Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Ecological links between emergent macrophytes and associated periphyton and benthic communities in a… Wilson, Sandra Joan 2006

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2006-0344.pdf [ 18.58MB ]
Metadata
JSON: 831-1.0092643.json
JSON-LD: 831-1.0092643-ld.json
RDF/XML (Pretty): 831-1.0092643-rdf.xml
RDF/JSON: 831-1.0092643-rdf.json
Turtle: 831-1.0092643-turtle.txt
N-Triples: 831-1.0092643-rdf-ntriples.txt
Original Record: 831-1.0092643-source.json
Full Text
831-1.0092643-fulltext.txt
Citation
831-1.0092643.ris

Full Text

Ecological Links between Emergent Macrophytes and Associated Periphyton and Benthic Communities in a Coastal Reservoir Littoral Zone by SANDRA JOAN WILSON B .Sc , University of British Columbia, 1992 A THESIS SUBIMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Resource Management and Environmental Studies) UNIVERSITY OF BRITISH COLUMBIA April 2006 © Sandra Joan Wilson 2006 Abstract A study conducted in Stave Reservoir, near Mission British Columbia, examined the linkages between flooded shoreline vegetation and associated periphyton and benthic communities in a coastal reservoir drawdown zone. Two native perennial species, woolgrass (Scirpus cyperinus) and lenticulate sedge (Carex lenticularis), and an annual agronomic species, fall rye (Secale cereale L.) were used in the study. A total of 108 pots containing either one of the plant species or barren substrate (i.e. no vegetation) were installed within three elevation bands in the inundated reservoir drawdown zone in June 2000. Samples were retrieved monthly over the period of inundation, with the final samples being removed in early September 2000. The effects of planting elevation, vegetation species and time on endpojnts including plant biomass and nutrient composition, periphyton density and biovolume, and benthos abundance were examined. Fall rye decomposed quickly, losing approximately 80% of its foliar biomass after one month of submergence. Woolgrass and sedge foliage decomposed much more slowly, losing only 30% and 55%, respectively, of their aboveground biomass by the final sample date. There was evidence that nitrogen was translocated from shoots to roots fairly quickly upon inundation in all three plant species. Phosphorus was lost rapidly from the fall rye foliage with no concurrent increase in the roots, suggesting that rather than being translocated to the roots, phosphorus was leached from the plant material. No trends were noted for phosphorus for the perennial species. The periphyton communities associated with all plants types were dominated by diatoms. The maximum diatom density was approximately 112,000 cells/m 2 observed in the final sample period. Oligochaetes and chironomids were the most dominant benthic taxa, accounting for 63% of all benthic organisms. The presence of vegetation increased the number of benthic organisms by 3 times, and benthic taxa by 1.7 times in comparison to barren substrate. There was no significant difference in total benthos abundance between the three plant species; however, distinct groupings of benthic community composition associated with the control, fall rye and perennial i i plant species were noted using the nonmetric multidimensional scaling (NMS) ordination technique. i i i Table of Contents Abstract Table of Contents i v List of Tables vii List of Figures x List of Illustrations xv Acknowledgements xvi 1.0 Introduction 1 1.1 Background 1 1.1.1 Objectives of this study 2 1.1.2 Research Questions 2 1.1.3 Significance of this research 4 1.2 Literature Review 5 1.2.1 Storage Reservoirs vs. Lakes and Rivers 6 1.2.2 Vegetation: Physiological reaction to flooding 8 1.2.3 Nutrient translocation and cycling 12 1.2.4 Microbial Loop 14 1.2.5 Aquatic Invertebrates 19 2.0 Methods and Materials 26 2.1 Experimental Procedure 26 2.2 Study Site 27 2.3 Methods 29 2.3.1 Preparation of vegetation samples 29 2.3.2 Installation of pots 29 2.3.3 Retrieval of pots 30 2.3.4 Environmental Data 31 2.3.5 Processing Samples 31 2.3.6 Foliar and Root Nutrient Analysis 31 2.3.7 Periphyton Analysis 32 2.3.8 Benthos counting methodology 34 2.4 Data analysis 3 5 2.5 Statistical Analysis 3 5 2.5.1 Experimental Design 35 IV 2.5.2 Analysis 35 3.0 Results 39 3.1 Environmental Data 39 3.1.1 Water Level 39 3.1.2 Temperature 40 3.1.3 Light 40 3.2 Vegetation Data 41 3.2.1 Biomass 41 3.2.2 Nutrient Content 46 3.3 Periphyton Data 61 3.3.1 Community Composition 61 3.3.2 Diatom Density 64 3.3.3 Diatom Biovolume 70 3.3.4 Diatom Taxonomic Richness 76 3.4 Benthic Organisms 78 3.4.1 Community Composition 78 3.4.2 Benthos Abundance 81 3.4.3 Benthos Taxonomic Richness 88 3.5 Relationship between diatom biovolume and benthos density 91 4.0 Discussion 94 4.1 Environmental data 94 4.2 Vegetation 95 4.2.1 Biomass 95 4.2.2 Nutrient cycling 99 4.3 Periphyton 102 4.3.1 Density 102 4.3.2 Taxonomic richness 105 4.3.3 Changes in community composition 106 4.3.4 Biovolume 107 4.4 Benthic Community 109 4.4.7 Abundance 109 4.4.2 Taxonomic Richness 112 4.4.3 Size of animals 113 4.4.4 The relationship between diatoms and benthos 113 v 5.0 Conclusions 114 5.1 Summary and recommendations for reservoir management 116 6.0 References 119 Appendix A. Plant Biomass, Nitrogen and Phosphorus Data 133 Appendix B. Complete ANOVA tables for Plant Analyses 138 Appendix C. Raw Data for Calcium, Magnesium and Potassium in Above and Belowground Plant Biomass 141 Appendix D. Periphyton Density Summary Data By Sample Period 147 Appendix E. Periphyton Biovolume Summary Data By Sample Period 153 Appendix F. Complete ANOVA tables for analyses related to periphyton 159 Appendix G: The Nonmetric Multidimensional Scaling Ordination Technique 160 Appendix H: NMS ordination diagrams for periphyton density analysis 168 Appendix I: NMS ordination diagrams for diatom biovolume analysis 179 Appendix J : Benthos Summary Data 188 Appendix K: Complete ANOVA tables for analyses related to benthos 194 Appendix L: NMS ordination diagrams for benthos abundance analysis 196 Appendix M: Complete ANOVA table for diatom to benthos ratio 205 vi List of Tables Table 1. ANOVA table for the significant interaction between species and time upon mean foliar N (%) (note: data was arcsine transformed) 47 Table 2. ANOVA table for the significant effect of species and time for root N% (arcsine transformed) content 49 Table 3. The A N O V A table for the significant interaction between species and time upon nitrogen rootshoot ratio 52 Table 4. Least square means of the nitrogen Root to Shoot ratio showing statistically significant designations of species by time 53 Table 5. ANOVA table for the significant interaction between species and time upon mean foliar P (%) (arcsine transformed) 55 Table 6. ANOVA table for the significant interaction between species and time upon mean root P (%) (arcsine transformed) 57 Table 7. Least square means of root P% data showing statistically significant designations of species by time 58 Table 8. ANOVA table for significant interaction between species and time on mean phosphorus R:S ratio 59 Table 9. Least square means of the phosphorus Root to Shoot ratio showing statistically significant designations of species by time 61 Table 10. List of periphyton taxa, and associated biovolume per cell, observed during the study. Stave Reservoir, 2000 62 Table 11. ANOVA table for the significant effect of time upon mean log-diatom density 65 Table 12. A N O V A table for the significant interaction of elevation and time upon mean diatom biovolume 71 Table 13. Least square means of the diatom biovolume data (urn3/cm2) 73 Table 14. ANOVA table for significant interaction between elevation and time upon mean diatom log-taxonomic richness associated with woolgrass and sedge vegetation 77 Table 15. List of benthic taxa associated with all foliage and root samples, and barren samples in Stave Reservoir, 2000 79 Table 16. ANOVA table for the significant effects of species and time upon mean benthos log-abundance 83 Table 17. A N O V A table for the significant interactions upon mean log-density of benthic organisms (log-number of individuals/gram of plant material) 84 Vll Table 18. ANOVA table for the significant interaction between elevation and time and the main effect, species, upon the mean number of benthic taxa 89 Table 19. Least Square Means of the benthic organism taxa data showing statistically significant designations of elevation by time 90 Table 20 . ANOVA table for significant effect of Time on the ratio of diatom biovolume to benthos density 92 Table A-1 Aboveground Plant Biomass, Nitrogen and Phosphorus Data 133 Table A-2 Belowground Plant Biomass, Nitrogen and Phosphorus Data 135 Table B-1. Complete A N O V A table for foliar N (%) arcsine transformed) content 138 Table B-2. Complete A N O V A table for root N (%) arcsine transformed) content 138 Table B-3. Complete A N O V A table for Foliar P (% arcsine transformed) content 139 Table B-4. Complete A N O V A table for Root P (% arcsine transformed) content 139 Table B-5. Complete A N O V A table for Foliar N (% arcsine transformed) content 139 Table B-6. Complete A N O V A table for Root N (% arcsine transformed) content 140 Table B-7. Complete A N O V A table for N% Root to Shoot ratio 140 Table B-8. Complete A N O V A table for Foliar P (% arcsine transformed) content 140 Table B-9. Complete A N O V A table for Root P (% arcsine transformed) content 140 Table B-10. Complete A N O V A table for P root to shoot ratio (arcsine transformed) 140 Table C-1. Aboveground Plant Calcium, Magnesium and Potassium Raw Data 141 Table C-2. Belowground Plant Calcium, Magnesium and Potassium Raw Data 144 Table D-1. Average periphyton density (cells/cm2) estimates associated with woolgrass and sedge foliage at each elevation on July 6, 2000 147 Table D-2. Average periphyton density (cells/cm2) estimates by associated with woolgrass and sedge foliage at each elevation on August 9, 2000 149 Table D-3. Average periphyton density (cells/cm2) estimates associated with woolgrass and sedge foliage on August 28, 2000 for 80 m samples and September 8, 2000 for 76 m and 78 m samples 151 viii Table E-1. Average periphyton biovolume (•m 3 /cm 2 ) estimates associated with woolgrass and sedge foliage at each elevation on July 6, 2000 15: Table E-2. Average periphyton biovolume (um3/cm2) estimates associated with woolgrass and sedge foliage at each elevation on August 9, 2000 15! Table E-3. Average periphyton biovolume (|im3/cm2) estimates associated with woolgrass and sedge foliage on August 28, 2000 for 80 m samples and September 8, 2000 for 76 m and 78 m samples 15' Table F-1. Complete A N O V A table for diatom density (log10(x) transformed data) 159 Table F-2. Complete A N O V A table for diatom biovolume (square root transformed data).... 159 Table F-3. Complete A N O V A table for diatom species richness (Iog10(x) transformed data). 159 Table J-1. Number of benthic organisms associated with woolgrass, sedge, fall rye and control samples 188 Table K-1.-Complete A N O V A table for benthic organism abundance (log10(x) transformed data) 194 Table K-2. Complete A N O V A table for benthic organism abundance per gram of plant material (log10(x) transformed data) 194 Table K-3. Complete A N O V A table for benthic taxonomic richness analysis (non transformed data) 195 Table M-1. Complete A N O V A table for ratio of diatom biovolume to benthos density 205 ix List of Figures Figure 1. Schematic of a food web for the littoral zone of Stave Reservoir. Under oligotrophic conditions, the microbial loop plays a key role in the transfer of nutrients to higher trophic levels 17 Figure 2. Schematic of an aquatic food chain under mesotrophic conditions. The size of the arrows is indicative of the relative importance of the flow of carbon between the two systems. 18 Figure 3. Schematic of experiment layout for Stave Reservoir vegetation submergence 26 Figure 4. Stave Reservoir Vegetation Study Sample Sites 28 Figure 5. Stave Reservoir daily mean surface water elevation June 1 to September 15, 2000. 39 Figure 6. Daily mean water temperature by elevation, Stave Reservoir 2000 40 Figure 7. Irradiance by depth on sampling dates, Stave Reservoir, 2000 41 Figure 8. Mean aboveground biomass (± SE) for three plant species over time (n=99) 42 Figure 9. Mean belowground biomass (AFDW) (± SE) for three plant species over time (n=99). 44 Figure 10. Mean root to shoot biomass ratio (± SE) for three plant species over time (n=99).45 Figure 11. Mean foliar N (%) (±SE) (non-transformed) for all three plant species over time (n=99) 46 Figure 12. Profile plot comparing the least square means of foliar N (%) fall rye, sedge and woolgrass over time (n = 99) 48 Figure 13. Mean root N (%) (±SE) (non-transformed data) for all three plant species overtime (n=99) 49 Figure 14. Profile plot comparing the least square means of root N (%) content pre and post-inundation for all three plant species (n=99) 50 Figure 15. Profile plot comparing the least square means of root N (%) for fall rye, sedge and woolgrass, pre and post-inundation (n=99) 51 Figure 16. Mean nitrogen root to shoot ratio (±SE) (non transformed data) for all three plant species over time (n=99) 51 Figure 17. Profile plot of the interaction effect of species and time upon the mean nitrogen Root to Shoot ratio for all three plant species (n=99) 53 Figure 18. Mean foliar P (%) (±SE) content (non-transformed data) for all three plant species over time (n=99) 54 x Figure 19. Profile plot of the interaction effect of species and time upon mean foliar P (%) (n=99) 55 Figure 20. Mean Root P (%) (±SE) content (non-transformed data) for all three plant species overtime (n=99) 56 Figure 21. Profile plot showing the interaction effect of species and time upon root P%, taking pre and post-inundation data into account (n = 99) 57 Figure 22. Phosphorus root to shoot ratio (±SE) (non transformed data) for all three plant species over time (n=99) 59 Figure 23. Profile plot showing the interaction of species by time effect on the phosphorus root to shoot ratio (n = 99) 60 Figure 24. Mean diatom density (cells/cm2) associated with woolgrass and sedge over time (n=54) 65 Figure 25. Profile plot showing time effect upon mean diatom density (cells/cm2) associated with woolgrass and sedge 66 Figure 26. Profile plot showing elevation effect upon mean-diatom density (cells/cm2) associated with woolgrass and sedge 66 Figure 27. Dominant diatom species by density (cells/cm2 of plant material) associated with woolgrass and sedge vegetation 67 Figure 28. Two-dimensional ordination of diatom density showing groupings by time. Distances between sample units approximate dissimilarity in species composition 69 Figure 29. Mean diatom biovolume (um 3/cm 2 plant material) (±SE) (non-transformed data) associated with woolgrass and sedge overtime (n=54) 71 Figure 30. Profile plot showing the interaction of elevation and time upon mean diatom biovolume (um 3/cm 2 plant material) associated with woolgrass and sedge (n= 54) 72 Figure 31. Dominant diatom taxa by biovolume (um 3/cm 2 plant material) associated with woolgrass and sedge plant species 74 Figure 32. Two-dimensional ordination of diatom biovolume showing groupings by time. Distances between sample units approximate dissimilarity in species composition 75 Figure 33. Diatom taxonomic richness associated with woolgrass and sedge (n=54) 77 Figure 34. Profile plot showing the interaction effect of elevation by time on mean diatom taxonomic richness 78 Figure 35. Dominant benthic organisms by abundance associated with all three plant species in Stave Reservoir, 2000 81 xi Figure 36. Mean benthos abundance (benthic organisms/sample) (±SE) for all three plant types and control samples over time (n = 108) 82 Figure 37. Profile plot comparing mean total abundance of benthic organisms associated with control and plant samples (n = 108) 83 Figure 38. Profile plot showing mean abundance of benthic organisms over time (n=108).... 