements of samples, when a sampling occurred within the same month, and at the same location and depth, the measurements of one sample (a combination of three replicates) were applied to taxon within that same month (months began and ended mid-month). Measurement of a species was not done when abundances were low. Dry weight biomass was estimated only for the most numerous groups found during counting; Diaptomus sp., Cyclops sp., B. longirostris, Daphnia galeata, and Epischura lacustris. Three other species, L. macrurus, D. retrocurva, and B. cederstroemi, whose average abundance was less than one percent of all species/groups over the year, were measured because of their relatively larger size. Dry weight estimation methods followed three courses of measurement: the counting of instars, length-dry weight regressions, and a volume to dry weight method. A length-dry weight regression already exists for B. cederstroemi (Carton and Berg 1990), but Burkhardt (1991) developed linear regressions of mean dry weights of each of the three instars of B. cederstroemi on epilimnetic temperature. Species specific length-dry weight regressions exist for B. longirostris, D. galeata, D. retrocurva, Epischura lacustris, and L. macrurus and.were used. The volume to dry weight method was used for Diaptomus sp. and Cyclops sp. because these were groupings of many species. 18 WW Zooplankton dimensions were measured using an ocular micrometer to within 0.02 mm (20 um). Individuals in a sample were measured consecutively as encountered to obtain a random sampling of the population. Those animals that were obviously contorted were not measured. Copepod length measurements were made from the furthest projection of the head to the insertion of the caudal setae. Lengths of B. longirastrls were measured from the furthest projection of the head process to the point of insertion of the caudal spine. The cyclomorphotic D. retrocurva and D. galeata were measured from the eye to the base of the tail spine, or standard length, (Lawrence et a1.,1987). e o D Wei ht Est tio Lawrence et a1.(l987) developed estimates of biomass of zooplankton using geometric formulas, and conversion factors to convert volume to dry weight. Diaptomus sp. and Cyclops sp. were measured volumetrically, because they are groupings of several congeneric species. Volume estimates were made by measuring length, width and depth and entering these measurements into formulas, where a-0.5 length, b-0.5 width, end c-0.5 depth: Cyclops sp. volume- (4/3)abc*pi Diaptomus sp. volume- (4/3)ab(l.25c)*pi The volumes were then converted to dry weight by multiplying the average volume of a combined sample by 0.07 (Lawrence et al.1987). Ninety five percent confidence intervals of the volume measured 19 species, Diaptomus sp. and CyclOps sp., were calculated using the dry weight values (0.07*volume). - o 3 io Species specific length-dry weight regressions (Table 2) used were from: Culver et a1. (1985), for B. longirostris and both daphnids, D. galeata and D. retrocurva; Conway (1977), for L. macrurus; and Lawrence et a1. (1987), for E. lacustris. Biomass estimates from length-dry weight regressions followed the guidelines of Bird and Prairie (1985), except for the use of the bootstrap variance. The bootstrap was performed on 30 dry weights of the most variable species Diaptomus sp., and the resulting confidence interval was nearly symmetrical and differed from both upper and lower limits by less than 0.005 mm. It was decided that the variances using the estimated dry weights would suffice. Table 2 Length-Dry Weight Regressions I Species L-W Regression (W in ug) Range (mm) r2 B. longirostris Ln We 2.8756+(2.2291LnL) (.22-.43) .98 D. galeata Ln We 2.3917+(l.5202LnL) (.36-l.81) .99 D. retrocurva Ln We 2.0213+(2.7552LnL) (.33-1.45) .99 L. macrurus Log1o - (0.98L) - 0.79 (1.10-2.83) .70 E. lacustris Ln We 1.4670+(2.4741LnL) (1.43-2.04) .86 Wigs Burkhardt (1991) developed a linear regression of mean dry weight of the three instars (TH-.927, ré-.883, r3-.930) of B. cederstroemi on 20 epilimnetic temperatures in Lake Michigan (1990 and 1991). The instars are distinguished by the number of lateral spines extending from the caudal spine. Neonates have no lateral spines, and none were found after August. Average surface temperatures of this study in July and August 1991 ranged from l9.2°C to 21.1°C, and an approximation of 70 ug was used from Burkhardt's (1991) data for neonates at 18°C . Counts of instars and neonates were done in entirety on every sample to reduce sampling bias. Instars with broken spines were multiplied by the average of the three regression values in order to salvage some estimate of biomass of these animals. RESULTS and DISCUSSION maximum Abundances are reported as both individuals'2 or areal abundances (Table '3) and as individuals’3 or cubic abundances (Table 4). Each reporting method has it's advantages. Areal abundances allow for the detection of peak abundances, and density abundances allow for direct comparisons of population abundances at different contours. The density abundance is appropriate when the water column lacks a hypolimnion, because hypolimnetic waters may have low numbers of zooplankton that would artificially lower the zooplankton abundances. The primary sampling location, Ludington, and West Grand Traverse Bay each developed a hypolimnion in August and cubic abundances from that time on should be considered cautiously. Abundances calculated as #/m?‘were dominated by the four taxon groups; Diaptomus sp., Cyclops sp., Bosmina longirastrls, and Daphnia galeata (Table 3). Average areal abundances over the entire sampling season were as follows: Diaptomus sp. (61.1%), Cyclops sp. (21.8%), B. longirostris (8.5%), and D. galeata (5.7%). Less abundant species areal abundances were, E. lacustris (1.5%), D. retrocurva (0.7%), and B. cederstroemi (0.1%), and the remaining species account for the rest (0.6%). Total zooplankton areal abundance from a single sample at the 30 meter contour peaked on August 29, 1991 for Ludington (>469,000/m2: 15,645/m3), on August 27 for Manitou Passage (>931,000/m2: 31034/m3), and on August 26 for West Grand Traverse Bay (>1,222,000/m?: 40742/m3) 21 22 0 3 mm 0 0 mm «9 0mm 0 0 83 308 F300 000$ 09 2 33:8 3.. 0 0 3 0 0 0 «0: 8v 0 30 «m8 800 0300 33m 03 2 33:8 3.. 0 0 N3 0 0 .0 .30 8 0 NF 80 F30 030m «83 FFF 3 33:8 3.. ,. ...0. 0 m8 30F 0 N00 003 N8 0 3033 003 0083 080 «8 00 33:8 3.. . ,0 0 0. .0: , 0 3 FE 0 0 «.800 800. 839 3an Km 00 33:8 3.. .0. ,0 0 88 .0. 3e 8: 30 0 088 030 300mm 30F N3 00 33:8 3... 0 0 0 0 0 0 vow 0 0 F2... 08 03.0— 88 m 9 330:8 as. 0 0 0 0 9 0 N9 0 0 300 03F 80mm 38 0 9 330:8 as. . .0 0 0 00 .0 00 8 0 0 300 9: 830 30m m 9 330:8 as. . .0 0 0 83 .0 3 30m 0 8 3:. 8mm 300: 080— 80 00 33:8 02 H . .0... .o . c we: .9. we «8 o as 88 83 $82 8a. m: 8 3828 u: .8. 0 0 89. .3 .3 82 0 0 080 83 0030. 0.33 08 00 33:8 0.2 0 8 0 0 0 0F 3 0 0 m8 58F 030 03F 0 9 330:8 0.. 0 8 0m 0 0 9 0n 0 0 8 RS 8an F3 3 9 33030 0... . .0 NF 0 0 0 0 3 0 0 0.3 «09 88 E0 0:. 9 330:8 o... :50 0 0 .0 0 we 02 0 0 08F 8.2 v8.3 8F 08 00 330:8 0.... .o o o o . o e mas o 8 3F .8 «85 sum. as 8 58:8 0F .. .0 0 0 0 0 NF 88 am 0 83 83. C88 0808 mi. 00 530:8 o... 0:. 02 mm 0 0h 3m 80 0 mm 30 033 82m 03: mm 0— 3:28 3.. 0 0m «9 0 cm «0— 80 0 Fe 08 339 000mm 080 3 9 3:330 3.. . 0 .3 we 0 0 0 3m 0 3 30 88F FKFF 38 «N 9 3:330 3.. 7... .0... .0 0 33, 0. 0 .300 0. .38 0008. 80.0. 33.2 33... 0F 00 330:3 3.. o as e o .o. ..o to 88 use 3.8 ~38 3: cause 83.... e... 8 E38 3 0.0 0 0 mm; 0_ 53 RF... 0 «33 033 3.: «830 83.: 0F 00 33:3 3.. 0 03 8m 0: 02 0 3 0 mm 03 ~03 ~03 88v 88¢ 30. 2 33:3 3.. 0 03F 99 N0 N0 0 80 8 0 00¢ 03 003 ~me0 83 3F 2 33:3 3.. .0 .. 0F. 03 N0 0 0 09 0 0 0:. 03 80v 880 88.: 09 2 33:3 3.. .0 0.. 0 «we «we 0 «0:. mm... .80 300 038 2.3. 0380 0%.: 00:. 00 F0:.:..F.::.0 3.. .0. 0.. 0 0 30F .39 3F «3 03 88 003~ 33 380 039 03 00 55:3 3.. .. .0 03 $0.. 0 3.. v3 0 08. 030:. 39 0300; 808 08. 00 33:3 3.. «.3013 >10 was: 5.... am... 03 Km .50 Ex 3:0 mOm <3 o>o ._.>m N MES 00.. .52 c. 80.8.2 9.5 8.8080: 5 «ago 52 new 0: on ace 0: Bacon 05 E E3: canvases .090 coicoEOON .0 2nd... 23 ~89. o o o 0 mm 2: o 9.0 o 0- :2 mm: 308 0.85 0 o o o 0 8 o ~0 o 9. 03. 3F ~§~ 85 o o o o o 2 0 F.~_ 0F 9 ~Fm 0:. 05: $82 0 o. o o 0 3... o o 0 Re 08: :3, F35 ~F~0- o. o, o 0 F8~ E. o o 0 0.8 ~32 8~F 8&3 FFFB~ o o... 0 0 one o o o 0 02 ~82 08°F 9%... va~F 0 we. 80 o o E 3 FF 0 o o ~3~ ~o~m F008 0 08 8. FF 0 an o 8 0 ~F.F o 8% 83. 80:- o o. o F.~ o -F o o o -F o 28. 98 8&8 0. 0 o 2. o 8~ w$~ F3... o 3.2 E... -~m~ 382 0.8.8.... 0 .0. o 0. o 29 0 £~F 22 ~88 80... 30% 20.08 8.33 . o 0 o 0 o o o as o «9.8 $3 ~83 FEE F8F~ o 0 o o 0 8 o F~F o 8 0 ~00 F32 $8.. 0 a 8 - o 8. o m-~ o 9... a. :8 20.3 302. o s. ~0. F~ 0 ~.. 0 ~00 o 8 _~ 33 88F 2.83 c o o oF~o 8.. F30 $~F 0 83m 8; :25 $82 $0.30 0 o o o o 03. :5 ma: 0 F8- ~F~o~ F38 FF0-F 385 o o 0 Q... 0 $3 88 Fm~ o 088 33 ~88 3.8: ms; 0 FF 0 o 0 FF 0 ~8~ 0 8~ 39. .56 ~83 03$ 0 o o o o ~F~ o 88 0 ~F~ 83. ~88 F8§ 33F 0 mm~ o o x: o o o 0 03 3.9. 0%: :30 828 .0. .0. o o 0... o~m o o~3~ o 08 SE 028 258 «0.8% o. .0. o o o «FF 9F 88.. o 0 805 33 F805 08.me .0 .0. o o o 0 o 3.x 0 0 208 ~83 3&8 m-~¢~ o o 0 o o o o o o o is 880 .59 828 o o o 0 o 9F o 89 9F 0 83 08:. 838 ~83 .0 .o o o o o o o o o ~F0~ $08 $89 . . .0 o 0 .0. 83 8: .88 o 0 £22 ~23 283. o 0 0. .0 .0 o 08F $3. 0 F088 0808 $05. 0 o o o 0 $8 $2. o 80.9 mB-F 3.83 59 .000 0% >00 0% 00. 0&0 000 E .0". mm 0.0 000 <5 88 0 2 53:8 83. F 2 F238 §~ F. 9 59:8 88F. FF F200 0~ 09.09 9 83 Fm 3% 8 ~F.F.~ Fm omemF mm ~me0~ .8 380 .2- «~00. 8 SB 8 -~e 9. 3~$ Fm 0080 FF 880 2. ~88 a may. 3 -Fm~ FF $82 8 meEF 8 9.32 E 038 - 3.9 - 22: 8 0509 m... 0.00m? .3. F039 8 96 FE 60:80 0228. 82 2 5:2: 8. 2 53:. .wm‘E & J 33.3 Factor 33:0— 24 53:9 6.503 dais. 25 m o 2 mm o o as 88 38:8 m o 9 S c on «8 88 3.. o, 8 h o. 9 8 NE 52 . mu 2, To . 8mm ,3. J or o n3 mm, .9 V R. .9 .m3N._. mm? o o o E. 8 o o o 3v 8* o o 0 as E ....o. .o m 9: a _ c; m; 8 «a, a. o. b 2: m; .o. o 0 mm «NF v o o m as N o F a 2: o o o mm 8 . no o .0 mm a.“ o F o mm an. 8 o m 3 as, N P 0 ¢ mm :8 a F o m 8 3n: . ac o 5 v3 2 .. .. yo .8 mm 80 .8 a; . Ho. o mm mm: ,3... has: m 3, 8 am 5:28 o 3 5 3 .228 o 3 3:28 .52 5 595:2 9.3 Eofiwofioc E was» .52 new “E on tau 0: méamu 92 E A 0:53 occaucanw 05:0 couEquoN .v 2%... 26 o— 53200 D.— o_. 33:60 D... 2 53:8 3 on ,. vaQmO: D.— ,..om..rumPPmo. .3 cggfiigfia 0.. PQQOo 3.. 9 «ngo 3.. e. . Q88 3 mm P6P 02 moon vm can «9 39 6666600 0‘) ID gig-3.}. in In (0 i. ox. «Po :8 58. 88 8P m8, 88, 2P. 2P P8 88... or 300000000100 0 o m m N o N m mv 9 5m ovmw * 8: 8m 82, .88 08 e8 88 88 8: 85 88 88 8. 8P. 8; 88 Pa 88 8m 83: 8 «8 8: Pam Pa 88 88. 88P. 88 8:, 8m: 88 .. .88 «8.8 P8P 88 REP 89 at. P89 P8 .88 88P 88. :8. P8: 3me amt . . .83an P. 0.. mom «mm o umm .5va mg a oo— .o a; F qo¢cobooooooooooo coo o 060‘ng 6 00.6760 0 o o 900 o 6 38° 6 o :0 ..t6 N... H8 co. w .. ..... :8P .P...P..P 3MP. 88.. . ,,.m8am .59 PP.” :8 2.6 1% .2... am: an... am, .5; mm .5 m8 (5 debooéEéOooodévooOOOoombboooo ODOOOOOOOOOOQONNMOOONNNOOOOCOO 6ombooomoooo 'MOOO‘omoo ‘ooooo mm , Egg or” we N c ”mflhmmflNEFFOFdfvmm?¥rwmmfiwéOOw 060 o o o o o 6.803 P. 2%» 27 9 3:2: 3.. o— 5:2: 3.. «23¢? 33:2 3E8“ «29¢ 28 (Table 3). Maximum abundance at the 10 meter contour was on the same dates in late August (27 and 26 respectively) for Manitou Passage (>244,000/m3), and West Grand Traverse Bay (>242,000/m?), while at Ludington the maximum 10 meter contour abundance was on October 18th (>335,000/m2: 33,536/m3) (Table 3) . Obvious differences in areal abundances between the two depths was apparent. For most species, abundance was greater at the 30 meter contour. Much of the difference may be a result of the greater volume of water at the 30 m contour. Areal abundance at the deeper stations was greatest for all three daphnids: D. pulicaria (77 X), D. retrocurva (43 X), and D. galeata (20 X), and a large calanoid copepod, L. macrurus (21 X). The daphnid and L. macrurus estimates are much greater at the deeper contour than can be explained alone by three times as much volume. The greater abundance of daphnids at the 30 meter contour than at the 10 meter may be caused by increased predation in shallower waters. Daphnia sp. are some of the largest opaque zooplankton found in the Great Lakes and is targeted by visually feeding planktivores. It is possible that Bythotrephes cederstroemi, alewife, or yellow perch, visually feeding predators, may be contributing to the smaller population abundances of daphnids found at the shallower depths. Greater numbers of L. macrurus at the 30 meter contour may be a result of the species affinity to colder waters (Balcer et a1. 1984). L. macrurus is a cold-water stenotherm that is generally restricted to the hypolimnion (wells 1960), and is seldom found in waters warmer than la'C (Balcer et al. 1984). Differences between seasons are not readily apparent from abundance 29 estimates alone, and must wait for the following PC analysis. W111! A scree diagram is a subjective method of deciding the number of principal components to analyze. The scree diagram (Figure 2) illustrates the declining percent of total variance explained by subsequent principal components. The number of principle components to be analyzed, 4, is one less than the point where the curve flattens at PC 5. This cut off point at PC 4 can also be explained intuitively. Dividing the 15 principle components into 100%, results in 6.66%. If all the variance was random, each component would have no more than 6.66% of the total variance. In both cases, PC 5 explained less than 6.66% for both the areal (6.5%) and the cubic (6.3%) abundance data. The first four PC's from the areal and cubic abundance data of 15 species account for more than 66% of the total variance in both cases. The first component, PC 1 accounts for 26.1% of the areal and 26.9% of the cubic abundance variance. Principle component 2 accounts for an additional 17.3% of the areal, and 17.2% of the cubic variance. The third accounts for 12.8% areal, and 13.3% cubic variance; and the fourth accounts for 10.1% areal, and 9.4% cubic variance. The first four principle components were analyzed with multiple regressions and transformed abundances were plotted on each of the significant variables (depth and month) to show slope. However, only the first two PCs were verified and plotted on the two-dimensional biplot. Because the biplot was necessary for the method chosen for species partitioning, correlation coefficients and verification beyond the first two principle components was not done. Percent of Total Variance Explained by Principle Component d ore-ammo 30 a. 12 3 4 5 6 7 8 9101112131415 Principle Components - Areal Abundance ' '4. 12 3 4 5 6 7 8 9101112131415 Principle Components - Cubic Abundance Flgxez.Screediagra'mshowhgpercentofvarianceexplahedbyeach componentforboharealandctbicabmdancedatamdthecutofl heforprincipiecorrponentsanalyzed. 31 Correlation Coefficients for PC 1 Correlation coefficients that were significant to each of the principle components of the transformed areal abundances are shown in Table 5. Ten species were positively correlated with PC 1. The species were: Cyclops sp., Diaptomus sp., B. longirostris, D. galeata, D. retrocurva, D. pulicaria, E. lacustris, E. coregoni, L. kindti, and L. macrurus (Table 5). Only one species, Chydorus sphaericus, had a significant negative correlation with the first component. The cubic abundance data had the same ten positively correlated species, and C. sphaericus was joined by B. cederstroemi as the only negatively correlated species (Table 6). Correlation Coefficients for PC 2 Five species were positively correlated with PC 2 of the areal PCA. The species were: Cyclops sp., Diaptomus sp., D. galeata, D. pulicaria, and E. lacustris. Species negatively correlated to the areal abundance PCA include: D. retrocurva, L. kindti, Diaphanosoma sp., Chydorus sp., and C. sphaericus (Table 5). With the cubic PCA, except for two species, Cyclops sp. and Eubosmina coregoni, all species reversed significant correlations from negative to positive and from positive to negative (Table 6). Cyclops sp. was significant only with the areal PCA (positively) (Table 5) and E. coregoni only with the cubic PCA (Table 6). V r c As would be expected with principle component analysis, the regression of the first principle component had the best fit (multiple rz-.689) (Table 7), and the fourth PC had the worst fit (multiple r2-.029) (Table 8). The same was true for the cubic abundance data 32 Correlation coefficients of principle components 1 - 4, Table 5. using areal abundance data. * SPECIES PC 1 PC 2 PC 3 PC 4 I nmommss 0.036 0.070 0.423 * 0.393 * F CYCLOPS SP. 0.469 * 0.250 * -.229 * -.034 DIAPTOMUS SP. 0.611 * 0.463 * -.058 -.027 BOSMINA 0.404 * 0.017 -.306 * 0.046 DAPHNIA GALEATA 0.802 * 0.431 * 0.083 -.116 D. RETROCURVA 0.674 * -.608 * -.l40 -.212 D. PULICARIA 0.392 * 0.335 * 0.165 0.056 EPISCHURA 0.229 0.542 * -.443 * 0.425 * EUBOSMINA 0.597 * -.117 0.161 0.656 * LEPTODORA 0.344 * -.293 * -.707 * -.080 LIMNOCAIANUS 0.509 * -.063 0.667 * -.l65 DIAPHANOSOMA 0.152 -.546 * 0.063 0.425 * CHYDORUS SP. -.135 -.606 * 0.183 0.449 * C. SPHAERICUS (.330 * -.478 * 0.072 0.303 * EURYTEMORA -.095 0.033 -.l78 0.268 * * Coefficient 'is significant (a - .05) either positively (> .230) or negitively (< -.230) with the principle components. 33 Table 6. Correlation coefficients of principle components 1 ~ 4, using cubic abundance data. * SPECIES PC 1 PC 2 PC 3 PC 4 l BNTHOTREPHES -.293 * 0.083 -.163 0.265 * F CYCLOPS SP. 0.359 * -.077 0.477 * 0.149 DIAPTCMUS SP. 0.502 * -.323 * 0.176 0.186 BOSMINA 0.400 * 0.067 0.523 * 0.662 * DAPHNIA GALEATA 0.889 * -.321 * -.209 0.144 0. 11311100th 0.599 * 0.667 * 0.119 -.349 * H D. PULICARIA. 0.411 * -.265 * -.121 0.183 EPISCHURA 0.156 -.587 * 0.539 * -.364 * EUBOSMINA 0.500 * 0.287 * 0.013 0.016 LEPTODORA 0.332 * 0.254 * 0.550 * -.170 LIMNOCALANUS 0.375 * 0.149 -.673 * -.057 DIAPHANOSOMA 0.065 0.540 * 0.092 0.179 CHYDORUS SP. -.164 0.595 * 0.102 0.393 * C. SPHAERICUS -.299 * 0.437 * 0.232 * 0.364 * EURYTEMORA -.089 -.072 0.237 0.078 * Coefficient is significant (a - .05) either positively (> .230) or negatively (< -.230) with the principle components. 