84 Figure 39. Profile plot showing the interaction of elevation by time upon mean benthic organism density 85 Figure 40. Profile plot showing the interaction effect of species by substrate on benthic organism abundance per gram of plant material 86 Figure 41. Ordination for non-transformed benthos abundance data showing a strong grouping by plant species. Distances between sample units approximate dissimilarity in species composition. The ellipses show approximate groupings 87 Figure 42. Profile plot showing interaction between elevation and time upon the mean taxonomic richness of the benthic communities associated with all three plant species and control samples (n = 108) 90 Figure 43. Profile plot showing mean number of taxa associated with each plant species and control samples (n = 108) 91 Figure 44. Ratio of diatom biovolume (pm3/cm2) to benthos density (organisms/gram plant material) (N= 54) 92 Figure 45. Profile plot showing the effect of time on the ratio of diatom biovolume to benthos abundance for communities associated with woolgrass and sedge 93 Figure 46 Changes in Total Diatom Density on glass slides at 1 m in Lake 240 in the Experimental Lakes Area, Northwestern Ontario (after Stockner and Armstrong 1971) 104 Figure G-1. PC-ORD output plot of stress (expressed as a percentage) vs. iteration number. The final solution is achieved when stress levels no longer decrease appreciably. In this ordination, the lowest stress level was reached after 35 iterations. The plot also indicates that the solution is relatively stable as the curve settles on an even stress level and does not fluctuate erratically 162 Figure G-2. A two dimensional ordination, showing axis 2 vs. axis 1 (Fig. G-2a), axis 3 vs. axis 1 (Fig. G-2b), and axis 3 vs. axis 2 (Fig. G-2c), of benthos abundance in species space. Distances between sample units approximate dissimilarity in species composition 165 Figure G-3. Overlay plot with sizes of symbols proportional to the magnitude of the variable, in this case, abundance of Nematoda in relation to time from this thesis 166 Figure G-4. PC-ORD output for coefficients of determination (r2 values) for the correlations between ordination distances and distances in the original n-dimensional space, in this case for the invertebrate abundance data from this thesis 167 xii Figure H-1. Two-dimensional ordination of diatom density by elevation (m) (Fig. H-1a) and plant species (Fig. H-1b). Distances between sample units approximate dissimilarity in species composition 168 Figure H-2. NMS ordination diagrams for Frustrulia sp., accounting for 22% of all diatoms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of Frustrulia sp. The different symbols in the ordination diagrams represent different sample times (Figure H-2a), elevation (m) (Figure H-2b) or plant species (Figure H-2c) 170 Figure H-3. NMS ordination diagrams for Achnanthes minutissima, accounting for 9% of all diatoms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of A. minutissima . The different symbols in the ordination diagrams represent different sample times (Figure H-3a), elevation (m) (Figure H-3b) or plant species (Figure H-3c) 172 Figure H-4. NMS ordination diagrams for Frustrulia rhomboides, accounting for 8% of all diatoms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of F. rhomboides. The different symbols in the ordination diagrams represent different sample times (Figure H-4a), elevation (m) (Figure H-4b) or plant species (Figure H-4c) 174 Figure H-5. NMS ordination diagrams for Tabellaria flocculosa, accounting for 8% of all diatoms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of T. flocculosa. The different symbols in the ordination diagrams represent different sample times (Figure H-5a), elevation (m) (Figure H-5b) or plant species (Figure H-5c) 176 Figure H-6. NMS ordination diagrams for Tabellaria fenestrata, accounting for 7% of all diatoms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of T. fenestrata. The different symbols in the ordination diagrams represent different sample times (Figure H-6a), elevation (m) (Figure H-6b) or plant species (Figure H-6c) 178 Figure 1-1. Two-dimensional ordination of diatom biovolume. Note that there are no clear groupings by elevation (m) (Fig. 1-1 a) or plant species (Fig. 1-1 b). Distances between sample units approximate dissimilarity in species composition 17$ Figure I-2. NMS ordination diagrams for Frustrulia sp., accounting for 26% of all diatoms by biovolume. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of Frustrulia sp. The different symbols in the ordination diagrams represent different sample times (Figure l-2a), elevation (m) (Figure l-2b) or plant species (Figure l-2c) 18 Figure I-3. NMS ordination diagrams for Frustrulia rhomboides, accounting for 15% of all diatoms by biovolume. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of Frustrulia rhomboides. The different symbols in the ordination xiii diagrams represent different sample times (Figure l-3a), elevation (m) (Figure l-3b) or plant species (Figure l-3c) 183 Figure I-4. NMS ordination diagrams for Eunotia sp., accounting for 15% of all diatoms by biovolume. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of Eunotia sp. The different symbols in the ordination diagrams represent different sample times (Figure l-4a), elevation (m) (Figure l-4b) or plant species (Figure l-4c) 18E Figure I-5. NMS ordination diagrams for Tabellaria fenestrata, accounting for 14% of all diatoms by biovolume. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of T. fenestrata. The different symbols in the ordination diagrams represent different sample times (Figure l-5a), elevation (m) (Figure l-5b) or plant species (Figure l-5c) 187 Figure L-1. Ordination diagram for benthos abundance data. Note that there is a weak grouping by time (Fig. L-1 a) and no clear grouping by elevation (m) (Fig. L-1b). Distances between sample units approximate dissimilarity in species composition 196 Figure L-2. NMS ordination diagrams for subclass Ostracoda, accounting for 22% of all benthic organisms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of Ostracoda individuals. The different symbols in the ordination diagrams represent different sample times (Figure L-2a), elevation (m) (Figure L-2b) or plant species (Figure L-2c) 198 Figure L- 3. NMS ordination diagrams for family Enchytraeidae (Class Oligochaeta), accounting for 21% of all benthic organisms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of individuals from the family Enchytraeidae. The different symbols in the ordination diagrams represent different sample times (Figure L-3a), elevation (m) (Figure L-3b) or plant species (Figure L-3c) 200 Figure L-4. NMS ordination diagrams for Phaenopsectra, the most abundant chironomid taxa noted during this study. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of Phaenopsectra. The different symbols in the ordination diagrams represent different sample times (Figure L-4a), elevation (m) (Figure L-4b) or plant species (Figure L-4c) 202 Figure L-5. NMS ordination diagrams for Phylum Nematoda, accounting for 8% of all benthic organisms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of nematodes. The different symbols in the ordination diagrams represent different sample times (Figure L-5a), elevation (m) (Figure L-5b) or plant species (Figure L-5c) 204 xiv List of Illustrations Plate 1. Representative woolgrass samples prior to submergence (left) and at T1 (right) 42 Plate 2. Representative sedge samples prior to submergence (left) and at T1 (right) 43 Plate 3. Representative fall rye samples prior to submergence (left) and at T1 (right) 43 Plate 4. Typical periphyton assemblages associated with plant material. Photo taken at 600x magnification 63 Plate 5. Typical periphyton assemblages associated with plant material. Photo taken at 300x magnification 64 XV Acknowledgements First and foremost, I would like to thank the members of my thesis advisory committee: Dr. Ken Hall (advisor) and Dr. Les Lavkulich from UBC, Dr. Will Carr (Carr Environmental Consultants); Chris Perrin (Limnotek Research and Development); and Dr. John Stockner (Eco-logic Ltd.) for their guidance, patience and advice. I am grateful to BC Hydro for providing funding for this project, and for providing reservoir elevation information and temperature monitoring records. I would also like to thank several individuals at BC Hydro for their support over the course of completing this thesis, including but not limited to Carol Lamont, Ed Hill, Greitje Van Dijk, Randy Bourne, Dave Cattanach, and Glen Singleton. I am indebted to many other individuals who assisted with various aspects of this project including: Tim Lissimore, Julie Beer, Glen Roberts, and Ian Wright who all assisted at various times in the field or lab; Danusia Dolecki who completed the invertebrate speciation and provided advice on periphyton analysis; SCUBA divers from Foreshore Technologies who installed and collected the vegetation samples in the field; Bev Herman and the rest of the staff at Pacific Soil Analysis Incorporated for vegetation nutrient analysis; and Nick Page for advice on statistical analysis using ordination techniques. Assistance from the Statistical Consulting Service at Simon Fraser University in Burnaby, B.C. regarding ANOVA was greatly appreciated. Special thanks go to Julie for her friendship and advice along the way. Last, and definitely not least, I would like to thank my husband Glen, and daughters Carly and Jenna for their love, encouragement and support. xvi 1.0 Introduction 1.1 Background Reservoirs generally follow a classic pattern where immediately after inundation, and for the following approximately 5 to 10 years they are highly productive. During this time period, nutrients from the decay of terrestrial vegetation and insects, as well as release of nutrients from the soil result in a tremendous trophic upsurge (Northcote and Atagi 1997). This source of nutrients, however, is exhausted within a few years and the system begins to 'crash' without an external input of nutrients (Baxter 1977; Thornton et al. 1990; Marotz er al. 1994; Soballe et al. 1992; Stockner er al. 2000; Stockner et al. 2005). Planting of shoreline vegetation in drawdown zones has been used to address a number of issues in British Columbia reservoirs. In Arrow and Williston reservoirs, vegetation has been planted to combat dust storms which are created when exposed shoreline sediments are mobilized by wind (Carr et al. 1993), and to improve aesthetics. More recently, a shoreline revegetation program was implemented at Stave Reservoir as a compensation measure to address a potential loss of biological productivity due to changes in the operating regime of the reservoir (Fisheries and Oceans Canada (DFO) 1999). The longest established reservoir shoreline planting program in British Columbia is the dust control program in the Arrow Reservoir, located in the interior of the province near Revelstoke. The full-scale program, including planting of the annual fall rye as well as a variety of perennials, began in 1990. Since then, a well-developed perennial community, consisting mainly of sedges and reed canary grass has become established, and in many areas fall rye is no longer planted. Planting of fall rye tends to expand the natural vegetation community in at least two ways: the act of mechanically drilling the fall rye seed into the substrate causes the seeds of native plants to also be incorporated into the sediment, thereby encouraging expansion into fall rye planted areas. Secondly, once the fall rye area is flooded and the aboveground biomass begins to decompose, a coarse mat of stubble and root mass remains (A. Moody, AIM Ecological Consultants, pers. comm.). This stubble mat provides a cover for native seeds and acts to stabilize the substrate. In addition to improved aesthetics and dust control, local residents have reported increased bird, wildlife and terrestrial insect activity, as well as improved angling in vegetated areas (B. Gadbois, BC Hydro, pers. comm.). 1 Nutrient loading from the decomposition of plant material after inundation may increase trophic production, ultimately increasing the availability of food for fish. The greatest area of influence is likely to be in the littoral zone itself; however the pelagic community may also realize some benefit. 1.1.1 Objectives of this study The first objective of this study was to evaluate the effects of flooding on three different vegetation species over time in the drawdown zone of Stave Reservoir. Vegetation species included in the study were lenticulate sedge, woolgrass, and fall rye. Lenticulate sedge and woolgrass are perennial species native to the Stave Reservoir drawdown zone. Fall rye, an annual agronomic species, has been used successfully in other British Columbia reservoirs for dust control, and was included in this study to examine its potential to increase biological productivity in the reservoir drawdown zone. The second objective was to examine the ecological links between the three vegetation species and the associated periphyton and benthic communities in the drawdown zone of Stave Reservoir. A set of five specific research questions were developed to meet these objectives. To address each question, the effects of planting elevation, plant species and time on endpoints important to biological productivity, as detailed in the following section, were tested. 1.1.2 Research Questions 1. Are there differences between the three plant species in terms of decomposition rates and nutrient release? To address this question, the effects of planting elevation, plant species, and time on the following endpoints were tested: • above and belowground biomass of the three plant species • biomass root to shoot ratio over time • percentages of nitrogen (N) and phosphorus (P) in foliage and roots 2 • N and P root to shoot ratios 2. Are there differences in the periphyton communities that colonize the plant species? To address this question, the effects of planting elevation, plant species, and time on the following endpoints were tested: • diatom density • diatom biovolume • diatom taxonomic richness The Nonmetric Multidimensional Scaling (NMS) ordination technique was used to identify patterns in spatial arrangement of the diatom community, and to see if there were any detectable colonization patterns for dominant diatom taxa. 3. Is there a difference in the number of benthic individuals associated with vegetated vs. non-vegetated areas? To address this question, the effects of planting elevation, plant species, and time on the following endpoints were tested: • total abundance of benthic organisms in samples containing vegetation vs. control samples • abundance of benthic organisms in the aboveground vs. belowground components of the vegetation and control samples • taxonomic richness of vegetation and control samples The NMS ordination technique was used to look for potential patterns in spatial arrangement of the benthic communities associated with the vegetation and control samples. 4. Is there a difference in the benthic communities that colonize the three plant species over time? To address this question, the effects of planting elevation, plant species, and time on benthic abundance per gram of plant material were tested. 