34 Table 7. Regression and ANOVA of principle components 1 and 2 using transformed areal abundance data. Principal Component 1 Regression N: 76 Multiple R: 0.830 Multiple R‘: 0.689 Ad]. Multiple R‘: 0.676 Standard error of estimate: 2.628 Dependent Variable: Factor(1) Variable Coefficient Std. error Std. ooef. T P (2 tall) CONSTANT 1 1.586 2.628 0.000 4.410 0.000 DEPTH 0.284 0.030 0.618 9.372 0.000 ** MONTH -2.044 0.275 -0.507 -7.433 0.000 ** SITE 0.059 0.313 0.013 0.190 0.850 Analysis of Variance Source Sum-of-Squares DF Mean-Square F-ratio P REGRESSICN 1101.928 3 367.309 53.184 0.000 ** RESIDUAL 497.257 72 6.906 Principal Component 2 Regression N: 76 MultipleR: 0.516 Multiple R’: 0.267 Ad]. Multiple R‘: 0.236 Standard error of estimate: 3.285 Dependent Variable: Factor(2) Variable Coefficient Std. error Std. coef. T P (2 tall) CONSTANT -13.445 3.285 0.000 -4.092 0.000 DEPTH 0.151 0.038 0.404 3.989 0.000 ** MONTH 1.189 0.344 0.363 3.458 0.001 ** SITE 0.287 0.392 0.077 0.732 0.466 Analysis of Variance Soume Sum-of-Squares DF Mean-Square F-ratio P REGRESSICN 282.588 3 94.196 8.727 0.000 ** RESIDUAL 777.181 72 10.794 ** - significant at alpha = .05 35 Table 8. Regression and ANOVA of principle components 3 and 4 using transformed areal abundance data. Principle Component 3 Regression N: 76 MultipleR: 0.387 Multiple R': 0.150 Ad]. Multiple R‘: 0.114 Standard error of estimate: 3.046 Dependent Variable: Factor(3) Variable Coefficient Std. error Std. coef. T P (2 tall) CONSTANT 7.836 3.046 0.000 2.573 0.012 DEPTH -0.005 0.035 -0.015 -0.134 0.894 MONTH -0.675 0.319 -0.239 -2.118 0.038 SITE -1.204 0.363 -0.373 -3.315 0.001 ** Analysis of Variance Soume Sum-of-Squares DF Mean-Square F-ratio P REGRESSION 117.621 3 39.207 4.226 0.008 ** RESIDUAL 668.045 72 9.278 Principle Component 4 Regression N: 76 MultipleR: 0.170 Multiple R': 0.029 Ad]. Multiple R’: 0.000 Standard error of estimate: 2.886 Dependent Variable: Factor(4) Variable Coefficient Std. error Std. coef. T P (2 tall) CONSTANT 4.174 2.886 0.000 1.446 0.152 DEPTH -0.018 0.033 -0.062 -0.629 0.599 MONTH -0.407 0.302 -0.162 -1.346 0.183 SITE 0239 0.344 -0.084 -0.694 0.490 Analysis of Variance Source Sum-of-Squares DF Mean-Square F-ratio P REGRESSION 17.862 3 5.954 0.715 0.546 RESIDUAL ' 599.784 72 8.330 ** - significant at alpha = .05 36 except PC 4 had a better fit than PC 2 and PC 3 (Table 9 and Table 10). Month, depth and site were regressed on the transformed areal abundance principle components (l-4)(Table 7 and Table 8). Month and depth were found significant (p<.05) for PC's 1 and 2 (Table 7). Site was found significant for only PC 3 of the areal data (Table 8). Sites 1, 2, and 3 are Ludington, Manitou Passage, and West Grand Traverse Bay, respectively. Site 4 was Holland, and this site was the only location sampled only once in 1991. Site 4 had a much reduced abundance compared to the other three sites. Because Holland was the only site with a significantly different abundance, and because of the lack of samples from Holland, the site variable was dropped from the verifications and species partitioning sections reported later in the text. The multiple regressions of the cubic abundance data were significant for PCs 1, 2, and 4 (p<.05) for both depth and month, and site was not significant for any of the principle components (Table 9 and Table 10). ZQoplankton Community Trends - Areal Transformed abundances (ln[# + 1]), of both areal and cubic abundances of the significant variables in the multiple regression models were plotted both temporally (month) and spatially (at different depth contours) to examine trends of significantly correlated species within the zooplankton community. It is expected, that the zooplankton community would decline in abundance with season (after peaking earlier in the season), or had a negative slope with the month variable. For PC 1 this is true (Figure 3a), but unexpectedly the slope reverses to a positive slope for PC 2 (Figure 3b). 37 Table 9. Regression and ANOVA of principle components 1 and 2 using transformed cubic abundance data. Principle Component 1 Regression N: 76 MultipleR: 0.751 Multiple R': 0.563 Ad]. Multiple R‘: 0.545 Standard error of estimate: 2.057 Dependent Variable: Factor(i) Variable Coefficient Std. error Std. coal. T P (2 tall) CONSTANT 8.124 2.057 0.000 3.950 0.000 DEPTH 0.152 0.024 0.502 -6.260 0.000 ** MONTH -1.347 0.215 «0.506 6.433 0.000 ** SITE 0.149 0.245 0.049 0.607 0.546 Analysis of Variance Source Sum-of-Squares DF Mean-Square F-ratlc P REGRESSICN 392.880 3 130.960 30.965 0.000 ** RESIDUAL 304.506 72 4.229 Principle Component 2 Regression N: 76 MultipleR: 0.456 Multiple R‘: 0.208 Ad]. Multiple R': 0.175 Standard error of estimate: 2.213 Dependent Variable: Factor(2) Variable Coefficient Std. error Std. coal. 1' P (2 tall) CONSTANT 8.995 2.213 0.000 4.065 0.000 DEPTH -0.065 0.025 -0.266 -3.677 0.013 ** MONTH -0.852 0.232 -0.401 -2.533 0.000 ** SITE -0.331 0.264 -0.136 -1.255 0.214 Analysis of Variame Source Sum-of-Squares DF Mean-Square F-ratlc P REGRESSION 92.780 3 30.927 6.316 0.001 ** RESIDUAL 352.527 72 4.896 ** {significant at alpha = .05 38 Table 10. Regression and ANOVA of principle components 3 and 4 using transformed cubic abundance data. Principle Component 3 Regression N: 76 MultipleR: 0.319 Multiple R': 0.102 Ad]. Multiple R‘: 0.064 Standard error of estimate: 2.076 Dependent Variable: Factor(3) Variable Coefficient Std. error Std. coef. T P (2 tall) CONSTANT -2.684 2.076 0.000 -1.293 0.200 DEPTH . -0.038 0.024 -0. 177 1.353 0.119 MONTH 0.294 0.217 0.157 -1.580 0.180 SITE 0.552 0.248 0.258 2.230 0.029 Analysis of Variance Source Sum-of-Squares DF Mean-Square F-ratic P REGRESSICN 35.217 3 1 1.739 2.724 0.051 RESIDUAL 310.335 72 4.310 Principle Component 4 Regression N: 76 MultipleR: 0.464 Multiple R': 0.216 Ad].MuitIpie R': 0.183 Standard error of estimate: 1.633 Dependent Variable: Factor(4) Variable Coefficient Std. error Std. coef. T P (2 tall) CONSTANT 6.092 1 .633 0.000 3.730 0.000 DEPTH -0.060 0.019 0333 -2.941 0.002 ** MONTH -0.503 0.171 -0.319 -3.180 0.004 ** SITE 0406 0.195 0225 -2.084 0.041 Analysis of Variance Source Sum-«Squares DF Mean-Square F-ratio P REGRESSION 52.764 3 . 17.588 6.596 0.001 ** RESIDUAL 191.994 72 ' 2.667 ** - significant at alpha = .05 39 For the depth variable, it would intuitively be expected that the slope would be positive if the zooplankton community inhabits all the volume of water at both the 10 and 30 meter contours. The community does have a positive slope for PC 1 (Figure 4a) and PC 2 (Figure 4b) for the areal abundance data. The site variable was significant only with PC 3 (p<.05)(Tab1e 8)(Figure 5). It is hypothesized that species positively correlated with PC 1 will either tend to decrease markedly with season or be more abundant at the 30 meter depth. Because of the reversal of the seasonal trends between the first and second PC, it is hypothesized that species that are positively correlated with PC 2 will either not be greatly affected by season (because of the reversal of the month trend of PC 2) or more abundant at the 30 meter contour. Zooplankton Community Trends - Cubic The expectation that the zooplankton community would decline in abundance with season is also true for the cubic abundance data. Principle components 1, 2, and 4 were all significant and they all had negative slopes as well (Figure 6). The reporting of the zooplankton community measured as a cubic or volumetric abundance, would not necessarily show an increased abundance at the deeper contour as was expected with the areal abundance data unless the abundances were actually greater at the 30 meter contour. Principle component 1 has a positive slope (Figure 7a), and P68 2 and 4 have a negative slope (Figure 7b and Figure 7c). It is hypothesized that species positively correlated with PC 1 of the cubic abundance data will respond in the same way to both depth and 40 ‘0 I I I I I PACTOR1 O I I N O 0 .0 O 11 'b) FACTOR 2 O I l N O Q .0 O 11 Figure 3. Category plots by month of principle components 1 (a) and 2 (b) using areal abundance estimates. 41 10 I I FACTOR 1 3 fi 1 1O 30 b) FACTOR 2 O I n 10 30 DEPTH Figure 4. Category plots by depth of principle components 1 (a) and 2 (b) using areal abundance estimates. 42 10 | l l I 5_ .1 IF) I O I- 0' " 0 < I]. -5- .. -10 I l n 1 2 3 4 SITE Figure 5. Category plot by site of principle component 3 using areal abundance estimates. 43 ‘0 I I I I I FACTOR 1 O I I ‘10 1° ' I I I I b) rAcroaa O I l q l a 3. ‘ ‘ 1O FAOTOB 4 a I l c‘ - u -‘o I I I I I 1 I I 10 11 '0'?“ Figure 6. Category plots by month of principle components 1 (a). 2 (b). and 4 (0) using cubic abundance estimates. 44 10 fl , a) s- - C o I- 0' ' o < u. ‘. C ‘10 ' J 10 I” 10 l I b) ‘_ u N a: o I- 9‘ " O 4 II- ..L .. -‘o I I 10 1” 10 u I C) e?- '- FAOTOR 4 O I I 01'?" Figure 7. Category plots by depth of principle components 1 (a). 2 (b). and 4(0) using cubic abundance estimates. 45 month as with the areal abundance data. In other words, a species positively correlated with PC 1 would be expected to either decline markedly with season or be more abundant at the 30 meter depth. However, the differences between PC 2 of the cubic PCA are opposites of the seasonal and depth trends of PC 2 of the areal PCA. The month variable of the cubic abundance PCA does not reverse slope from PC 1 to PC 2, as was the case with the areal abundance PCA. And the other variable, depth, unlike the areal PCA, reverses from a negative to a positive slope. The result is that a species negatively correlated with PC 2 of the cubic PCA ordination would be hypothesized to respond in a like manner (to both the month and depth variables), as a species positively correlated with PC 2 of the areal PCA ordination. ZQgplankton Community Biplots Each principle component, is a linear compound of the transformed abundances, and has an associated eigenvector of each component coefficient giving the weighting of each species in the linear compound (Sprules 1977). The biplot uses the first two eigenvectors as (x,y) coordinates of the original variables. These coefficients are unitless and the first two principle components of the zooplankton community are plotted on the areal (Figure 8) and cubic (Figure 9) biplots. Distance and direction of the specie component coefficient has meaning, and species plotting close to the origin would not be expected to have their abundances influenced by the significant variables depth or month, and species abundances plotting far from the origin would be expected to be influenced by these variables. The biplots use the component coefficients of the principle 46 2 EP1 N I PUL .. DIA " = 0.5 oYc' 0 I t: EUR BYT 80$ 3 0 I I l E UM I o I EUB o 0.5 0 Lil” a SPH APH '; '1 I CHY I c .I E -1.5 .2 BET -2.5 -1 -0.5 0 0.5 1 1.5 2 2.5 Principle Component 1 Figure 8. Biplot of first two principle components using areal zooplankton abundance data from Northeastern Lake Michigan in 1991 (axes are unitless). 