3 NMS was used to see if there were any detectable colonization patterns for dominant benthic taxa associated with the vegetation and control samples. 5. Is there a link between the periphyton and benthic communities associated with the three plant species? To address this question, the effects of planting elevation, plant species, and time on the ratio of diatom biovolume to benthos abundance were tested. 1.1.3 Significance of this research Low biological productivity is common in reservoirs found in temperate climates around the world. Management options used in British Columbia to increase biological productivity, with varying degrees of success, have included whole reservoir fertilization (Ashley and Slaney 1997; Ashley era/. 1997; Wilson era/., 2000; Perrin, 1997; Stockner et al. 2005), introduction of exotic organisms as food sources for key species (e.g. introduction of Mysis relicta in a number of British Columbia lakes and reservoirs to act as a food source for kokanee, as described in Ashley et al. 1997), changes in reservoir operating regimes (BC Hydro 1999), and shoreline planting programs (Fisheries and Oceans Canada (DFO) 1999). It is hoped that the present research will expand the scientific knowledge base regarding the ecological links between littoral zone vegetation and reservoir food webs; and therefore, encourage reservoir managers to consider the role of littoral zone vegetation in overall reservoir biological productivity when making future decisions regarding reservoir operations. 4 1.2 Literature Review Macrophytes are of significant importance to aquatic food web interactions and quality of aquatic environments, even with relatively low plant coverage. Plants not only act as a source of nutrients into the aquatic environment, they also play a role in predator/prey relationships between fish and invertebrates, and between predacious, planktivorous and benthivorous fish. Changes in these relationships may have implications for the entire food web in both the pelagic and littoral zone (Hunt and Jones 1972; Jeppesen er al. 1997). Aquatic macrophytes play an important role for macroinvertebrates both directly, by providing structure to sediment and aboveground habitat (Palmer er al. 2000) and resources for invertebrates (i.e. additional substrate area provided to periphytic algae, which may in turn be a food source) (Northcote and Atagi 1997), and indirectly by effecting light penetration, oxygen concentration and temperature. Fresh macrophyte tissue is used directly in the diets of some herbivores (Lodge 1991), and after senescence and death, plant material becomes available to shredders and deposit feeders (Suren and Lake 1989; Kornijow et al. 1995; Diehl and Kornijow 1997). The abundance and species of aquatic macrophytes influences fish-macroinvertebrate interactions by increasing the diversity of habitats and resources for macroinvertebrates; reducing susceptibility of macroinvertebrates to fish predators; and reducing the vulnerability of prey fish to piscivores (Dibble et al. 1996; Jeppesen et al. 1997; Northcote and Atagi 1997). Plants also provide a resting area for migrating birds, and breeding site for birds, fish and invertebrates (Crow and MacDonald 1978). This literature review summarizes the current understanding of ecological links between submerged macrophytes and aquatic food webs. The role of benthic invertebrates in providing those links as a bridge in the flow of energy and carbon between the macrophytes with an associated biofilm and fish is discussed. The review focuses mainly on north temperate ecosystems, and wherever possible, reservoirs. Since there is a paucity of information available on shoreline vegetation in reservoirs, much of the literature in this review relates to wetlands which are similar to reservoir littoral zones in the sense that the water table is generally at or near the surface, or the land is covered by shallow water at some time during the growing season of each year. This results in the substrate becoming "saturated for a long 5 enough period to promote wetland or aquatic processes as indicated by hydric soils, hydrophytic vegetation, and various kinds of biological activity which are adapted to the wet environment" (Tarnocai 1979). Similarly, the information presented on invertebrates has been summarized from lake and reservoir studies whenever possible; however, stream studies have also been used when appropriate. 1.2.1 Storage Reservoirs vs. Lakes and Rivers Storage reservoirs are regulated bodies of water that may be managed for a variety of activities including provision of drinking water, flood control, recreation, wetland enhancement projects, navigation, and the production of hydroelectric power. Water release is controlled by the reservoir manager, not nature, usually resulting in a release pattern quite different than the natural hydrologic cycle. The resulting waterbody is essentially a river-lake hybrid (Baxter 1977; Thornton 1990; Soballe et al. 1992.) that displays characteristics of both types of waterbody, but is neither. For example, reservoirs have vertical gradients, similar to lakes, related to parameters such as light penetration, plankton growth, nutrient concentrations, and temperature stratification (Thornton 1990). However, they also exhibit horizontal gradients normally associated with rivers, such as current velocity, turbidity, transparency, channel depth and width, sediment load and bottom type (Thornton 1990; Soballe et al. 1992). In addition, reservoirs typically have a longitudinal gradient, with the waterbody becoming less river-like and more lake-like as the channel broadens and deepens moving from the upstream end of the reservoir downstream toward the dam (Soballe et al. 1992). Lake-like characteristics also increase with water mean residence time (Thornton 1990). An important distinguishing feature between natural lakes and storage reservoirs is the depth of discharge. In natural lakes, the discharge is always over the top of the lowest edge; when the water drops below this edge, discharge stops. In summer months, this epilimnetic water is generally warm, oxygenated, and nutrient depleted. In contrast, almost all storage reservoirs discharge from subsurface outlets in order to maximize the potential for power generation. This relatively cold, anoxic, and moderately nutrient rich hypolimnetic water can continue discharging from the reservoir system even at lower water levels, resulting in greater water level fluctuations, or drawdowns, in comparison to natural lakes (Furey era/. 2002). 6 It has been well documented that drawdowns result in a loss of biological productivity in reservoirs (Fraley et al. 1989; Thornton 1990; Marotz et al. 1994; Northcote and Atagi 1997; Chambers et al. 1999; Kaster and Jacobi 1978). The irregular and often extreme water level fluctuations associated with reservoir storage and release patterns inhibit aquatic plant growth during the drawdown phase, and terrestrial plant growth during the periods of flooding. Many plant species are able to tolerate either the drought or flooding, but few are able to tolerate both. This generally limits the number of plant species that can become established within the littoral zone, and often prevents development of a stable vegetated littoral zone (Kaster and Jacobi 1978; Wetzel 1990). In addition, increased erosion resulting from the repeated inundation and desiccation of the littoral zone can cause increased turbidity in the water, thereby reducing light transmittance to this area (Peltier and Welsh 1970; Thornton 1990). Other effects on biological productivity resulting from reservoir drawdown include reduced water volume, which decreases the overall area in which all organisms can live, as well as a decrease in the surface area available for phytoplankton growth (Marotz et al. 1994). Hunt and Jones (1972) report that a decrease in reservoir volume results in a decrease in biological productivity in lower trophic levels, resulting in reduced food availability for fish. Additional effects of reservoir drawdown include altered thermal stratification; changes in water chemistry; erosion due to wave action; loss of zooplankton and phytoplankton through the dam; and reduced fish growth in the late summer and fall (Kaster and Jacobi 1978; Carpenter and Lodge 1986; Fraley et al. 1989; Northcote and Atagi 1997; Thornton 1990; Furley er al. 2002). One of the primary causes of reduced biological productivity in reservoirs in comparison to natural systems as a result of water level fluctuations, however, is the exposure of emergent macrophytes and the associated periphytic and benthic communities to solar radiation, wind and other elements which results in desiccation of the communities (Thornton 1990; Marotz et al, 1994; Paterson and Fernando 1969; Hunt and Jones 1972). In addition, the photic zone shifts as water levels change, forcing communities at the lower elevations to adapt to higher water temperatures and light levels than they would normally be subjected to. The result being that the normal seasonal growth cycles of the communities are frequently interrupted and they never reach a climax state of succession (i.e. stable state or equilibrium condition). Kaster and Jacobi (1978) indicated that recolonization of benthic communities within the drawdown zone of a Wisconsin reservoir required 3 months, whereas Marotz et al. (1994) reported that benthic communities in Hungry Horse and Libby reservoirs required at least 2 years to recover from large water level drawdowns. 7 Plant species that are capable of adapting to this extreme environment are usually quite successful. They form a vegetation community that at certain times of the year functions like a submergent macrophyte bed, providing refuge and food sources to fish which are able to freely move among the macrophytes, and at other times, like a terrestrial flood plain or wetland (Wetzel 1990a). Adaptations in emergent plants in many cases can result in growth rates which exceed those of most terrestrial or submergent vegetation (Wetzel 1983). Coastal reservoirs in British Columbia are operated such that there are two annual periods of drawdown. The first drawdown provides room to store spring run-off from the melting snowpack, and the second is to make room for heavy fall rains. In Stave Reservoir, where the field research for this thesis was conducted, seasonal water levels may fluctuate by up to nine meters (Hirst 1991; BC Hydro 2000). 7.2.2 Vegetation: Physiological reaction to flooding The wetland environment exposes plants and animals to stresses that many organisms are unable to contend with. These may include long periods of exposure for aquatic organisms, and conversely, long periods of inundation for terrestrial organisms. Once the photosynthetic parts of a plant are flooded and excluded from the ambient air, a reduction of photosynthesis occurs and the energy budget of the plant is severely affected (Ernst 1990). This results in a reduction in biological productivity, with the primary suppressing mechanisms being exponential light attenuation with depth, and the reduction of gas and nutrient exchange (Raven 1984). Flooding also results in a reduction of the oxygen concentration of the soil, leading to a rapid decrease in oxygen supply to the roots of the plant. This causes aerobic metabolism in the roots to shut down in flood sensitive plants. The energy status of the cells becomes impaired and almost all metabolically driven processes such as cell division, cell extension, and nutrient absorption decrease. Within 24 hours of anoxia, mitochondria and organelles of flood sensitive plants are generally destroyed. Flood tolerant plants, however, use various morphological and physiological growth and reproductive adaptations, including development of aerenchyma and specialized tissues or roots, to survive the stress of flooding and associated anoxic soil conditions (Ernst 1990; Raven 1984; Mitsch and Gosselink 2000; Robe and Griffiths 1998; Steed er al. 2002; Visser et al. 2000) 8 Aerenchyma and adventitious roots To avoid root anoxia, flood tolerant plants may develop air spaces called aerenchyma in their roots and stems (McCarron et al. 1998; Steed er al. 2002; Visser et al. 2000; Muthukumar 2004). The spaces allow diffusion of oxygen from the aerial portions of the plant into the roots rather than from the soil, which is the usual mode of oxygen uptake in terrestrial plants. Mitsch and Gosselink (2000) report that pore space for normal terrestrial mesophytes (land plants that need moderate amounts of moisture for growth) is 2-7% of their volume, whereas up to 60% of the volume of wetland species may consist of pore space. The formation of aerenchyma is thought to occur either by cell separation during maturation of the root cortex or by cell breakdown, possibly due in part to production of ethylene. Ethylene, which is formed in response to a hormonal change in the plant's hypoxic tissues stimulated by flood conditions (Jackson 1985), concentrates rapidly in flooded tissue because its diffusion rate in water is approximately 10,000 times slower than in air. In a number of plants, ethylene stimulates cellulase activity in the cortical cells, which results in the collapse of cell walls (Kawase 1981). Wetland trees and plants may also respond to flooding by developing specialized tissue or organs, such as adventitious roots, lenticels, and pneumatophores (air roots); or by growth adaptations such as stem elongation. Adventitious roots, or roots that arise from a stem rather than from the primary root, are characteristic of Carex and Scirpus species (Visser et al. 2000), as well as other flood tolerant herbaceous plants and trees. Adventitious roots may also develop in flood intolerant plants just above the anaerobic zone, where the roots function normally in the aerobic environment (Jackson 1985). Root aeration has a number of secondary effects, including physiological changes in the plant to limit water uptake, oxygenation of the surrounding rhizosphere, and changes in nutrient absorption (Mitsch and Gosselink 2000). Water Uptake: Flood intolerant plants show a decreased ability to uptake water upon flooding, likely as a response to overall reduction of root metabolism. The plant responds to flooding by closing its stomata, which reduces its uptake of C 0 2 , decreases transpiration, and results in wilting. It is 9 likely that this adaptation is to minimize water loss and accompanying damage to the plant's cytoplasm (McKee et al. 1989). An additional, undesirable, result is the depression of photosynthetic activity, possibly due to the slow rate in C 0 2 diffusion (Robe and Griffiths 1998). Similar symptoms are noted in plants under drought conditions (Mitsch and Gosselink 2000). Rhizosphere oxygenation: The availability of many nutrients in the soil is modified by anoxic conditions, especially for flood intolerant species (Robe and Griffiths 1998; Mitsch and Gosselink 2000). As the period of flooding progresses, more nitrogen, manganese, iron, sulphur and carbon in the soil changes from the oxidized to reduced state, resulting in potentially toxic levels in the plant tissue (Ernst 1990). In mildly anoxic substrates, diffusion of oxygen through the stems to the roots is often enough to supply the roots with oxygen, and to allow for some oxygen to diffuse out of the roots into the surrounding anoxic soil (Izumi et al. 1980, Sand-Jensen et al. 1982; Ottosen et al. 1999; Palmer et al. 2000). It is thought that this oxygenated rhizosphere may help to moderate the toxic effects of the soluble reduced ions by reoxidizing the ions