47 2 RET 1.5 I N E 1 g cm 0 SPH I APH g 0.5 ' I LEP 5:8 I 8 an un- o o I I 808 .9 EUR cvc 2 '- DIA ': -o.5 0- GAL PgL I -1 EPl I .1 .5 -1 .-o.5 0 0.5 1 1.5 2 Principle Component 1 Figure 9. Biplot of first two principle components using cubic zooplankton abundance data from Northeastern Lake Michigan In 1991 (axes are unitless). 2.5 48 components, and the species were assigned correlation coefficients of these components. They both respond in similar intensities either positively or negatively. Because of this, the use of a biplot will visually aid in the verification process and later, in the splitting of the zooplankton community into species associations. o lan n vir nmenta ends The variables, month and depth, may or may not represent meaningful ecological relationships, and the synthetic principle component variables must now be closely scrutinized. This scrutiny requires that the community trends of the principle components accurately portray the trends of positively correlated species, and that an inverse of the trends is portrayed by the negatively correlated species of the areal abundance data (Table 5), and the cubic abundance data (Table 6). The regression of the variables depth and month were both significant when regressed on PC 1 with the areal abundance data (Table 7). The hypothetical trends month (Figure 3a) and depth (Figure 4a) of these variables on the zooplankton community with respect to PC 1 is illustrated in Figure 10. These same variables were also significant when regressed on PC 2 and the hypothetical trends of month (Figure 3b) and depth (Figure 4b) are illustrated in Figure 11. Species significantly correlated with a principle component either positively or negatively would be expected to be influenced (by relative abundance) by one or both of the variables found significant in the regressions. Species that were not correlated with both principle components may or may not be influenced, and for this reason the uncorrelated species of both PC 1 and PC 2 will not be used in the verification process or the delineation of species into species 49 2 EPI 1.5 I (SI-\L N 1 - -- - ............... 30903.. .. «I . Jg’i‘k’”. ":{w- ~ :- c 0.5 '. . 0 I: C a. 0 E O 0 -0.5 0 O. - -1 c u. ..., ..... .. ,L -1.5 -2 -2.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 Principle Component 1 Figure 10. illustration of the significantly consisted variables. depth and month. of principle component 1. expected to influence a species position on both the areal and cubic abundance biplots. 50 Principle Component 2 -1 -0.5 0 0.5 1 1.5 2 2.5 3 Principle Component 1 Figure 11. illustration of the significamiy correlated variables. depth and month. of principle component 2. expected to influence a species position on the areal abundance biplot. 51 associations. Figure 12 illustrates only those species that were significantly correlated with both PC 1 and PC 2 (Table 5). Following are the areal, then the cubic abundance verifications. Because it is unknown which variable, either month and depth (or both), is affecting the relative abundance of a species, each species transformed abundance will be plotted by both month and depth. Verification of each component will first plot the transformed abundances of one or two of the most positively correlated species by first month then depth. Then one or two of the most negatively correlated species by first month than depth. Areal Abundance Verification Most species correlated both positively and negatively with the areal abundance data (Table 5) were also correlated in the same manner as with the cubic abundance data (Table 6). Only B. cederstroemi was correlated (negatively) with the cubic abundance data, and not correlated with the areal abundance data. Therefore, except for B. cederstroemi, verification of PC 1 for the areal abundance data will also represent the verification of the cubic abundance data as well. The first two plots are the log transformed abundances of the two species with the highest positive correlation with PC 1, D. galeata and D. retrocurva (Table 5). It is hypothesized that these species would either decrease in abundance with season, or be in greater abundance at the 30 meter contour (Figure 10). A plot of these two species by month (Figure 13) shows that only D. retrocurva is much affected by season. D. retrocurva was no longer found in the community after September, while D. galeata was present in abundances similar to samples taken in July, and was present until November. The graph of these two species plotted 52 Figure 12. Biplot illustrating species that are significantly correlated with principle components 1 and 2 using the areal abundance data. 53 17 1 2 fl. f i X Daphnia galeata - D. retrocurva 14 - . - 25 X 11 r- ; x - XX % X X :3 g x + __ x 3.., 8 T :- x - x X - E X 3% ): X \ :X x X :3 5 7 — 5}" — x X x ‘ E -: Y _ 8 ”K .—x — 2 — ._ -=-,__ ‘75:. _=- _1 i g? i :fix T 6 7 8 9 10 11 12 M O N T H Figure 13. Log transformed (in) areal abundances of Daphnia galeata and D. retrocurva for the months of July through November in 1991. 54 by depth (Figure 14) shows both species appear to have greater abundances at the 30 meter contour. Many samples for both species had absences, however, D. galeata was absent only at the 10 meter contour which may indicate the species is more affected by depth than by season. The only species with a significant negative correlation to PC 1 was C. sphaericus. It is hypothesized that this specie's abundance would either not differ much throughout the season or has a greater abundance at the 10 meter contour (Figure 10). When this species was plotted against month (Figure 15), the species does not appear to follow the seasonal trend of PC 1. However, abundances were highest in August and October and absent in September. It appears that this species is indeed not greatly affected by season. A plot of this species by depth (Figure 16) clearly shows that C. sphaericus corresponds very well with the second part of the hypothesis, that it is more abundant at the 10 meter contour. The species was totally absent at the 30 meter depth. The next two plots are the log transformed abundances of the two species with the highest positive correlation with PC 2, D. galeata and E. lacustris. It is hypothesized that these species would either decline markedly with season or have greater abundance at the 30 meter depth (Figure 11). The first plot, is the graph of the two species by month (Figure 17). Neither species shows decline in the later months and both are present in all samples in November. The next plot of the two species is by depth (Figure 18), and it appears that both species are present at greater abundances at the 30 meter contour. The next two plots are of the two species most negatively correlated with PC 2, D. retrocurva and C. sphaericus. It is hypothesized that these two species would declined markedly with season 55 17 1 1 l x Daphnia galeata — D. retrocurva 14 - .— x 11 — g? - ’1: x x 25 + x x )0?K 3‘ 8 ” Xx? figs: " E § X213 fit - x V 1_ _— X _. I: 5 7:3: =_ " 1% 2 __ _ m f: _, a”: . . 0 . 10 20 30 40 DEPTH Figure 14. Log transformed (in) areal abundances of Daphnia galeata and D. retrocurva at two depths. 10 and 30 meters. 56 7 I l i l l 6 O O '- 09 o o 5- O o 0 O o 4_ _ :: o 8 o + 0 A3” '- N E 112- ‘ E —1_ ._ 0- 9e 99 8%. ~ _1 l J I I l 6 7 8 9 1O 11 MONTH Figure 15. Log transformed (in) areal abundances of Chydorus sphaericus for the months of July through November in 1991. 12 In((#lm2)+1) 57 mCOO O O O 00% DEPTH Figure 16. Log transicrrned (in) areal abundances of Chydorus sphaericus at two depths. 10 and 30 meters. 40 58 17 1 . 1 T 1 1 X Daphnia galeata — Epischura lacustris 14 ~ 1 ~ X X XXX 11 H x - x .28” —_ A I )2(X 1 3 "' "‘— g: 1»: Q. ~ «E >2); 2:8; :11 ix ’5. N f -- 111 5 a x- ‘* V 5 —- -— — x - v __ :xc- : .5 .3 >5. "1‘ X 2 - _ : 2.1.: =- j: _1 1 _i 1 A 1 6 7 8 9 10 11 MONTH 12 Figure 17. Log transformed (in) areal abundances of Daphnia galeata and Episcura lacustris for the months of July through November in 1991. 59 17 1 : 1 X Daphnia galeata — Epischura lacustris 14 - - X >Z< 11 1 g? _ a. 3 fi— .— §x 48% + e L 39% - 5‘ _ Ex 5 I: t 5 " g x—_— -1 5 is}? w 2 - ._ _1 . g I .1: 0 10 20 30 40 DEPTH Figure 18. Log transformed (in) areal abundances of Daphnia galeata and Episcura lacustris at two depths. 10 and 30 meters. 60 or have a greater abundance at the 10 meter depth (Figure 11). The plot of the twp species by month (Figure 19) shows that only D. retrocurva decreases with season and is absent in the samples after September. The plot of the two species by depth (Figure 20) shows that only C. sphaericus is more abundant at the 10 meter contour. Cubic Abundance Verificgtion As was mentioned above, the only species whose correlation coefficient was found significant in the cubic and not in the areal abundance, was B. cederstroemi. This species was negatively correlated with PC 1 of the cubic abundance data (Table 6). It is hypothesized that this species either does not decline markedly with season or it is more abundant at the 10 meter contour (Figure 10). When B. cederstroemi was plotted by month (Figure 21) the species showed no decline with season, instead the species was absent in the September samples then in October and November rebounded to areal abundance levels similar to July and August. When B. cederstroemi was plotted by depth (Figure 22), it showed no greater abundance at either the 10 or the 30 meter contour. Hypotheses of the species responses to principle component 2 of the cubic abundance data (Figure 23) are reversals of the PC 2 areal abundance hypotheses (Figure 11). In other words, most species in the upper two quadrants on the areal biplot (Figure 8) will be in the bottom two quadrants of the cubic biplot (Figure 9) and vice versa. However, positively and negatively correlated species from PC 2 of the cubic PCA (Table 6) also reverse direction (positive to negative and vice versa) from the areal (Table 5). This results in both biplots grouping many of the same species into groups with hypothetically similar responses to both month and depth, or species associations. Therefore the following 61 1 17 ' l 1 1 u i [X Daphnia retrocurva — Chydorus sphaericus 14 — 1 - 11 '" §>< - :: Xx .1 8 _ xx x _ N 5 1. x 28‘ '11: xx 5% x _- t: 5 ' 33‘ - ‘ I; xx X X _ ._ ._ x x 2 — x ._ -_—__ -—: 13"“ 4%: -1 T § -1— xia 7: 6 7 8 9 10 11 12 MO NTH Figure 19. Log transformed (In) areal abundances of Daphnia retrocurva and Chydorus sphaericus for the months of July through November of 1991. 62 17 1 1 X Daphnia retrocurva — Chydorus sphaericus 14 — - 11 H x X - X A xx ‘- xx 1'. 8 - XX — a — x,< E ’38 z _ 1 :3 5 ’2: § — E. ici- x 2 ~ xx - _1 g l g 0 10 20 30 40 DEPTH Figure 20. Log transformed (in) areal abundances of Daphnia retrocurva and Chydorus sphaericus at two depths. 10 and 30 meters. 63 7 I I I I I e_ — 5- O " 41 — : O 0 6:. {‘3' (30 O@ O O '— E 8 o 32' i O o " E 000 8 -1~ a. - o % % g o 0* O T _1 1 1 1 1 1 6 7 8 9 10 11 12 MONTH Figure 21. Log transformed (in) cubic abundames of Bythotrephes cederstroemi for the months of July through November In 1991. 64 7 1 l i 6‘ —. s— 0 - A 4“ " .- 630 00 ' + (g 0 6‘ 3" 0% .— E % c290 3 2' 5. ‘ 2' $22 _ 1 _ g _ 86 % 0— 08 o _ _1 l l J O 10 20 30 DEPTH Figure 22. Log transformed (in) cubic abundances of Bythotrephes cederstroemi at two depths, 10 and 30 meters. 40 6S Principle Component 2 GAL -1 -0.5 0 0.5 1 1.5 2 2.5 .Princlpie Component 1 Figure 23. Illustration of the significantly correlated variables. depth and rncnth. of principle component 2. expected to influence a species position on the cubic abundance biplot. 66 verifications of the cubic abundance environmental trends need only be compared to the areal abundance verifications to ensure that the significantly correlated species respond in a like manner. The next two plots (month and depth) are of the most positively correlated species to PC 2 of the cubic abundance PCA, D. retrocurva and C. sphaericus. These species are the same as the two most negatively correlated to PC 2 with the areal abundance PCA. The response of the cubic (Figure 24) and the areal abundance (Figure 19) of the two species to season are very similar, except that the cubic measurement is less marked than the areal. Next is the comparison of these same two species response to the different depth contours, and again the cubic response (Figure 25) is very similar to the areal response (Figure 20). The two most negatively correlated species to PC 2 of the cubic PCA are D. galeata and E. lacustris. Comparison of the cubic responses to both month (Figure 26 and Figure 17) and depth (Figure 27 and Figure 18) are also similar. Verification of Correlated Species If the grouping of species into species associations is reduced to using only the significantly correlated species of the first two principle components, then the two types of species associations, will differ by only a few species (Figure 12 and Figure 28). In other words, Cyclops sp. and E. lacustris are in the proposed areal association illustration but not the cubic, and E. coregoni is in the cubic association illustration, but not in the areal. 67 17 1 1 1 . 1 x Daphnia retrocurva — Chydorus sphaericus 14 - - 11 F - + ‘ -—1 9 x" : 5 — 2;; x _ E xX % >35 __ _& “Z. _ 9" x _ ._ 2 % x x f - 1% ——:— 11 1.1 _1 J I . I 6 7 8 9 10 11 12 MONTH Figure 24. Log trandormed (in) cubic abundances of Daphnia retrocurva and Chydorus sphaericus for the months of July through November of 1991. 68 17 1 s I. X Daphnia retrocurva - Chydorus sphaericus 14 — - 11 — 2 - i 8 T x x ‘ E 23‘ XX 3 5 1 ' as 1 v xx .5 i >22 .2, 2 - 1%? § _ X _1 g l g 0 10 ' 20 30 40 DEPTH Figure 25. Log transformed (in) cubic abundames of Daphnia retrocurva and Chydonis sphaericus at two depths. 10 and 30 meters. Ir‘? 3"”“1 17 14 11 ln((#lm9)+1) 69 x‘ Daphnia galeata — Epischura lacustris x x x xX :1" ”is. x -_ X‘ x x _ _ .98 g x : — _. x- — X _ X“ 21$ x — -x -:-7_. g x g: _ _ 71—35 __ X _ 408‘ — -x 1 1 1 1 1 7 8 9 1 0 1 1 MONTH 12 Figure 26. Log transformed (In) cubic abundames of Daphnia galeata and Epischura lacustris for the months of July through November in 1991. 7O 17 1 : 1 X Daphnia galeata — Epischura lacustris 14 _ _ 11 — _ C x + “E 12,. Sir : s. ’1 :3 5 - .4— 313 - c 212x - 5% f; f; 1’ 2 1- 2.x _ pa _1 g l —l—:- 0 10 20 30 40 DEPTH Figure 27. Log transformed (In) cubic abundances of Daphnia galeata and Epischura lacustris at two depths. 10 and 30 meters. 71 N 1.: 7 . -/////////////////~ m; %‘ ' -1 -0.5 1.5 2 2.5 Principle Com1ponent 1 Figure 28. Biplot illustrating species that are significantly correlated with principle components 1 and 2 using cubic abundance data. ' 72 W W Species will first be assembled into quadrants until the actual species associations are described later in the text. Figure 29 illustrates the species that are hypothesized to be in the areal species associations. Each quadrant has text that describes what the hypotheses are for each quadrant and the responses of the verified species to each of the variables of the hypotheses. The unverified species (uncircled) are species that are significantly correlated to both PC 1 and PC 2. The verified quadrant I species of the areal PCA, E. lacustris and D. galeata were both more abundant at the 30 meter contour and did not decline markedly with season. It is therefore hypothesized that D. pulicaria, Diaptomus sp. and Cyclops sp. will respond in a similar manner and the species could then be characterized by these environmental trends. Both Diaptomus sp. and Cyclops sp. did not decline markedly with season (Figure 30), and were more abundant at the 30 meter contour (Figure 31). D. pulicaria differed from the other quadrant I species with respect to month and declined with season and was absent from the samples by October (Figure 32). However, D. pulicaria did respond similarily to the depth variable (Figure 33). The verified quadrant II species of the areal PCA, D. retrocurva declined markedly with season and was more abundant at the 30 meter contour. The other hypothesized quadrant II species, L. kindti also declined with season and.was absent from four of six samples in October and totally absent by November (Figure 32). Like D. retrocurva, L. kindti when present in the samples, was more abundant at the 30 meter 73 Epischura and Daphnia galeata both are more abundant at the 30 meter contour Neither species declined markedly with season. EPl GAL N PUL E DlA 2 mm Quadrant i o a. E Quadrant ill Quadrant ll 0 g LEP .3 Daphnia retrocurva declined markedly c with season and was more abundant E at the 30 meter contour Chydorus sphaericus did not decline markedly with season and was more abundant at the 10 meter contour Principle Component 1 Figure 29. illustration of significantly correlated species of the areal abundance biplot. Verified species and their responses to hypothetical expectations related to month and depth in the verification process (circled). and unverified species (uncircled) within proposed species associations or quadrants. 74 17 l l i l l 14 —- __ - a— — —— __ L :— % X’— =’5< E 11 " x— - § _ A *- :- f ’53-: ’3. I §— xx X x A 3 ‘ >2‘x ‘ “E x>§< it 5 — - E x Cyclops sp. 2 .._ — — Diaptomus sp. _1 I l J I l 6 7 ' 8 9 1O 11 12 MONTH Figure 30. Log transformed (in) abundances of the significantly cor- related quadrant i groups. Cyclops sp. and Diaptomus sp.. for the months of July through November in 1991. 75 17 l i l 14 — _ - :93.- 11 F — g - X :: t? i 3 ” >1< " (\l E x it 5 - - E 2 __ x Cyclops sp. _ — Diaptomus sp. -1 l J i O 10 20 30 4O DEPTH Figure 31. Log transformed (in) abundances of the significantly cor- related quadrant l groups. Cyclops sp. and Diaptomus sp.. at two depths. 10 and 30 meters. ' 76 15 I I I I I X D. pulicaria O Leptodora kindti 1 1 — - z: 059‘ + _ x — a? 7 ‘3 5120) s 8% é o 3 (SQ) 00 0 V o '5 3 ” {5 Q x ‘ X @ x06 @ %e _1 l l I l 6 7 9 ‘i O 1 1 1 2 MONTH Figure 32. Log transformed (in) abundances of the significantly cor- related quadrant ! species. Daphnia pulicaria and the quadrant il species. Leptodora kindti for the months of July through November in 1991. 