and precipitating them in the rhizosphere, making them unavailable for plant uptake at dangerous levels (Mitsch and Gosselink 2000; Ernst 1990). It is possible that the root systems of the flood tolerant plants may help to oxidize the soil so much that nearby flood intolerant plants may also be able to survive (Mitsch and Gosselink 2000). Water conservation mechanisms in wetland plants may also act to reduce the rate at which soil toxins are drawn toward the root, which increases the probability of oxidizing and precipitating the toxins, making.them less available for uptake as they move through the oxygenated rhizosphere (Mitsch and Gosselink 2000). Uptake of specific nutrients Algae and submergent macrophytes obtain many of their required nutrients from the water column by foliar absorption. Emergent plants, such as sedge and woolgrass, on the other hand, obtain many of their nutrient requirements from the sediment (Mitsch and Gosselink 2000). In anoxic soil, nitrates are generally replaced by the reduced form, ammonium. Most plants preferentially absorb the oxidized form nitrate; however, wetland plants seem to be able to maintain normal rates of nitrogen uptake upon flooding. This could be a result of ammonium 10 possibly being oxidized to nitrate in the oxygenated rhizosphere as described above, or it may be that some species are able to absorb ammonium directly (Ottosen er al. 1999), or by direct uptake of organic N as low-molecular-mass amino acids (Muthukumar et al. 2004). The availability of phosphorus generally increases in saturated soils; however phosphorus is precipitated by iron/which also increases in availability in saturated soils. Studies have shown a general trend of decreased uptake of phosphorus by flood intolerant species and either enhanced or no effect with flood-tolerant species (Mitsch and Gosselink 2000). Iron and manganese both become reduced in wet soils, and are therefore easier for the plant to uptake. Plants, however, require extremely small amounts of iron and manganese, and due to their high concentrations, they may become concentrated in plant tissues and can quickly reach toxic levels (Ernst 1990). In addition to the moderating effects of the oxygenated rhizosphere, wetland plants are able to tolerate these high concentrations through several adaptations: • Much of the minerals taken up can be sequestered in cell vacuoles, in shoot vascular tissue, or in senescing tissue where they do not influence the metabolism of healthy cytoplasm. • Many wetland plants may have a higher metabolic tolerance of these elements (Ernst 1990). Sulfur is reduced to sulfide in anaerobic soils; a form that is toxic to plants. The uptake of sulfate is metabolically controlled, however uptake of sulfide occurs without control in wetland plants and elevated concentrations are often found in the tissues of flood-tolerant species under reduced conditions. Ernst (1990) describes a variety of detoxification mechanisms utilized by wetland plants to survive high sulfide tissue concentrations, including oxidation of sulfide to sulfate in the oxygenated rhizosphere; accumulation of sulfate in the vacuoles, and the conversion to gaseous hydrogen sulfide, carbon disulfide and dimethylsulfide, and then the subsequent loss to diffusion; and a metabolic tolerance to elevated sulfide concentrations. Avoidance Strategies Many plants have also evolved escape strategies by life-history adaptations, such as timing of seed production to occur during the non-flooded season by either delayed or accelerated 11 flowering (Robe and Griffiths 1998; Blom and Voesenek 1996); production of buoyant seeds that lodge on high, unflooded ground; production of a large persistent seed bank; and production of tubers, roots, and seeds that can survive during long periods of submergence (Mitsch and Gosselink 2000). 1.2.3 Nutrient translocation and cycling In general, wetland plants survive periods of cyclical inundation by undergoing active growth in spring and summer, followed by translocation of nutrients from senescing aboveground biomass to belowground storage organs in the fall. The majority of the aboveground biomass decomposes over the fall and winter, and in the spring, nutrients are moved from the roots to shoots to initiate growth. Plants that are adapted to survive extended periods of inundation and exposure in the reservoir environment survive by developing shoots in the late summer which overwinter as small shoots. These shoots experience rapid growth in the spring, partially due to translocation of stored nutrients from the roots (Bernard and Hankinson 1979; Robe and Griffith 1998; AIM Ecological Consultants Ltd. et al. 2000; Roseff and Bernard 1979; Kistritz et al. 1983). For example, Roseff and Bernard (1979) estimated that 15-25% of spring growth was related to translocation of nutrient reserves. Kistritz er al. (1983) found that aboveground growth of Carex lyngbyei appeared to rely almost entirely on reallocated nutrients. Their study showed that the most active period of translocation of N and P was May to August in a C. lyngbyei marsh, with belowground nutrient levels diminishing to minimum levels in May and June, while aboveground levels increased to maximum levels during the same time frame. Maximum aboveground standing crop was measured approximately one month later. Belowground nutrient losses were in excess of aboveground gains, suggesting that the balance of nutrients were lost to the water column via aboveground leaching. From late July to the end of August, Kistritz et al. (1983) calculated a downward nutrient translocation rate of 44.8 mg N per m 2 per day, and 12.2 mg P per m 2 per day in C. lyngbyei. Belowground nutrient reserve gains in excess of aboveground losses likely reflected uptake of nutrients by the roots from interstitial waters of the hydrosoils, and from microbial degradation of residual plant tissue and epiphytic microorganisms (Kistritz et al. 1983; Wetzel 1996). In a submergence study completed in Arrow Reservoir, AIM Ecological Consultants and Carr Environmental Consultants (2000) noted that sedges began to grow replacement shoots for dying leaves while plants were still underwater. Similarly, they found that reed canarygrass, another species 12 adapted to surviving inundation, began to grow new above and belowground biomass while still submerged. Upon inundation, the plant material begins to senesce and is subject to biological and mechanical fragmentation (Puriveth 1980, Brinson et al. 1981). Some of the nutrients from the plant material are leached into the water column (Boyd 1970, Gosselink and Kirby 1974); some are translocated to roots and rhizomes as described above (Kistritz er al. 1983); while others are eventually exported through decomposition to the wetland surface (Goldman and Home 1983; Mitsch and Gosselink 2000) or through consumption by herbivores. Wetland plants generally have a large foliar surface area to volume ratio. This can greatly enhance gas exchange between the plant and surrounding water (Robe and Griffiths 1998) and the interception of light (Sculthorpe 1967). This also increases the amount of substrata available for colonization by periphyton (i.e. the assemblage of epiphytic algae, cyanobacteria, and other microbes that colonizes macrophytes). Upon inundation, the plant material begins to leach nutrients and dissolved organic carbon which supports the growth of a bacteria and cyanobacteria assemblage. The bacteria and cyanobacteria secrete a polysaccharide matrix which forms a biofilm on the surface of the macrophyte. The biofilm provides a rich food source for epiphytic algae, including flagellates, ciliates and diatoms (Stockner and Antia 1986). Initially, the algal assemblage is dominated by cells that tend to be low posture, prostrate or apically attached. Over time, the community becomes more three-dimensional as protrusive, stalked, and or filamentous forms grow (Jones er al. 1997). The resulting periphyton communities are often multi-layered with an adherent understory of basal cells, encrusting green algae, and tightly attached diatoms. Above that are protrusive long diatoms, stalked diatoms and filaments (Allan 1995). The large surface area provided by the submerged macrophyte allows the attached microbial community to extend higher into the water column where it is exposed to more light and dissolved gases from photosynthesis and decomposition (Wetzel 1996; Wetzel and Sondergaard 1997). Chlorophyll a associated with the attached microbial community on submerged plant surfaces can be up to several hundreds of micrograms per square meter in density (Jones er al. 1997). This epiphytic microbial mass may actually compete with the plant for light, inorganic carbon and nutrients, and in many cases, the biological productivity of the attached microbial community on the macrophyte and associated detritus exceeds that of the macrophyte itself 13 (Kistritz er al. 1983; Wetzel 1996; Stockner and Antia 1986; Porter et al. 1988; Wetzel and Sondergaard 1997). As such, the attached microbial communities are often major regulators of nutrient dynamics in many freshwater environments (Wetzel and Manny 1972; Wetzel 1983, 1990b; Stockner and Antia 1986; Stockner and Porter 1988). Direct Energy Transfer to Higher Trophic Levels Direct energy transfer from the primary producers to higher trophic levels occurs via the grazing and/or detritus food chains (Jones er al. 1997). Within the grazing food chain, herbivores consume macrophytes directly or graze on the attached microbial biofilm. Within the detritus food chain, detrivores feed on highly nutritious detritus and the associated biofilm (Jones et al. 1997; Wetzel 1996; Soballe et al. 1992). The following section explores the role of the microbial biofilm in nutrient cycling and transfer of nutrients and carbon to higher trophic levels. 1.2.4 Microbial Loop It was originally thought that microorganisms in aquatic environments were strictly decomposers, with the ability to reduce organic detritus to dissolved and particulate organic matter (DOM and POM), soluble nutrients nitrogen (N) and phosphorus (P), and gases (C0 2 , C H 3 , and H 2S). Researchers now believe that bacteria and small cyanobacteria are, in fact, primary carbon producers in pelagic microbial food webs, and that the remineralization of N, P and dissolved organic matter is largely mediated by excretion from protistan (e.g. flagellates and ciliates) and micro-metazoan (e.g. rotifers and nauplii) grazing activity on picoplanktors (0.2 - 2.0 urn diameter) (Pomeroy 1974; Stockner and Antia 1986; Stockner 1991, 2001). The total activities of all microorganisms, including bacteria, cyanobacteria, flagellates, ciliates and other microzooplankton, within the aquatic food web, form the microbial loop - a complex microbial food web that facilitates flow of carbon and nutrients. Carbon and nutrient flows within the microbial loop are tightly coupled. The dynamic nature of the loop is facilitated by three ecological processes: commensalism (production of DOM by phytoplankton and subsequent utilization by bacteria); competition for nutrients between bacteria and phytoplankton (both bacteria and cyanobacteria are effective competitors of larger phytoplankton and attached algae for bioavailable N and P); and predation/grazing, which 14 provides nutrients and DOM due to cell lysis during "sloppy eating" and excretion (Stockner and Antia 1986; Stockner and Porter 1988). Grazing within the microbial loop Grazing activities are very important for the maintenance of efficient flows of energy and carbon in aquatic food webs. In the absence of grazing, biomass accumulates, and the overall biological productivity of the ecosystem declines (Robinson et al. 1997). It is currently believed that grazing microflagellates are the most important 'link' organisms in the transfer of carbon from picoplankton to higher trophic levels in aquatic food webs (Stockner and Antia 1986). Grazers such as mayflies and stoneflies in streams, and chironomids in littoral zones, are predators on the microflagellates and attached bacteria in the periphyton community (Allan 1995), allowing for carbon transfer to higher trophic levels (Stockner 2001). Nutrient Recycling The relative significance of the microbial food web to total nutrient recycling and carbon production is highest in oligotrophic systems. Under these conditions, bacteria are able to out-compete phytoplankton for the available nutrients, and the low food supplies often preclude the populations of larger keystone grazers (e.g. Daphnia) from becoming established. The bacteria sequester dissolved N and P, and due to their small size, are able to stay in the epilimnion rather than settling out in the hypolimnion or sediments (Stockner and Porter 1988). As a result, bacteria become a carbon 'link' to higher trophic levels rather than a carbon 'sink'; and the microbial food web becomes the primary source of nutrient regeneration and material flux (Stockner and Antia 1986; Stockner and Porter 1988; Stockner and Maclsaac 1996). Similarly in hyper-eutrophic systems, where excessive nutrients, anoxic conditions, or toxic algal blooms may result in unfavorable conditions for keystone grazers, microbial food webs (via grazing) may also play the key role in transferring carbon and nutrients to higher trophic levels (Stockner and Antia 1986). It should be noted, however, that each transfer of carbon and nutrients has a high respiration cost resulting in a loss of carbon in the form of C02 . Therefore, the longer and more complex a food chain is, the less efficient it is with respect to carbon transfer (Stockner 1987; Stockner and Shortreed 1989). Despite these high respiratory costs, it has been demonstrated that mild fertilization of oligotrophic systems can stimulate microbial 15 food webs, resulting in at least a doubling of carbon at all trophic levels (Stockner and Maclsaac 1996). The microbial food web generally plays a less important role in mesotrophic systems, where Daphnia tend to be highly abundant. Daphnia, as keystone grazers, are able to graze on all sizes of organisms contributing to the microbial loop, effectively shutting the process down. When this occurs, the predominant pathway of carbon flow reverts to the larger organisms within the more efficient mesotrophic food chain, and most nutrient and DOM recycling results from damaged cells and lysis due to sloppy grazing and excretion (Pace et al. 1990). At times of the year when keystone grazers are absent or in low abundance the microbial loop again plays the key role in nutrient cycling (Stockner 2001). The microbial loop in detritus-based ecosystems In the past, most research on the importance of microbial communities has focused on the pelagic environment, and little was known of their role in detritus based ecosystems (i.e. littoral zones of lakes, shallow lakes with emergent macrophytes, wetlands, swamps, etc.) (Porter er al. 1988). More recently, the processes and principles apparent in the pelagic environment have been studied in other aquatic habitats, such as streams and littoral zones of lakes, with similar findings (Stockner 2001). In many of these systems, the exceptionally high rates of primary productivity of the attached communities are only possible due to the intensive internal recycling of nutrients, including carbon and gasses, within the attached microbial communities (Jones er al. 1997; Wetzel 1993). Figure 1 shows the microbial loop related to the periphyton and benthic assemblage associated with macrophytes in an oligotrophic system, such as Stave Reservoir. Figure 2 shows the linkages from the periphyton assemblage to the pelagic food web in a mesotrophic system. Indirect Energy Transfer to Higher Trophic Levels Indirect transfer of energy from the primary producers to higher trophic levels occurs via a number of pathways, including leaching from aquatic macrophyte shoots (Kistritz er al. 1983; Wetzel and Manny 1972; McRoy er al. 1972); excretion from animals; and export of invertebrate larvae (Jones et al. 1997; Crow and MacDonald 1978; Wetzel 1996; Wetzel and Sondergaard 1997). 