15 11 in((#lm2)+1) Figure as. 77 i I l X D. pulicaria O Leptodora kindti 4O DEPTH Log transformed (in) abundances of the significantly cor- related quadrant l species. Daphnia pulicaria and the q'uadrant li species. Leptodora kindti at two depths. 10 and 30 meters. 78 contour. Quadrant III contains only C. sphaericus and this species declined with season (Figure 15), but not as drastically as other species such as D. retrocurva, D. pulicaria or L. kindti. This species placement on the left side of the biplot is more likely because of the remarkable characteristic of having a greater abundance at the 10 meter contour. Wm Figure 34 illustrates the species that are hypothesized to be in the cubic species associations. The unverified quadrant I species, L. kindti and E. longirostris are hypothesized to decline markedly with season and be more abundant at the 30 meter contour (Figure 34). L. kindti declines gradually. In September and October, two of six samples, then four of six samples lack this species, respectively, and in November the species was totally absent from the samples (Figure 35). E. coregoni also declined, but not gradually. It was totally absent in September, and one in six samples had the species present in both October and November (Figure 35). Both species. when present, were more abundant at the 30 meter contour (Figure 36). The unverified quadrant II species, D. pulicaria and Diaptomus sp. are hypothesized to not decline markedly with season and be more abundant at the 30 meter contour. Only Diaptomus sp. responded in this manner to both season (Figure 37) and to depth (Figure 38). D. pulicaria, when present, was more abundant at the 30 meter depth (Figure 38). but declined significantly with season and was absent by September. W Only two variables, month and depth, were significant in the regressions of the principle components on the zooplankton community. 79 Chydorus sphaericus did not decline markedly with season and was more abundant at the 10 meter contour Daphnia retrocurva declined markedly with season and was N more abundant at the 30 meter .- contour - C 2 Quadrant iV LEP EUB Quadrant i 3 E Quadrant m DIA Quadrant ii 0 o a '8 .5 PUL 0- Daphnla galeata did not decline markedly with season and was more abundant at the 30 meter contour Principle Component 1 Figure 34. illustration of significantly correlated species of the cubic abundance biplot. Verified species and their responses to hypothetical expectations related to month and depth in the verification process (circled). and unverified species (uncircled) within proposed species associations or quadrants. 80 l l l l I x Leptodora kindti 7 O Eubosmina coregoni :: 5 ' + 6‘ E X 1h 3 P x x X — Z: X o o s X x 1 —- ._ o s g %a ._1 i i l 6 9 1O 11 12 MONTH Figure 35. Log transformed (In) cubic abundames of the significantly correlated quadrant i species. Leptodora kindti. and Eubosmina coregoni for the months of July through November In 1991. 81 9 i i l X Leptodora kindti 7 O Eubosmina coregoni X o A 5 " "' 1; 3% 6‘ E 3 _ 1“: a 8 E o 1 .. C? _ -1 I J J 0 10 ‘~ 20 30 DEPTH 40 Figure 36. Log transformed (in) cubic abundances of the significantly correlated quadrant l species. Leptodora kindti. and Eubosmina coregoni at two depths. 10 and 30 meters. 82 1 5 I I I I I X Diaptomus sp. 0 D. pulicaria n P as: - X x x § A £2... 2}? a; 1- X X XX Xx X '1' 7 -— X g X -1 6‘ x E g ‘2) t: 0 O O C _. O O O O @ 0 ‘% 6 7 8 9 1 O 1 1 1 2 M O N T H Figure 37. Log transformed (in) cubic abundames of the significantly correlated quadrant ii species. Diaptomus sp. and D. pulicaria for the months of July through November in 1991. 83 15 I I I X Diaptomus sp. 0 D. pulicaria 11- “ )§(x a: xwaw " In((#/m3)+1) 0 10 20 3O 40 DEPTH Figure 38. Log transformed (In) cubic abundances of the significantly correlated quadrant ll species. Diaptomus sp. and D. pulicaria at two depths. to and 30 meters. 84 Following are the three proposed species associations. Species significantly correlated to the first two principle components can hypothetically be grouped together in these associations as a means of describing the groups of species within the zooplankton community. a ec e s cc at on There was one species, C. sphaericus, that showed a negative correlation to both PCl and PC2. In both the areal and the cubic PCA the species had abundances greater at the 10 meter contour, and was never found at the 30 meter contour (Table 3 and Table 4).The rest of the bedoridae family, though not significantly correlated with PC 1 in both the areal (Table 5) or the cubic PCA (Table 6), was significantly correlated with PC 2. The species was in the same quadrant as C. sphaericus in both biplots (Figure 8 and Figure 9), and was absent at the 30 meter depth in all but one sample in July (Table 3 and Table 4). It is proposed that these two species be grouped in a species association on the basis of their affinity to the shallower depths in the Great Lakes (Balcer et al. 1984). The seasonal trends of the pair differ. Both groups had their highest abundance in July. C. sphaericus, although rare, was found in the samples until November, and Chydorus sp. was no longer found in the samples after August. The known life history and habitat preferences for the sediment and littoral zones of both species conforms with this association (Balcer et al. 1984). - S s s It was hypothesized from both the areal and cubic abundance PCA's that because D. retrocurva was in greater abundance at the 30 meter depth and declined significantly in the fall months, that the other 85 species included within a quadrant with this species would also react similarily to the variables. L. kindti was in this quadrant for both the areal and cubic PCA. L. kindti both declined gradually with season (Figure 32) and was more abundant at the 30 meter depth (Figure 33). Another species, E. coregoni found significant only in the cubic PCA, though more abundant at the 30 meter depth (Figure 36) persisted in one of six samples in both October and November (Figure 35). D. pulicaria is another species that should be included in this grouping although the species was never placed in the same quadrant as D. retrocurva. In each of the biplots, D. pulicaria was placed on the other side of the x-axis, in the quadrant that is proposed to be the persistent species. For the areal PCA, D. retrocurva, L. kindti, and D. pulicaria; and in the cubic PCA, D. retrocurva, L. kindti, E. coregoni, and D. pulicaria are grouped in the limnetic non-persistent species association. Limnetic Persistent Species Association It was hypothesized from both the areal and cubic abundance PCA's that because D. galeata (and E. lacustris in the areal PCA), was in greater abundance at the 30 meter depth and persisted into the fall months, and that the other species included within a quadrant with this species would also react similarily to the variables. As mentioned above, D. pulicaria was included within these quadrants, yet declined appreciably with season. The other two species within the respective quadrants, Diaptomus sp. in both biplots, and only Cyclops sp. in the areal biplot. did 86 however persist. However, it is impossible to tell what is happening with both of these large groupings of species with respect to season. It may be many species peaking at different times within the genus, or one species dominating over the entire season. For the areal PCA, D. galeata. Diaptomus sp., Cyclops sp., and E. lacustris; and in the cubic PCA, D. galeata, Diaptomus sp., and E. lacustris are grouped in the limnetic non-persistent species association. W V i ati s 0 mass Mean dry weights (ug) (used to estimate biomass), and 95% confidence intervals (n-30) for the volume estimated species Diaptomus sp. and Cyclops sp. are shown in Table 11. Both of these species dry weight estimates were based on both length and either depth or width measurements. Because of this, size differences of many different sized adults and copepodid stages can give widely differing dry weight estimates. Compounding these variations is that both the Diaptomus sp. and Cyclops sp. groupings are composed of different sized species. The significantly smaller dry weight average estimates of Diaptomus sp. and Cyclops sp. at the 10 meter contour in West Grand Traverse Bay on 8/26/91 (Table ll) was because the population was dominated by the shorter and thinner early copepodid stages. Diaptomus sp. lengths ranged from O.38-0.64 mm, and Cyclops sp. from O.34-O.6O mm. These measured lengths were less than half the measured dimensions of Diaptomus sp. and Cyclops sp. on comparable dates at Ludington on 8/29 and at Manitou Passage on 8/27 (Table 11). This lag in development was expected because 87 2.00. 3.00.0. 000.0 0000.. 80...... .000. 000.0 0000.0 0. .20 2.. 3.. “0.0000000. 00.0.0. 0005.0 80.0.0000. 000.0 00.3.0 00 .20.... 3.. 800. ..000.0. 0. ..0 0000.. 30......0000. .30 0000.0 0. .20 :0. 3.. A000...00...0. 00.0 0000.. 830.000... 000.0 .000. 0. .2020. 3 ...0.00....000.0. 0. ...0 0.0.0 80020.00. .00.. 0.00.0 00 .2020. 3.. 300.0060. 00.0 0000.. 830.000... 000.0 .000. 0. .20200 3.. 800...,0000. 0. ..0 0000.0 $00.50 .00. 000.. 0.00.0 00 .20 :00 3.. $00.... .00. .0..0 0000.. 3.000.000. 000.0 02.050 0. .20200 3.. 5.0.0.100 .0. 2.0.0 0000.0 800......000. 000.0 000 ..0 00 .20200 3.. «0.0.0....0000. 000.0 00.0.0 3.0.0.00... 00.0 0000.. 0. .20200 or am.m.0.w..w......0.000. .00... .0 02. ... . .0 00.0.00... .. 000.0 2.0 . .0 00 .20200 0..., . $00......00.0. 000.0 .0000 3.0.0.000... 000.0 030.. 0. .200200 0.5.. .A0.0.0...._..0.s0..0. .0 . .0 000 ... 