16 Fish ^ K 1 \ V Chironomids (Keystone Order) Oligochaetes D O M ^ A l w A A J Nematodes (detrivores) Microbenthos (20 - 200 nm) tfl^ J)S Slopp) ,} A graze J whicl — r » bacter l y eating and excretion by r rs releases DOM, N & P, which are cycled back to the ia and picocyanobacteria Nanobenthos (2 - 20 (xm) (flagellates, ciliates, diatoms) Bacteria and (0.2 - 2 (im) picocyanobacteria DOM, N &.;P Decay of plant material provides DOC and nutrients to support v ^ D O C , hi & P development of the bacteria and picocyanobacteria Periphyton assemblage Figure 1. Schematic of a food web for the littoral zone of Stave Reservoir. Under oligotrophic conditions, the benthic microbial loop plays a key role in the transfer of nutrients to higher trophic levels. The size of each arrow is indicative of the relative importance of the flow of carbon. 17 Dissolved and Parti GUI ate Organic Material : (DOM; & POM) Fish (carnivore) Macrozooplankton (i.e. Daphnia) (secondary producers) Phytoplankton (i.e. „ . diatoms, dinoflagellates) (primary producers) N & P J Bacteria/Fungi •••• N & P Large amounts ofN&P are introduced to the system from external, usually human, sources (i.e. sewage, agricultural drainage, etc...) "pelagic" Chironomids Oligochaetes Nematodes (detrivores) DOM, N & P Microbenthos (20 - 200 |im) 1 I / Nanobenthos (2 - 20 nm) (flaaellates, ciliates. diatoms) I Bacteria and (0.2 - 2 jim) picocyanobacteria Sloppy eating and excretion by grazers releases DOM, N&P which are] cycled back to octbie bacferia and pico-N&P cyanobacteria eriphyton assemblage DOC and nutrients released from decaying plant material support bacteria and pico-cyanobacteria, which in turn, form a biofilm to support the periphyton assemblage "littoral" Figure 2. Schematic showing the linkage between the littoral and pelagic zones in a mesotrophic lacustrine system. The size of each arrow is indicative of the relative importance of the flow of carbon. 1.2.5 Aquatic Invertebrates Invertebrates perform important ecosystem and community functions in lakes and wetlands. Invertebrate communities are intermediates in the transfer of energy to higher trophic levels, and many invertebrate species are a major food source for fish, birds, amphibians and mammals at some point in their lifecycle (Persson 1988; Osenberg et al. 1992; Olsen et al. 1995; Weller 1994). Invertebrates commonly found in lentic (lake-like) waterbodies, include oligochaetes, nematodes, microcrustacea (Cladocera, Copepoda, Ostracoda), amphipods, isopods, mysids, chironomid larvae, caddisfly larvae, and mayfly larvae (Jones er al. 1997). Generally, anoxia tolerant species such as chironomids and oligochaetes tend to dominate the reservoir benthos community (Kaster and Jacobi 1978). Habitats provided by both flooded emergent and submerged vegetation are important to the development of the invertebrate community (Voigts 1976). Generally, there is a strong correlation between the densities of submerged macrophytes and epiphytic macroinvertebrates (Kornijow and Kairesalo 1994; Rasmussen 1988), with a higher abundance of macroinvertebrates found in vegetated areas versus non-vegetated areas (Cowell and Hudson 1968; Diehl and Kornijow 1997; Northcote and Atagi 1997; Levings 1997; Cyr and Downing 1998; Gregg and Rose 1985). This is likely due to a number of factors including an increase in substrate area offered by the plants (Krull 1970), the food source offered by periphytic alga and detritus (Cattaneo and Kalff 1980), and lower risk of predation in the more complex habitat (Hargeby 1990). The standing crop of macroinvertebrates varies considerably among different species of aquatic macrophytes, and the diversity and density of the epiphytic macroinvertebrate communities tends to be related to the growth forms of the macrophytes they are associated with (Jones et al. 1997). Lamberti and Moore (1984) describe four general categories, or functional feeding groups, of macroinvertebrates based on common feeding characteristics: 1. Filter-feeders construct tubes on plant stems or in the bottom sediments where they construct coarse, irregular nets on or below the substrate surface. The main food source appears to be planktonic alga or detritus. If sufficient plankton is unavailable, some species will revert to other feeding methods such as deposit-collecting or shredding (Walshe 1951). 19 Midge larvae in subfamily Chironominae are the most common type of filtering insects in lentic environments. 2. Deposit-collectors commonly live on the sediments of lakes, ponds, and slow moving streams where currents do not interfere with their movements and food sources. These invertebrates feed primarily on fine particulate organic matter that is often conditioned or processed by microbiota. They generally have reduced mandibles and maxillae (Hynes, 1970), as they do not scrape attached material. Their mouthparts may be covered with dense brushes of hairs or setae, generally associated with the maxillae and labium, which allow them to browse on periphyton. Some species are able to switch between deposit-collecting and other functional groups. 3. Scrapers are most abundant in flowing waters and possess blade-like mandibles and other mouthparts, with limited setation that allow them to remove alga and other attached organisms from the surface of rocks or other solid objects. There is great variation in the morphological adaptations that allow these groups to be effective scrapers. 4. Shredders generally have cutting mandibles. Some consume large amounts of living higher plants; however, many species bore directly into higher plants and construct a net at the front of their tube to filter plankton. Some shredders have cutting mandibles that are used to bore into wood or to consume large particles of fallen detritus. Merritt and Cummins (1996) also identify predators as a functional feeding group. Predators feed upon other animals, either by ingesting the animal, or by sucking its bodily fluids. Macroinvertebrates found in the littoral zone can be split into two further categories: Benthic (i.e. dwelling in bottom sediments), or epiphytic (i.e. associated with macrophytes). Generally, the same few classes, orders and families can be found in both areas; however, different genera are usually associated with plants in comparison to genera associated with the benthos (Kornijow er al. 1990, Kornijow and Kairesalo 1994). Benthic and epiphytic invertebrates differ with respect to their seasonal dynamics, food sources, and predators (Diehl and Kornijow 1997). 20 The ratio of epiphytic macroinvertebrates to the total number of macroinvertebrates tends to increase with increasing vegetation density. In very dense areas, invertebrates associated with aboveground biomass tend to outnumber invertebrates associated with belowground biomass (Kornijow and Kairesalo 1994). Epiphytic invertebrates tend to be smaller than benthic invertebrates, therefore relative biomass effects may be less noticeable (Jones et al. 1997). Role of Aquatic insects in Translocation of Nutrients Aquatic insects facilitate nutrient turnover in aquatic environments directly through nutrient uptake and excretion associated with feeding actions such as grazing or deposit-collector activity (Graneli 1979; Gallepp era/. 1978; Gallepp 1979; Gardner era/. 1981). Aquatic invertebrates may also facilitate the translocation of nutrients indirectly through enhancement of microbial activities associated with particulate organic matter; translocation of nutrients at sediment-water interfaces through physical activities such as bioturbation and mineralization (Merritt et al. 1984; Soballe et al. 1992; Mitsch and Gosselink 2000); or through emergence and mobility (Cowell and Hudson 1968; Likens and Loucks 1978). Direct pathways for nutrient translocation due to aquatic invertebrates Grazing is a very important mechanism for translocation of nutrients in aquatic environments (Cattaneo 1983). Effects of invertebrate grazing on periphyton may include direct consumption of the periphyton resulting in an alteration of the abundance and composition of the periphyton community, nutrient mobilization, and physical disruption or dislodging of the organisms. In turn, the dislodged material may be an important food source for deposit-collecting detrivores (Jones er al. 1997). Supply of nutrients to the grazer-resistant algal understory is enhanced by invertebrate grazing activity either directly by excretion or indirectly by removing periphyton overstory and facilitating diffusion from the water column to the understory (Jones er al. 1997). Most grazers are mobile and focus on distinct patches of periphyton; however, there are species, such as some caddisfly, chironomid, and lepidopteran species, which live in constructed structures and feed in a localized area (Lamberti and Moore 1984; Bergey 1995). The diet of grazers depends not only on the periphyton community present, but also on attributes of the grazer itself, including size, motility, morphological specialization for feeding, 21 and digestive capabilities. Most grazers are scrapers (e.g. chironomids, caddisflies, and mayflies); however, a number of collector-gatherers, such as some mayfly larvae may also have mouthparts which allow them to browse on periphyton (i.e. dense brushes of hairs or setae). Other taxa use additional body parts to remove periphyton from the macrophyte before eating it. Differences in morphology relate to different effects on periphyton (Jones er al. 1997). In addition to attributes of the grazer, quantity, quality, and composition of the periphyton community are also important factors in determining the abundance, composition, and growth rate of the invertebrate grazer community, and in turn, the biological productivity of that community (Lamberti et al. 1989; Vaughn 1986). Important periphyton characteristics with respect to grazing include species, size, growth form, density, ease of harvest, as well as reproductive traits and chemical defenses (Gregory 1983). Physical features, such as variation in protein and lipid content and cell wall thickness may affect nutritional value and palatability of periphyton, and in turn, the rate at which it is consumed (Gregory 1983). For example, Jones er al. (1997) noted that although cyanobacteria are high in protein content, they are generally a poor food source for invertebrates as the mucosaccharide sheath is indigestible and toxins may be produced when eaten. There are conflicting views in the literature as to whether quantity or quality of the periphyton food source is the most important feature in development of the grazer community. Vaughn (1986) noted that selection for algal abundance was of overriding importance in streams, where it was observed that even tough algal types had a marked effect on the development of the grazer community. Gressens and Lowe (1994), however, concluded periphyton quality was more important than quantity. They studied the dispersal of chironomid larvae among algal patches differing in abundance of several diatom species and a green algae Stigeoclonium. Patch preference of the larvae was negatively correlated with abundance of green alga, concentration of chlorophyll a, and algal biovolume, and positively correlated to algal diversity. Generally, higher numbers of grazers are associated with a higher abundance of periphyton, as invertebrate life cycle processes (hatching, migration from mud to plants and back, pupation, cocoon formation and emergence) are often timed to coincide with food availability (Cowell and Hudson 1968; Feminella and Hawkins 1995; Fairchild 1981; Fairchild and Lowe 1984; Jones et al. 1997). For example, Mason and Bryant (1975) noted that some species of chironomid larvae moved from mud in late spring onto plants and grazed periphyton, and then moved back to the 22 sediment in the fall jn response to falling temperatures. As a result, there is often a spring peak in invertebrate and periphyton abundance (Tokeshi 1986). Insect grazing influences community structure and turnover rates of the algal and microbial populations that comprise their food. Effects on a community level depend on the composition of the original periphyton community (Hart 1981; Kohler 1984), nature and size of grazers (i.e. scrapers tend to be more devastating than browsers) (Mason and Bryant 1975; Bowker er al. 1983; Karouna and Fuller 1992), their abundance (Dodds 1991); and the localized environmental conditions (Dodds 1991; Poff and Ward 1992). In streams, grazing activity by scrapers may shift the structure of the affected periphyton communities from large, slow-turnover species to small, rapid-turnover species that are less palatable or easy to harvest, such as coarse filaments or cyanobacteria (Merritt et al. 1984). Lamberti et al. (1989) observed initial food assimilation rates for a snail of 70 to 80% in a stream environment. These levels declined to 40% over time, coinciding with a shift in composition from diatoms and unicellular greens to filamentous green and blue-green algas. A similar pattern may occur due to activity of shredders and collectors on fungal and bacterial communities (Barlocher 1980). Eichenberger and Schlatter (1978) reported that grazing may result in increases in species richness in the stream environment. Studies on the effects of invertebrate grazing on primary productivity have shown a range of results. Some studies in streams have shown grazing to reduce periphyton productivity (e.g. Rosemond et al. 1993). Other results have shown an increase in primary productivity on a biomass specific basis (Lamberti and Resh 1983; Gelwick and Matthews 1992), or an increase of areal primary productivity (Lamberti et al. 1989; Power, 1990). Enhancement of production has been attributed to a variety of factors including nutrient renewal by algal cell disruption and excretion by grazers (Underwood 1991), reduced competition for nutrients (Lamberti and Resh 1983), increased light levels to the grazer resistant algal understory, and removal of senescent cells by consumption or dislodgment (McCormick and Stevenson 1991; McCormick 1994; Lamberti era/. 1987). Competition between grazers is an important factor in determining the relative size of the grazer and periphyton populations. Competition between grazers may interfere with feeding activity, resulting in less selection as food abundance is decreased, and may also affect the physical location of grazers due to competition for dwelling construction (Tokeshi 1986). Architecture of 23 the host plant, degree of competition, time of year and life history phase also play roles in determining the type of grazers present, and thereby the overall effect of grazing on the ' periphyton community and on primary productivity. Nutrient translocation from the sediment Researchers have found that aquatic insect feces may comprise a large portion of fine particulates in lakes (e.g. Brundin 1949; Davis 1974). Deposit-collectors ingest the fecal matter from the surficial sediments, thereby playing a significant role in the continued mobilization of deposited organic matter (MacFayden 1961, McLachlan and McLachlan 1976; Petr 1977). Jonasson (1972) found that one or two species of Chironomid midges may use this fecal material and recycle as much as 20% of the gross primary production annually in lakes. Gardner et al. (1981) calculated that phosphorus excretion by tubificids and chironomids in Lake Michigan represented between 13-20% of the total phosphorus released from the sediments. Species release nutrients differently, which can affect their relative importance in nutrient cycling. Gardner er al. (1983) showed that tubificid worms release nitrogen continuously through nephridial openings located on each individual body segment and that excretion products are released apart from the gut and in a continuous pattern. Chironomids, on the other hand, release nitrogen in discrete pulses, several times per hour. Indirect pathways for nutrient translocation due to aquatic invertebrates Detrivores shred plant material and feed on detritus and the associated microbial community. Their actions change the size of particulate organic matter and increase the amount.of surface area available for microbial colonization and decomposition. This, in turn, may facilitate an increased turnover of the microbial populations and higher turnover times for nutrient resources (Merritt er al. 1984). Aquatic insects are thought to be responsible for the movement of large amounts of phosphorus and other nutrients from the lake sediments into the water column (Merritt er al. 1984) through bioturbation, the physical translocation of sediments as a result of burrowing and feeding activity, and active transport of particulate material both to the sediment surface and into deeper sediment layers. Chironomids and mayflies are the major groups involved in these 24 activities. Through their actions, nutrient rich water is pumped out of the substrate, and is replaced by oxygen rich water (Merritt er al. 1984). This in turn allows for the transfer of nutrients from the sediment to the water column (Graneli 1979). The large burrows of chironomids physically allow much more water passage and subsequently much greater mineralization and release of nutrients at the pore water interface in comparison to the passage of water around tubificids. The tube building activities of chironomids have been shown to increase the effective area for nutrient exchange by up to 50%. In the absence of macroinvertebrates the transfer of nutrients at the sediment-water interface is mainly through diffusion, which is a much slower process (Graneli 1979). Aquatic insects may also impact the form and concentration of phosphorus through mineralization of sediments (Gallepp 1979; Tessenow 1964; Davis 1974). Nutrient storage in aquatic insect biomass is not thought to be a significant sink for nutrients in most aquatic systems (Merritt er al. 1984). Emergence results in a loss of carbon and energy from the littoral zone of a lake, and subsequent translocation of nutrients as the insects move elsewhere (Cowell and Hudson 1968; Likens and Loucks 1978). 25 2.0 Methods and Materials 2.1 Experimental Procedure Three vegetation types were installed over a period of two days (June 7 t h and 8 t h, 2000) at three specific elevations within the Stave Reservoir drawdown zone. The three plant species were woolgrass (Scirpus cyperinus), lenticulate sedge (Carex lenticularis), fall rye (Secale cereale L) . Woolgrass and lenticulate sedge are perennial species that are native to the study area, whereas fall rye is an annual agronomic species. Fall rye was included in the field study as it is commonly used for shoreline vegetation programs for dust control in British Columbia reservoirs, and may help to increase biological productivity. Control samples, containing no vegetation (i.e. barren), were also installed during the same time period. Three samples of each vegetation type and three control samples were installed in the Stave Reservoir drawdown zone at each elevation (80, 78, and 76 m above sea level (asl)), so the samples could be retrieved in a time sequence (i.e. at each reservoir elevation, 9 plants and 3 barren pots were installed), approximately monthly. This installation was repeated at three locations in the reservoir to provide replication for statistical purposes. In total, 108 vegetation samples were installed. A schematic of the layout is provided in Figure 3. Time 1 Time 2 Time 3 (July 6) (August 9) (September 8) 80 m 78 m 76 m >± • = Woolgrass • = Sedge • = Fall Rye • = Barren Sediment Figure 3. Schematic of experiment layout for Stave Reservoir vegetation submergence study. This layout was repeated at three locations within the reservoir. 26 2.2 Study Site Stave Reservoir is located approximately 75 km east of Vancouver, near Mission, British Columbia. The reservoir was created in 1911 when the Stave Falls Dam was constructed. Stave Reservoir is approximately 26 km long, and has a perimeter of 80 km. It consists of the main basin (originally Stave Lake) and a 9.5 km outlet arm which was formerly part of the Stave River. The reservoir surface area covers approximately 6,200 ha, and its maximum depth at full pool is 101 m. The majority of the basin is steep-sided; however, the north and south ends of the reservoir have extensive shallows. The licensed operating range of Stave Reservoir is between 73.0 m to 82.1 m; however, it is usually operated between 76 m to 82 m. Similar to other coastal storage reservoirs in British Columbia, the majority of inflow to Stave Reservoir is from seasonal rainstorms and spring snowmelt, resulting in a cyclical drawdown pattern. Typically, the reservoir is drawn down beginning in September to allow for storage of fall rains, and again in early spring to allow for storage of spring snowmelt and run-off. Water levels are usually highest during the summer to facilitate recreational use, and in the winter to respond to increased demand for electricity (BC Hydro 2003). Three locations were chosen within the Stave Reservoir drawdown zone to provide replication for the experiment. Each location had similar site characteristics including shallow gradient, fine substrate and natural vegetation growth. Sites A and B were located in the southwest arm and Site C in the southeast arm of the reservoir. Figure 4 shows the reservoir location and the three installation locations. 27 Figure 4. Stave Reservoir Vegetation Study Sample Sites 28 2.3 Methods 2.3.1 Preparation of vegetation samples Live lenticulate sedge and woolgrass plants were removed from a donor site in the southwest arm of Stave Reservoir prior to reservoir flooding in spring 2000. Plants of similar size were chosen to minimize confounding effects due to age, and Stave Reservoir vegetation was used to ensure that the plants would be acclimated to the local flooding regime. Live plants were used in order to mimic the natural vegetation response to flooding as closely as possible. The sedge samples had an average basal diameter of 30 cm (range = 1 7 - 4 3 cm) and the height of the aboveground vegetation was on average 48 cm (range = 33 - 65 cm). The woolgrass samples had an average basal diameter of 37 cm (range = 24 - 47 cm) and aboveground vegetation height of 72 cm (range = 5 2 - 8 7 cm). Fall rye was grown in a greenhouse to ensure maximum growth prior to installation in the reservoir. The samples were grown in standard greenhouse flats (26.7 cm x 53.0 cm) using potting soil as the growth medium. The fall rye grown in each flat was divided into two samples for use in the field sampling program. The barren samples (controls) consisted of sediment collected from the Stave Reservoir drawdown zone. Each live plant was transplanted into a one gallon plant pot (top diameter of 10 in) with a series of one inch holes drilled in it to allow for transfer of water and gases to the plant roots upon submergence. The pots were lined with colour-coded burlap sacs to aid with identification of vegetation species and to contain soil and any benthic invertebrates that may enter the pots over the course of the experiment. A brick was placed in the bottom of each pot to weigh the pot down, and sand was included as filler to allow the top of the plant roots (i.e. the sediment surface) to sit flush with the lip of the pot. This was to ensure that the lip would not act as a barrier to any fauna that may chose to colonize the vegetation sample. 2.3.2 Installation of pots The vegetation samples were taken to Stave Reservoir the day before installation and placed in the water to allow the root material to become saturated. This reduced the risk of pots buoying to the surface once they were planted at depth in the reservoir. At the time of installation, the water elevation of Stave Reservoir was 81.7 m (asl) therefore divers were required to install the vegetation samples at the three locations within the reservoir. 29 Commercial divers randomly installed the appropriate number of each vegetation type at each elevation. Pots were randomly color coded for time of retrieval using coloured zap straps, (white, green or black) so divers weren't required to identify plant species when it was time to remove them (e.g. At T1, the divers would simply have to look for all plants with a white zap strap; at T2, they would be looking for the black zap straps, etc.). To easily relocate the plant samples on subsequent visits, a two to three foot length of yellow polypropylene rope was attached to each pot. The ropes floated underwater, but did not reach the surface so they provided a good marker for the divers, while not attracting attention to the sample pots from the general public. Divers installed the vegetation samples by digging holes in the reservoir substrate at the appropriate elevation and randomly burying the pots so the top of each was flush with the substrate. The installation occurred on two successive days: June 7th and 8th, 2000. 2.3.3 Retrieval of pots The vegetation samples were retrieved in a monthly time series between June and September 2000. "T1" was on 6 July 2000 (30 days post-inundation); "T2" was on 9 August 2000 (63 days post-inundation); and "T3" was on 23 August 2000 (78 days post-inundation) for the samples from the 80 m elevation band, and on 8 September 2000 (94 days post-inundation) for the 78 and 76 m samples. The "T3" 80 m samples were removed early due to rapidly decreasing reservoir elevations beginning in late August. Upon retrieval, divers eased each plant out of the substrate, being careful to not disturb any fauna associated with the foliage or sediment. Each plant was placed in a large mesh bag (dimensions = 71 cm x 102 cm), made of 1 - 2 mm wide strips of woven nylon. The bags were used to contain any organisms that may have washed off while the plant was being brought to the surface, while also allowing water to escape. Prior to bringing each plant sample on-board, a foliage sample was taken for later analysis of periphyton abundance and species composition. The vegetation was placed in a 100 ml plastic sample container and immediately preserved with Lugol's iodine-potassium iodide solution. 30 The mesh bags were then lifted onto the boat, tied shut and transported back to the soil sciences lab at the University of British Columbia for processing. 2.3.4 Environmental Data Daily surface water elevation readings recorded at the Stave Falls Dam forebay at 1300 hours were provided by BC Hydro for the course of the study. A temperature data logger was attached to the "T3" barren pot at each location, for each elevation (i.e. 9 loggers in total) to record hourly changes in water temperature over the study period. A LI-COR submersible quantum sensor (LI-250) was used to collect a profile of Photosynthetic Photon Flux Density (PPFD) (urnol s"1 m"2) at 1 m depth intervals at each location on each vegetation retrieval date. 2.3.5 Processing Samples In the lab, each sample was sectioned vertically into 2 approximately equal parts. For each half of the plant, the foliage and root were separated and processed as follows: • Foliage sample "A" was rinsed and air dried to be used for nutrient analysis; • root sample "A" was frozen to be later rinsed free of sediment and used for nutrient analysis; and • foliage and root samples "B" were placed in sealed plastic bags and preserved with a 10% formalin (3.7% formaldehyde) solution to be used for invertebrate analysis at a later date. 2.3.6 Foliar and Root Nutrient Analysis Nutrient analysis of the foliage and root material was completed by Pacific Soil Analysis Incorporated (PSAI) in Richmond, British Columbia. At PSAI the foliage and root tissue samples were rinsed with tap water over a series of nested sieves and allowed to air dry. The #8 (2.38 mm), #10 (2.00 mm) and #40 (0.42 mm) sieves were used. The samples were then oven dried at 70°C, and later moisture corrected to 100°C. 31 Tissues were ground to a powder in a coffee mill (Lavkulich 1977, 1978), and then digested using the Parkinsen and Allen method (1975). For this method, 0.5 grams of plant tissue were digested initially in concentrated sulfuric acid (H 2S0 4), followed by addition of 30% hydrogen peroxide (H 20 2) mixed with lithium sulfide (Li 2S0 4) and selenium (Se). Samples were analyzed to determine the total nitrogen (N), phosphorus (P), potassium (K), calcium (Ca) and magnesium (Mg), expressed in % (Lavkulich 1978), and either loss on ignition (i.e. organic content) or ash content (i.e. inorganic residual), also expressed in %. Total Ca, Mg and K were determined using a Perkin-Elmer Atomic Absorption Spectrophotometer (model #373). Total P was determined colorimetrically, using the ascorbic acid method for color development (Tran and Simard 1993) and measuring absorbance at 700 nm (T. Ballard, University of British Columbia Soil Science Department, pers. comm.). Total N was determined colorimetrically using a Technicon Autoanalyzer II (Technicon Autoanalyzer II Methodology, nd.). Loss on ignition was determined by muffle furnacing a sample of 100°C oven dried sample overnight at 480°C (Lavkulich 1977, 1978; McKeague 1987). Results were expressed either in % of sample lost (loss on ignition), or % of sample remaining (ash content). This provided a measure of how much sediment was associated with each sample, and was used to calculate the total organic component of each sample for use in later analyses. 2.3.7 Periphyton Analysis Periphyton samples were analyzed for species abundance and composition. A qualitative assessment was completed for all three species of vegetation, and a quantitative analysis was completed for woolgrass and sedge to determine species abundance per cm 2 of leaf material. A quantitative count was not possible for the fall rye samples due to the advanced state of decay of the foliage. Analysis procedures were developed in consultation with Dr. John Stockner based on his experience on similar projects. All periphyton analyses were completed using a Kyowa Medilux 12 compound microscope with 10 times optical lens, and 4, 10, 20, 40 and 100 objective lens. The microscope stage was equipped with a graduated micrometer to allow for easy calculation of transect lengths. 