800.00 .0... . 00.0 0. 00.0 00 .20200 a s. . 300.0000. 2.00.0 0000.0 8.....00E0. 000.0 0000.0 0. .200200 0... $300000. .. 000.0 .00.... 300.0. .000. 000.0 0000.0 00 .20200 0...... «000......00. .30 0000.. 2.000.000. 000.0 2.02.0 0. .20200 3 . _ “00.0.0...00..0. .00....0 0000.0 8000.300. 000.0 000 ..0 00 .20 :00 3.. 80.00.0060. 000... 500.... 8000.000... 000.0 0000.0 0. .20200 as. . ...0....,.0_.,..s00.0. 00.0 0. .0.. 3000.000... 2.00.0 0000.0 00 .20 :00 as. 8.0.03.0. 0....0 0000.0 0000.500. 000.0 0000.0 0. .20200 0.. H.000...€ .00. .0 . ..0 .0. . ... 8000.20. .. 000.0 0000.. 00 .20200 0... . 2.000.200. 000.0 .0000 8000000.... .000 5.00... 0. .2..200 3.. . _ a.«0,020.0...000.... 000.0 0000.0 200.0080. .000 0000.0 00 02.02.00. .3 $000030. 000.0 .0000 8000000.... .000 200.0 0. .2050 3.. ..A..0._0~.0..00~.. .. 000.0 0000.0 £00....0000. .000 0000.0 00 .2022 3... 80a... .0 00 0002 80".... .0 00 000.2 0 make .00.. d0 0ao.o>0 d0 0380505 d0 0ao_o>0 oce d0 0360320 to. 0.023.... eocep 0:00 0000 oce cote 0.0.33.0 .81. £0.03 >5 ceeE .0 00.05.30 .0053)ch .p. 030... 88 of observed differences in both temperature and zooplankton abundances between Grand Traverse Bay and other locations on Lake Michigan the previous year (Barner unpub. data). In addition, the previous sample at west Grand Traverse Bay (8/6/91) contained the largest proportion of nauplii of all the samples. It was suspected to be an important component to the total biomass, but estimates found the estimated weight to be insignificant (922 ug), or less than 0.2% of the total biomass for that sample. Lgng;h-Egigh§ gaggggsiog Estimations Dry weights were split into two tables because 8. longirostris and D. galeata, in most cases, had at least 30 animals measured and there is a higher degree of confidence in the length measurements than in the case of D. retrocurva, E. lacustris, and L. macrurus. Mean dry weights (ug) used to estimate biomass from length-dry weight regressions, the number of animals measured (n), standard error and 95% confidence intervals are shown in Table 12 for B. longirostris and D. galeata, and Table 13 for D. retrocurva, E. lacustris, and L. macrurus. The biomass weights of this study are similar to those estimated by Hawkins and Evans (1979). The exception, was the weight of D. galeata, which averaged from 3.7 to 9.6 ug per individual in this study, and just 4.0 ug in the Hawkins and Evans (1979) study. The discrepancy may be explained by the differing methods of estimation. Hawkins and Evans (1979) estimates should be the most accurate because they were done directly using a microbalance, whereas in this study they were only estimated by a length-dry weight regression and unverified. 89 FQVCFw 89;.mwi oofio mw ..F 2» :o F. F 286 F 93m F 3.: 39° NF. §.o.wui . Avmfimmé 286 h .. .. .. .. .fi. $4.0. my 3+6 5 ficimmdv 03.0 on 3.5 dag. «mod «mm :m.r.wm..: mm; on 3&4 £me med on. 8m;.vo..: owed on am_d..om.c had or $359.: 30.0 on .3... d. .v, «and mm $3.3 , . . 33.3.8 «mod on. _. .6 . a .. a: a Sciopi 30.0 on .3 r q. :9» 3. 3.1.3.: 3.0.0 on .3 5...; _..Z.P...o. ,. on. 0 mm 2 dado—cu cicada .cooohoo «EEuom .Suoio c.5360 ucu .cooouoo 65825 .0 2028...: oocovccoo $ma uca cote Bone...» .35 £0.25 >5 can... .0 0366900 $05.02.: .Nw 030... 9O . ..§.m..mu.e Q _. .9 «EEouE 3530055.. «£300.2328. w «230952 Sc: do 6228c. 3:830:53 ted 050306. Sacco—am 62:09:! «Isaac .5.— oiZoE. oocouccoo «8a 9.6 coho 9.69.88 .33 «£0.03 bu EOE .5 0015.30 $05325 .9 030.... 91 i s 3 ma 0 Bythotrephes dry weights (ug), derived from regressions of instar weights on epilimnetic temperatures (Burkhardt 1991) are reported in Table 14. Weights are reported for each instar, and neonates. Animals that had broken spines could not be aged and were multiplied by the average of all instar weights of the sample. Counts of instars and temperatures (’C) used are in APPENDIX B. Walsh Biomass (mg/m2) for each contributing species, from all three methods are combined in Table 15. Total biomass for individual samples ranged from 24 mg/m? to 3782 mg/m?. Total biomass for each location, date, and depth, (averaged over replicates), ranged from a low of 53 mg/m3 in Ludington on both August 29th and September 13th to a high of 2691 mg/m2 in West Grand Traverse Bay on August 26th. Copepods dominated the biomass over the entire season. Diaptomus sp. averaged 60.5% of the total biomass, followed by Cyclops sp. with 11.1%. The third most abundant species, Bosmina longirostris, ranked fifth in total biomass (4.8%), and the fourth most abundant, D. galeata ranked third in total biomass (10.5%). B. cederstroemi was only 0.1% of the total abundance, yet it was ranked fifth in biomass with 8.4% of the total biomass. The remaining species measured for biomass were: E. lacustris 3.0%, L. macrurus 0.8%, and D. retrocurva 0.7%. Both Manitou Passage and West Grand Traverse Bay increased at the 30 meter depth, and decreased in average biomass at the 10 meter depth in August of 1991. Grand Traverse Bay began at a very low average biomass (210 mg/m?) on August 6th at the deeper station, and increased 92 Table 14. Estimated dry weight blomass (ug) of three Instars. neonates, and broken splne animals of Bythotrephes cederstroemi at two depths and four sites In northeastern Lake Michigan In 1991. Instar 2 #lm2 3rd 2nd 1a neonate broken' .. 9.5800". . 4. ,. LU 07/17 1 0 287 63800 1 4455 3780 1 540 8856 92.4 LU 07/17 1 0 1 91 34650 7965 4900 560 5576 53.7 LU 07/17 1 0 1 89 25300 7965 7000 21 0 4264 44.7 93 Table 14. (oont'd) lnstar Loc. Date 2 #lm2 3rd 2nd 1st neonate broken* total(mg) H0 0828 10 33 6240 3760 1050 0 1900 130 HO 08/28 10 43 6240 61 10 1 575 0 1900 1 5.8 H0 081/38 10 33 3120 6580 1050 0 950 1 1 .7 6660 35.5 3552 10.6 6993 19.9 4.1.115;- 14.-6 572' 3.1 . .04 1 -0 468 5.1 884 6.4 ooooooooooooooooooooooooofi 14352 39.8 94 yrs-mot N — «mum-o o 0 0'0 0. o o o 0:05;: o o o c.3340 o o c or 38 :8 or 33 too .3 3.? .28 .3 £83. as... 2 3328 2 353° 3 3628 2.. 5.23%.» a.“ -533. 8 3:28 .8 653° ,3. 85.5.... 8 .223 .2 SEE 0 mm, mbum So o>o ._.>m N .59 c. 59:82 9.3 Eofimofio: 5 83m 52215 8 Ba o: £386 9: a 38% 8589: _o cease ago: 835:8 8532.: ...: 28.. 9S an Ov- Em 0° hum N0 .20 wOm <5 o>o 2. 38:8 3 3 33:8 3 OF F.N.FmN..F8 oF FmFmNFNo oF FmFmNFmo .o.F.. F.¢.FNN..F,8 , oF FNFNNFS or. [QMNBO o— F3330 ta .N 28 2: 628. .2 2.3 96 : 9. or 5?:5 D.— 3th: :4 ID v- It)“: 0 '- F . O N GO ON Ow —O\O:O_. 3.— gm own O O Ov : Ow wQOCOF :4 50m O Ow NV ON Ow FOE—ko— 2.. OON O a ON an O p rm? :2. 3.— .VO O F o O O _, 332° F 3.. mm OOF O N-. Ow —. Ow wOEQOw D..— O Q «macaw F “mafia? é..>< £28 2.... am fiESO .2 29¢ 97 over 12 times to the greatest average biomass of all stations to 2,691 mg/m2 on August 26th (Table 15). The large increase in both abundance (Table 3 and Table 4) and average biomass over replicates (Table 15) may illustrate that Grand Traverse Bay lags behind Lake Michigan in the timing of plankton blooms and that this study was able to capture the first population peak of zooplankton in Grand Traverse Bay in the 1991 season. It may also demonstrate that Grand Traverse Bay, like other embayments in Lake Michigan, may have a higher productivity than the other nearshore waters of Lake Michigan. In 1990, the year previous to the study, nearshore water temperatures at numerous sites on Lake Michigan had already increased above 4 °C by the end of April, and Grand Traverse Bay was still at 4°C throughout the water column in the shallow nearshore waters at the southern end of the bay, where the first warming of the bay would be expected. Warming of Grand Traverse Bay is inhibited by depth and a lack of water movement with the open waters of Lake Michigan (Lauff 1957). Spring mixing, and the subsequent primary and secondary production, occurs much later in Grand Traverse Bay than in Lake Michigan (Lauff 1957). MW Ludington was the only station that was sampled more than twice and will be used to examine seasonal community biomass trends in 1991. The 30 meter contour at Ludington showed the greatest seasonal decline in biomass. The first samples taken (on 7/17/91) was the season high average biomass (over replicates) for the 30 meter site with 1523 mg/m?. The low for Ludington was on October 3rd (454 mg/mZ) and the last sample on November 14th, the average biomass rose slightly to 580 mg/m2 98 (Table 15). The 10 meter contour at Ludington also declined in biomass as the season progressed. An outlier to this declining seasonal trend was the highest average biomass at the shallower station in Ludington on October 18th (537 mg/m?). On that day, Diaptomus sp., Cyclops sp., B. Iongirostris, and B. cederstroemi all greatly increased in abundance from the previous sample at the site just 15 days earlier (Table 3 and Table 4). The increase was too sudden to be explained by growth or reproduction in the cold waters (13 'C), and most likely reflects a combination of offshore and nearshore animals that were concentrated by stormy weather that day, because the 30 meter sample could not be taken that day because of high seas. The decreasing average biomass in Ludington reflects the decrease in abundance of all species (Table 3 and Table 4). This decline may indicate that the 1991 field season either started at peak productivity or some time after maximum abundance. In 1974, Duffy (1975) studied the nearshore zooplankton adjacent to the Ludington Pumped Storage facility. The sampling for that study started in June and the numbers of zooplankton (#/m3)‘were similar to the present study except the samples in June found about ten times as many Cyclops sp. than in any other samples of Duffy's, or this study. This considerable increase to the zooplankton community from a single species suggests that the Duffy (1975) study may have sampled the spring peak of zooplankton abundance and the present study did not (Figure 39). 