32 Qualitative counts: A small piece of vegetation from each sample was initially observed under low power (200 times) and then under high power (400 times), to see the overall structure of the plant material and attached periphyton community. A representative field of view was diagrammed for each sample. A rubber "policeman" was used to gently scrape the periphyton community from each strand of woolgrass and sedge (both sides) back into the original sample container. The foliage was then measured to provide an estimate of total surface area so a quantitative estimate of cells/cm 2 could be calculated at a later date. Samples were diluted if necessary and the total dilution volume was recorded on the count sheet. The sample containers were hand shaken for exactly 1 minute. Using an eyedropper, an aliquot of liquid was immediately retrieved from the center of the container, at approximately one third of the distance from the bottom. A drop of the sample was placed on a regular glass slide and covered with a 22 mm 2 coverslip. Excess liquid was carefully removed from the slide using a tissue. Each sample was scanned to estimate the percent biomass and abundance of each class present (Bacillariophyta; Chlorophyta; Cyanophyta; Other). The presence of genera within each class were rated based on the scale of rare (<5%); uncommon (5 - 25%); common (25 - 50%); abundant (50 - 75%); and dominant (>80%). The percent biomass and abundance of the genera in each class was also estimated. Quantitative Counts: The same sample slide was used to conduct both quantitative and qualitative counts. Using the stage micrometer to measure the transect length, all cells observed in the field of view under 400x power were counted along a series of 10 mm transects. Counting was continued until either 100 cells were counted in the sample, or a cumulative transect length of at least 80 mm was covered. Empty diatom frustules, distinguished from "living" cells by the absence of a 33 chloroplast, were not counted, nor were cells that were only partially in the field of view. The total transect length covered was recorded on the count sheet. All cells were identified to genus level, and to species whenever possible. The total numbers of cells/mL and cells/cm 2 of plant material for each species observed were calculated. Species identification was completed using the following sources: Prescott (1978); Bloomqvist and Olsen (1981); Canter-Lund and Lund (1995); and Dr. John Stockner (pers. comm.). 2.3.8 Benthos counting methodology Benthos counts were completed by Ms. Danusia Dolecki, from the University of British Columbia. The preserved root and vegetation samples were washed through a 250 |im and 1 mm mesh sieve, and all animals were retained, identified and enumerated. Micro (<1 mm) and macrobenthos (>1 mm) were counted separately. All macrobenthos was sorted manually from the 1 mm sieve. Microbenthos was enumerated by splitting the fraction of the samples that were retained on the 250 \xm sieve into subsamples using a zooplankton Folsom splitter. The subsamples were viewed in a gridded petri dish and sorted under the dissecting microscope at appropriate magnification. All visible organisms were removed from the sample and identified to genera level, with the exception of the following: • The chironomidae family was identified to tribe level; • oligochaetes were identified to family/genera level; • ostracods were identified to order; and • nematodes were identified to phylum The invertebrates were sorted, counted and identified using a G S Z Zeiss stereo microscope under suitable magnification (10 -100 times). Additional examination of crucial body parts was completed using an inverted microscope with magnification up to 400 times. Edmondson (1959), Merritt and Cummins (1984), and Pennak (1978) were consulted as taxonomic references. 34 Areal benthos densities associated with each plant were calculated by dividing the exposed soil surface area of the sample pot (i.e. 0.0314 m2) into the total number of animals associated with that sample. 2.4 Data analysis Temperature data was averaged over all three planting locations, to give one mean daily temperature for each elevation. The mean daily temperature values were then added together to give a cumulative total number of degree*days that the plants at each elevation were exposed to between TO and the time of retrieval. The degree*day data was used in the nonmetric multidimensional scaling analyses. 2.5 Statistical Analysis 2.5.1 Experimental Design The experiment is a split-split plot design with three sample locations, and three elevations (main plot) within each location. The first split is the three plant species (plus a control consisting of only sediment and no plant material) randomly assigned within the elevation bands. The second split is a time factor; where one sample of each the three species and control were randomly harvested on a monthly basis over a three-month period (Hicks and Turner, 1999). In cases when differences between foliage and root material were examined a third split was added to the analyses (SFU Statistical Consultation Service 2005). 2.5.2 Analysis Parametric Statistics Parameters of interest were analyzed using analysis of variance (ANOVA). The main effects tested were elevation, plant species, time, and in some cases, plant substrate type. The A N O V A results were examined for significance starting with the interaction effects. If an interaction was statistically significant (p = 0.05), the profile plot was used to help interpret the interaction. Further, a multiple comparison Tukey test was undertaken to show which means were statistically different and to provide an estimate of the difference between those means. 35 The estimated difference between the means and 95% confidence interval were reported. In the event that a data transformation was used, the estimated difference between the means and 95% confidence interval were back-transformed prior to reporting the results. If none of the interaction effects were statistically significant, the A N O V A results for the main effects were examined. All parametric tests were run using JMP-IN v5.1.2 (SAS Institute Inc. 2005), and a significant probability level was set at p = 0.05. ANOVA Assumptions and Data Transformations Distribution plots of the data from each treatment unit were examined and the Shapiro-Wilk test was used to test for normality. Homogeneity of variance within each treatment group was examined by looking at the sample standard deviations for each group. Variance was assumed to be homogeneous unless the ratio between the standard deviations was larger than about 5:1 (Schwarz 2005). The Levene's test statistic for homogeneous variance for each treatment unit was also examined for significance. In addition, plots of the residuals by predicted were examined for randomness, and to identify any outliers or unusual points. Data transformations were used when appropriate to improve normality and homogeneity of variance. It should be noted that A N O V A is fairly robust to unequal variances as long as the design is balanced (Schwarz 2005). A fixed model A N O V A is also relatively robust to violations of the assumption of normality with a balanced design; however non normality can affect the estimates of the components of variance for a random model (Hicks and Turner 1999). The following data transformations were undertaken to improve normality and homogeneity of variance: Vegetation nutrient data (%) was arcsine (square root (x)) transformed. Nutrient root to shoot ratios were log10(x) transformed. 36 Diatom density data was log10(x) transformed. Diatom biovolume data was square root transformed. Diatom species richness data was log10(x) transformed. Benthos species richness data (i.e. number of taxa) was not transformed; abundance data (i.e. number of individuals) was |og10(x) transformed. The ratio of diatom biovolume to benthos density was log™ (x) transformed. Another key assumption for A N O V A is that experimental units are independent from each other. This assumption is satisfied as complete plants were harvested and destructively sampled at each sample period, rather than taking samples from the same plant over time. In addition, areas that contained natural vegetation were chosen as sample sites to minimize the chance that the responses related to specific plants were influenced by other sample pots located nearby. The assumption that observations are collected randomly from a population was satisfied by ensuring that sample pots were randomly assigned during installation to T1, T2, or T3 retrieval dates using a coloured zapstrap attached to each sample pot. This ensured that the divers collected the correct number of each plant species at the proper time, without having to identify species. Plants were installed in the substrate randomly, regardless of species type or retrieval date. Non-parametric statistics The nonmetric multidimensional scaling (NMS) ordination technique was used to examine the spatial layout of periphyton and invertebrate populations associated with the three plant species and control samples at the three elevations over time. PC-ORD v4.14 (McCune and Mefford 1999) was used to run the NMS analyses. The Sorenson distance measure was used for each ordination. Ordination is used in ecology to seek and describe patterns in species composition. These techniques often help to identify the most important factors from multiple factors; separate strong patterns from weaker ones; and reveal unforeseen patterns and suggest unforeseen processes. NMS iteratively searches for the best positions of n entities on k dimensions (axes) 37 that minimizes the stress of the /c-dimensional configuration (McCune and Grace, 2002). The result is a multi-dimensional solution with the least difference, or stress, between dissimilarity (distance) in the original data set and those generated for the same samples in the reduced ordination space of the final solution (Page, 2003). According to McCune and Grace (2002), NMS is the most generally effective ordination method for examining ecological community data primarily because of biological meaningfulness and mathematical robustness. In addition, NMS can be used to assess the dimensionality of a data set. This technique is well suited to data that are non-normal. 38 3.0 Results 3.1 Environmental Data 3.1.1 Water Level The daily average surface water level at the time of sample installation (June 7 t h/8 t h) was 81.7 m. The surface water elevation at T1 (July 6th) was 81.0 m. The surface water elevation at T2 (August 9th) was 80.2 m. The final 80 m samples were retrieved on August 23 r d when the surface water elevation dropped below 80.0 m to avoid desiccation of the samples1. The surface water elevation at the final time of retrieval for the 78 m and 76 m samples (September 8 th) was 78.2 m (Figure 5). > ra o> in > o .Q ra c o '*-» ro > HI 83.0 82.0 81.0 80.0 79.0 + 78.0 77.0 76.0 75.0 -X-J2P r£ nSf> JJ2P sjfi vfiP J$> . .O 0 o,0° „SP Date — — Water Elevation (m) 80 m — 7 8 m , ,: ,: ,: , , 7 6 m • Installation (June 7/8) • T1 (July 6) • 12 (August 9) • T3 (September 8) X T3 (80) (August 23) Figure 5. Stave Reservoir daily mean surface water elevation June 1 to September 15, 2000. The early retrieval time for the 80 m T3 samples meant that the periphyton and benthic invertebrate community had approximately two weeks less to develop than the 76 and 78 m samples extracted at T3. The 80 m T3 data was included in the statistical analyses; however, in cases where the elevation or time effect on periphyton or invertebrate metrics were found to be statistically significant (either as an interaction or as a main effect), the different extraction times were taken into account for interpretation purposes. 1 Note that the 80m T3 samples at Site C were completely exposed at the time of retrieval. 39 3.1.2 Temperature The daily mean water temperature in Stave Reservoir was warmer in the shallower water (i.e. higher planting elevation), ranging from a low of approximately 9 - 1 0 . 5 °C in mid June to a peak of 18 - 21 °C in late August. Daily mean water temperature began to decline after the peak in late August until the final sample period in early September (Figure 6). Figure 6. Daily mean water temperature by elevation, Stave Reservoir 2000. 3.1.3 Light The irradiance data provided a "snap-shot" of light conditions at the time of sample installation and retrieval. This data indicates that the plants were all situated within the photic zone, defined as the area within the water column that receives enough light to support photosynthesis (Figure 7). The depth at which light attenuates to 1% of its surface value defines the lower limit of the photic zone, and is known as the compensation depth (Wetzel 1983). 40 Irradiance (umol/m2/sec) Figure 7. Irradiance by depth on sampling dates, Stave Reservoir, 2000. 3.2 Vegetation Data 3.2.1 Biomass Aboveground biomass Statistical analyses to examine the effects of elevation and plant species upon aboveground biomass over time were not undertaken due to the high variability in original plant biomass within species. A visual assessment, however, indicated that the aboveground portions of woolgrass and sedge both degraded slowly in comparison to fall rye. On average woolgrass and sedge lost approximately 30% and 55% of their initial aboveground biomass (measured as ash-free dry weight (AFDW)), respectively, by the end of the experiment. Small increases in aboveground biomass for both woolgrass and sedge were observed later in the summer. Fall rye lost approximately 80% on average of its initial A F D W by the first sample date. Very little fall rye structural material remained after the initial sample date (Figure 8 and Plates 1, 2, and 3). 41 6 0 H 0 1 2 3 0 1 2 3 0 1 2 3 * * i 11 1 Woolgrass Sedge Fall Rye Time within Species Figure 8. Mean aboveground biomass (± SE) for three plant species over time (n=99). Plate 1. Representative woolgrass samples prior to submergence (left) and at T1 (right). 42 Plate 2. Representative sedge samples prior to submergence (left) and at T1 (right). Plate 3. Representative fall rye samples prior to submergence (left) and at T1 (right). Belowground biomass Similar to the aboveground samples, no statistical analyses were undertaken to examine the effects of elevation, species and time upon belowground biomass due to the high degree of variability in sizes of the individual plants used for the study. Visual observations and preliminary, pre-inundation, measurements of basal diameter for each plant suggested that between the perennial species, the woolgrass samples tended to have slightly more belowground biomass than sedge, and that the perennial species tended to have a much larger belowground biomass in comparison to fall rye. During the study a marked decrease in fall rye belowground biomass was noted between TO and T1. The belowground biomass of sedge samples showed a decreasing trend over time, and no clear trend was apparent for woolgrass belowground biomass (Figure 9). 0 1 2 3 0 1 2 3 0 1 2 3 • i 11 1 Woolgrass Sedge Fall Rye Time within Species Figure 9. Mean belowground biomass (AFDW) (± SE) for three plant species over time (n=99). 44 Root to shoot biomass ratio The mean root to shoot ratio for biomass increased between TO and T1 for all three species, however the increase was the most pronounced for woolgrass, followed by sedge. The increase in fall rye was relatively small. In the case of woolgrass and sedge, the mean biomass of roots was at least twice that of shoots (Figure 10) at all times, reaching a maximum average of approximately 13 times the amount of root biomass to shoot biomass for sedge at T2. Note that there was a great deal of variation in sample sizes indicated by the large standard error associated with the sedge samples at 12. 0 1 2 3 0 1 2 3 0 1 2 3 • • • • i Woolgrass Sedge Fall Rye Time within Species Figure 10. Mean root to shoot biomass ratio (± SE) for three plant species over time (n=99). Biomass data for the plant above and belowground components are included in Appendix A. 45 3.2.2 Nutrient Content 3.2.2.1 Nitrogen Foliage The mean foliar nitrogen (N) content at TO, prior to flooding was highest in fall rye at 3.04%. Lenticulate sedge and woolgrass both had lower mean initial levels of nitrogen at 1.99% and 1.74%, respectively. Foliar N content declined for all species between TO and T1, and then remained fairly constant for the remainder of the study (Figure 11). 