99 wt: 28 83F _ m FFmo vao mNFmo #5: i=8 Ba 58: >3.» £5 50289 an EEO $0595an 0330 coscmaoon .89 8553.. F0 comtcano .8 059.... 0 {mo FoFmo omko Es SRO 8F8 EON III. EN_. 1?. who F 80-. 3.0.! E00 IT 52 om oo .05 (peqno 191ew/suenpwpul spuesnom) SUMMARY W93. The zooplankton community was dominated by three species/groups, Diaptomus sp., Cyclops sp., and Bosmina longirostris. However, because Diaptomus and Cyclops were groups of possibly many species, it can only be concluded that the two families, Diaptomidae and Cyclopoidae and one species, B. longirostris dominated. Abundances of species/groups were found in greater number at the deeper contour even if the zooplankton were measured as individualsqx When replicates were averaged, and same day comparisons made between depths, the 30 m:10 m ratio was 7.81 for the areal abundance and 2.60 for the cubic abundance. The daphnids, D. galeata, D. retrocurva, D. pulicaria, and one copepod Limnocalanus macrurus were found at much greater abundances at the deeper station. Species Associations The proposed species associations utilized the first two PC's. Although the first PC of to both PCA's, #/m? and #/m?, were significantly correlated to both month and depth, the split appears to be more influenced by depth than by month. All of the positively correlated species of PC 1 are more abundant at the 30 meter contour and negatively correlated species, more abundant at 10 meters. The second PC split was less definitive and involved different degrees of relative abundance as the season progressed and water temperatures became cooler. One group included the species, D. 100 101 retrocurva and L. kindti, (and E. coregoni only with the cubic PCA) which were characterized as declining markedly as the season progressed. The other group was characterized as having abundances that did not decline rapidly with season. All the species except one, D. pulicaria, did not decline appreciably as the season progressed. D. pulicaria declined precipitously with season. Because of this D. pulicaria could not be included in the limnetic persistent species association. Regarding the attempt to twice split the species into groups, it is deduced that the second split is inconclusive, and only the first split is acceptable. Differences between the analyses of the areal and cubic data were minor, and because of this the analysis could have been done adequately with one measurement or the other. Dry Weight Biomass In almost every case the 30 meter contour total biomass (mg/m?) is at least 3 times that of the 10 meter contour. A declining seasonal trend is also evident at both contours. The total biomass at the 10 meter stations declines very gradually. The 30 meter contour decline is much steeper (Table 15), and between August 16 (1469 mg/m?), and October 3 (33 mg/mz), the decline is sharp. W Peak zooplankton biomass values at Grand Traverse Bay exceeded peak biomasses at both Ludington and Manitou Passage, but were less than the biomass peaks of other nearshore studies (Evans et al. 1980, Roth and Stewart 1973, Gannon 1972) in southern Lake Michigan. This indirectly supports the generalization by Stoermer et a1. (1972), that the primary productivity and phytoplankton standing stock in Grand Traverse Bay were 102 intermediate between high values of inshore waters in Lake Michigan and those of the offshore. W In 1987, on Lake Michigan (Lehman 1991), Bythotrephes was characterized as more abundant (#/m3) offshore (>40 m) than inshore. Lehman suggested that a gradient of fish planktivory exists from inshore to offshore, and that this factor may be responsible for the lower abundances nearshore. It is of interest that this new invading species, could not be characterized by PCA as occurring in greater numbers at the deeper stations. High abundances (>100/m2) occurred sporadically from mid-July to mid-August, particularly at the deeper station at Ludington, Manitou Passage, and West Grand Traverse Bay (Table 3). High abundances were also found at the shallower station in October and November. A possible explanation for the similarity of abundances (>100/m2) between the nearshore of this study and the offshore (Lehman 1991) may be that wave and wind actions concentrate the animals in the nearshore, and sampling was done before the dispersal of the physically developed swarm. In July 1991 samples were taken when waves were 2-4 feet, and winds were out of the southwest at 5-15 knots. Wave and wind action, however cannot explain the high average abundances at 30 meters in mid- August at Manitou Passage (234/h3 on 8/10) when waves were 1-2 feet and the wind was only 0-5 knots. Another proposed suggestion may be that in 1987, B. cederstroemi might not have reached equilibrium yet, or that the species had not yet increased to the point where the predation by fish could no longer control the population in the nearshore region. 103 Bythotrephes also could not be characterized as having a seasonal gradient from July to November, or a summer maximum density. Sprules et.al. (1990) reported maximum densities of B. cederstroemi in Lake Michigan in 1987 as 29/m9‘during August and September, and Mordukhai- Boltovskaya (1958) reported August maximum densities in Rybinsk reservoir in the USSR as 44.3/mP. The maximum values at the two depths of this study were found not in mid-summer, but at the beginning and end of the sampling season. Densities were averaged over replicates, and resulted in 20.0/m9 on July 17 at 30 meters, and 42.6/m9‘on November 14 at 10 meters (Table 4). Densities in August and September were for the most part much lower in these months. Except for mid-August at Ludington, and at 30 m in early August at Manitou Passage and West Grand Traverse Bay, densities were less than 5/m3. Also of interest is the large proportion of biomass Bythotrephes is estimated to contribute. Bythotrephes abundance, on average, was less than 0.1% of the total abundance, yet was fifth highest in average biomass, comprising 8.4% of the total biomass. Copepod Dominance Cladocerans, primarily the smaller cladoceran Bosmina longirostris, reached summer maximums and dominated over all species in southeastern Lake Michigan (Evans et.al. 1980, Roth & Stewart 1973), and off of Milwaukee Wisconsin (Gannon 1972). Dominance by Bosmina may be the result of a more prolific nearshore phytoplankton population. Nearshore phytoplankton standing stocks were reported to be higher than offshore in southeastern Lake Michigan by Ayers & Siebel (1973), and inshore carbon fixation (measured as mgC/maflhr), is twice that of the offshore (Schelske and Callender 1970). 104 Gannon (1975) described Bosmina as the best adapted organism to rapidly respond to nutrient loading conditions in the nearshore waters. Diaptomus sp. dominated over Bosmina, or at the least, copepods dominated over cladocerans in both abundance (Table 3 and Table 4) and biomass (Table 15) at both depths and throughout the season in 1991. In the earlier study (1974) at Ludington (Duffy 1975), this same dominance by copepods also existed. Although Bosmina dominated once on July 1, 1974 at the shallower station (12 m), the seasonal peak in Ludington that year was attributed to a large abundance of the copepod Cyclops sp., not Bosmina. Diaptomids use chemoreception (smell) and are more selective than cladocerans, preferring phytoflagellates over the less edible blue-green and green algae (Sterner 1989). Cladoceran morphology, with their appendages enclosed in a carapace, is a source of interference that limits their selectivity. Cladocerans use their thoracic legs to produce a constant current of water between the valves of the carapace (Pennak 1989), and has a much higher ingestion rate of phytoplankton than do the Diaptomids (Scavia et.a1. 1988). Theoretically, these differences in preferences and feeding rates can affect the seasonal succession of phytoplankton (Sterner 1989). B. longirostrls responds to local conditions of temperature and food abundance that controls growth rate and fecundity (it determines a monocyclic or dicyclic reproductive pattern) (Balcer et a1. 1984). Therefore, low abundance of phytoplankton will result in lower abundances of cladocerans, particularly Bosmina. APPENDICES 105 APPENDIX A. Temperature Profiles. Flgure 40. Temperature (°C) proflle at 30 meter depth at Ludington on August 16. 1991. 0. ‘O N 0. "(O ,_ 0 o 2., ’9‘. 92 23 H CO L— (D 8 .9 <0 (n l- -0. <- -0. O l J l l 14" 1 A O N V (O (D O ' N v- 1— (919mm) utdag 106 APPENDIX A. Temperature Profiles. Figure 41. Temperature (°C) proflle at 30 meter depth at Ludlngton on September 13. 1991. Temperature (”0) 8.0 4.0 0.0 l I l l l l 1 $- '1‘ to 00 v- (smew) utdea 107 APPENDIX A. Temperature Profiles. Flgure 42. Temperature (°C) proflle at 30 meter station (over 110 m of water) at West Grand Traverse Bay on August 26. 1991. 8.0 4.0 0.0 l l 1 l J l l v o N v (0 co 2 9‘. T. (318190.!) utdec] Temperature (°C) 108 APPENDIX A Temperature Profiles. Flgure 43. Temperature (°C) proflle at 30 meter station at Manltou Passage on August 27. 1991. \ \9 ”O (\I O. ”(O ,_ O. 53, T— t— 3 H (U L— (D E _O. m m I'— -0. <- -0. O l l J l l J O N o m