0 1 2 3 0 1 2 3 0 1 2 3 • i • • i i Woolgrass Sedge Fall Rye time within species Figure 11. Mean foliar N (%) (±SE) (non-transformed) for all three plant species over time (n=99). Since the TO plant samples were not harvested from the elevation bands within the reservoir, direct statistical comparison between pre and post-inundation foliar nitrogen content could not be completed unless it was shown that elevation (or any interaction involving elevation) was not a significant factor. Statistical analyses were initially run on data from T1 to T3, and there was 46 no evidence of an effect of elevation on mean foliar N (refer to Appendix B for the A N O V A results table). Elevation was therefore removed from the model to allow comparison of pre and post-inundation nutrient levels. The main effects tested were species and time. Foliar nitrogen data was arcsine transformed to improve normality and homogeneity of variance. When comparing pre and post-inundation foliar N levels, there was strong evidence of an interaction between species and time (F = 6.98, p <0.0001). Table 1 shows the A N O V A table for the significant interaction. The profile plot of the species by time interaction indicates that the percent foliar N changed differently over time between the plant species that were tested (Figure 12). Sedge and woolgrass behaved very similarly over time; however, fall rye had a significantly higher foliar N content at TO than all other species, and at all other time periods. Fall rye foliar N content continued to decline and reached a low at T2, whereas the foliar N content in the perennial species did not change significantly between T1 and the remaining sample periods. The estimated difference in mean foliar N content of fall rye at TO and T1 was 0.35%, with a 95% confidence interval ranging from 0.12% to 0.83%. The estimated difference in mean foliar N content of fall rye and sedge (the next highest foliar N content) at TO was 0.12%, with a 95% confidence interval ranging from 0.007% to 0.40% (the estimated differences and 95% confidence intervals have been back-transformed). Table 1. A N O V A table for the significant interaction between species and time upon mean foliar N (%) (note: data was arcsine transformed) Source Nparm DF S u m of Squares *F Ratio Prob > F species*time | 6T 6 ] 0 . 0 0 8 2 3 9 6 3 *6.9782 < 0001 * Indicates a significant F-test result Although it appears that sedge tended to have a slightly higher foliar N content than woolgrass at all time periods, the differences were not statistically significant. As well, there was no evidence that the mean foliar N content differed between any of the species at T3. The complete A N O V A table is included in Appendix B. 47 Woolgrass Sedge Fall rye 0 1 2 3 Time Figure 12. Profile plot comparing the least square means of foliar N (%) fall rye, sedge and woolgrass over time (n = 99). Roots The mean root N (%) content at TO was highest in fall rye at 1.25%. Woolgrass and sedge both had lower mean initial levels of nitrogen at 0.76% and 0.63%, respectively (Figure 13). Similar to the analyses undertaken for foliar N (%) content, there was no evidence of an effect of elevation (or any interaction involving elevation) on root N (%) content (refer to Appendix B for the A N O V A table), therefore elevation was removed from the model to allow a comparison of pre and post inundation root N (%) content between species. Data was arcsine transformed to improve normality and homogeneity of variance. There was no evidence of an interaction between time and species upon mean N root content (F = 1.31, p = 0.26). There was evidence, however of an effect of both time and species (F = 4.23, p = 0.0077 and F = 200.02, p <0.0001, respectively) upon mean N% root content. The A N O V A table for the statistically significant effects is shown in Table 2. The complete A N O V A table is shown in Appendix B. 48 1 . 6 H 0 1 2 3 0 1 2 3 0 1 2 3 • 11 1 1 i Woolgrass Sedge Fall Rye time within species Figure 13. Mean root N (%) (±SE) (non-transformed data) for all three plant species over time (n=99). Table 2. A N O V A table for the significant effect of species and time for root N% (arcsine transformed) content Source Nparm DF Sum of Squares *F Ratio Prob > F time 3 3 0.00058900 M.2265 0.0077 species 2 2 0.01858236 *200.0152 <0001 * Indicates a significant F-test result Time The profile plot for mean root N (% arcsine transformed) is provided in Figure 14. Mean root N % levels at TO were significantly different than those at T1. The estimated difference in mean root N content was 0.005%, with a 95% confidence interval ranging from 0.0003% to 0.0162% (the estimated differences and 95% confidence intervals have been back-transformed). There 49 was no significant difference between mean root N (%) levels at TO, T2, and T3, or between T1, T2, and T3. Time Figure 14. Profile plot comparing the least square means of root N (%) content pre and post-inundation for all three plant species (n=99). Species An examination of the profile plot showed that the root N (%) content of fall rye was significantly different than both woolgrass and sedge, but that woolgrass and sedge were not significantly different from each other (Figure 15). The estimated difference between the means for fall rye and woolgrass was 0.076% with a 95% confidence interval ranging from 0.055% to 0.100%, and between fall rye and sedge was 0.093%, with a 95% confidence interval ranging from 0.070% to 0.119% (the estimated differences and 95% confidence intervals have been back-transformed). Nitrogen root to shoot ratio The root to shoot (R:S) ratio for N was similar in all species at TO, ranging from 0.32 for sedge to 0.44 for woolgrass. Between TO and T1, the R:S ratio increased for all species, with the most dramatic change taking place in fall rye, which increased to 1.13. The R:S ratio remained fairly constant for the remainder of the study for all species (Figure 16). 50 c 1.4 n o> 2 1.2 H a> I 1 (O 0.8 •*-» (0 S 0.6 ^ 0.4 z 0.2 -4-1 I ° Fall Rye Sedge Species Woolgrass Figure 15. Profile plot comparing the least square means of root N (%) for fall rye, sedge and woolgrass, pre and post-inundation (n=99). 1.3H •2 1.2-1 11.0-I r^0.9H 5 0.8-J § 0-7-1 ^ 0 . 6 H &0.5-I 2 0.4-1 0 . 3 -S0.2H 0.1-0.0-0 1 2 3 0 1 2 3 0 1 2 3 ; • • • i Woolgrass Sedge Fall Rye Time within Species Figure 16. Mean nitrogen root to shoot ratio (±SE) (non transformed data) for all three plant species over time (n=99). When ANOVA was used to compare the R:S ratio for the three plant species there was strong evidence of an interaction between species and time upon the N root to shoot ratio (F = 5.67, p 51 <0.0001), therefore the interaction was examined more closely. The A N O V A table for the significant interaction is presented in Table 3. The complete A N O V A table is included in Appendix B. The ratio of root to foliar nitrogen content was log 1 0 transformed to improve normality and homogeneity of variance. Results presented here have been back-transformed. Table 3. The ANOVA table for the significant interaction between species and time upon nitrogen root:shoot ratio Source Spec iesT ime Nparm Sum of Squares *F Ratio Prob > F 0.3320085I *5.6708| < 0001| * Indicates a significant F-test result The profile plot of the effect of species by time on the nitrogen R:S ratio showed that the R:S ratio changed differently over time between fall rye and the perennial species. There was no significant difference between the pre-inundation nitrogen root to shoot ratios for any of the plant species. The R:S ratio for fall rye increased rapidly upon inundation and did not change significantly thereafter. There was a small but significant difference between the nitrogen R:S ratios for sedge TO and T1. The nitrogen R:S ratio for woolgrass did not change significantly pre and post inundation (Figure 17). The geometric mean R:S ratio for fall rye at T1 was 2.7 times that at TO (95% confidence interval was 1.9 to 4.0 times as much); the geometric mean R:S ratio for sedge at T1 was 1.7 times that at TO (95% confidence interval was 1.1 to 2.5 times as much); and the geometric mean R:S ratio for fall rye at T1 was 1.9 times that of woolgrass (the next highest ratio) at T1 (95% confidence interval was 1.3 to 2.7 times as much). Table 4 shows the mean R:S ratios for the various species by time combinations and indicates which combinations are statistically different from each other. 52 1.4 as * CO 2 « o o> *! o c CO 2 °" c jg a> ™ O) o 0> 1.2 1 H 0.8 0.6 0.4 0.2 -•—Woolgrass Sedge Fall Rye Time Figure 17. Profile plot of the interaction effect of species and time upon the mean nitrogen Root to Shoot ratio for all three plant species (n=99). Table 4. Least square means of the nitrogen Root to Shoot ratio showing statistically significant designations of species by time Level Least Sq Mean Fall Rye,2 A 1.18851 Fall Rye,1 A 1.099519 Fall Rye,3 A 0.9051 Woolgrass,3 B 0.583167 Woolgrass, 1 B 0.577666 Sedge, 1 B 0.531236 Woolgrass,2 B 0.51052 Sedge,3 B 0.505211 Sedge,2 B 0.488643 Woolgrass,0 B C 0.435571 Fall Rye.O B C 0.40321 Sedge.O C 0.31881 l e v e l s not connected 3.2.2.2 Phosphorus Foliage by same letter are significantly different. The pre-inundation mean foliar phosphorus (P) content for fall rye (0.63%) was much higher than that observed in either woolgrass (0.16%) or sedge (0.17%). Upon inundation, the mean foliar P content of fall rye fell sharply and was similar to the levels found in woolgrass and sedge foliage by the first sample period. The mean foliar P content remained relatively constant for woolgrass and sedge pre and post-inundation (Figure 18). 53 0 1 2 3 0 1 2 3 0 1 2 3 i 1 1 11 1 Woolgrass Sedge Fall Rye time within species Figure 18. Mean foliar P (%) (±SE) content (non-transformed data) for all three plant species over time (n=99). Similar to the analyses performed for nitrogen content, direct statistical comparison between pre inundation and post inundation foliar phosphorus content could not be completed unless it was shown that elevation (or any interaction involving elevation) was not a significant factor. Statistical analyses were therefore initially run on arcsine transformed foliar P data from T1 to T3. There was no evidence of an elevation effect (either as a main effect or an interaction) on mean foliar P content (F = 0.29, p = 0.76), therefore, elevation was removed from the statistical model, and a comparison of pre and post-inundation mean foliar P content was undertaken. Refer to Appendix B for the complete A N O V A table. When comparing pre and post inundation mean foliar P % content, there was evidence of an interaction between species and time (F = 45.54, p <0.0001). The A N O V A table for the significant interaction is presented in Table .5. The complete ANOVA table is included in Appendix B. 54 Table 5. A N O V A table for the signi f icant interaction between spec ies and time upon mean fol iar P (%) (arcsine transformed). Source Nparm DF S u m of Squares *F Ratio Prob > F species*time 6| 6 0.00570926 M5.5441 <0001 * Indicates a significant F-test result The profile plot of species by time indicates that the percent foliar P content of fall rye changed differently than that in the sedge and woolgrass samples between TO and T1 (Figure 19). The mean pre-inundation P% foliar content for fall rye was much higher than that found in either woolgrass or sedge, and was significantly different from all other species at all other sample periods. Upon inundation, however, the mean foliar P content in fall rye declined to a similar level as that found in woolgrass and sedge, and remained constant throughout the remainder of the experiment. The estimated difference between the means of fall rye and sedge at TO (the next closest species) was 0.15%, with a 95% confidence interval from 0.09% to 0.22% (the estimated difference and 95% confidence intervals have been back-transformed). There was no evidence that the mean foliar P content differed between any of the species at T1, T2, or T3. o -i , , , 0 1 2 3 Time Figure 19. Profi le plot of the interaction effect of spec ies and time upon mean fol iar P (%) (n=99). Roots A similar pattern to that noted for foliar P content was observed for the root material of the three species. Fall rye had a higher mean pre-inundation root P content (0.25%) than either woolgrass (0.08%) or sedge (0.06%); however, the P content fell quickly post-inundation and reached slightly higher levels to those observed for woolgrass and sedge. Percent phosphorus 55 content in woolgrass and sedge root material remained relatively constant pre and post-inundation (Figure 20). 0 1 2 3 0 1 2 3 0 1 2 3 Woolgrass Sedge Fall Rye time within species Figure 20. Mean Root P (%) (±SE) content (non-transformed data) for all three plant species over time (n=99). As with the previous analyses undertaken for foliar and root nutrient content, there was no evidence that elevation (or any of the interactions involving elevation) had an effect upon the post-inundation mean root P% (F = 0.33, p = 0.743) (the complete A N O V A table is included in Appendix B); therefore, elevation was excluded from the statistical model, and an analysis of pre and post-inundation P levels in the root material was undertaken. Data was arcsine transformed to improve normality and homogeneity of variance. When testing the effects of species and time on root P % (arcsine transformed data), there was evidence of an interaction between species and time (F = 15.25, p <0.0001). The A N O V A table 56 for the significant interaction is presented in Table 6. The complete A N O V A table is included in Appendix B. Table 6. A N O V A table for the signif icant interaction between spec ies and time upon mean root P (%) (arcsine transformed) Source Nparm DF Sum of Squares *F Ratio Prob > F species*time 6 | 6 0.00095123 *15.2480 < 0001 * Indicates a significant F-test result The profile plot of the interaction between species and time indicates that the percent root P content changed differently over time between fall rye and the perennial plant species (Figure 21). Although the mean P content in fall rye roots remained higher than woolgrass or sedge during all sample periods, the initial P content fell rapidly upon inundation. The estimated difference between the means of fall rye at TO and T1, the next highest mean, was 0.023%, with a 95% confidence interval from 0.009% to 0.042%. The estimated difference between the means of fall rye and woolgrass, the next highest mean, at TO was 0.047%, with a 95% confidence interval from 0.024% to 0.079% (the estimated differences and 95% confidence intervals have been back-transformed). There was no evidence that the mean root P% content of woolgrass and sedge differed from each other during any sample session or over time. Table 7 shows the least square means of each plant species by time, and indicates which combinations are statistically different from each other. 2 o Time Figure 21. Profi le plot showing the interaction effect of spec ies and time upon root P%, taking pre and post- inundat ion data into account (n = 99). 57 Table 7. Least square means of root P% data showing statistically significant designations of species by time Spec ies by Time (*level) Least Sq Mean ((%) Fall Rye.O A 0.250101 Fall Rye,1 B 0.121694 Fall Rye,3 B C 0.115354 Fall Rye,2 C D 0.088355 Woolgrass.O C D E 0.07977 Woolgrass, 1 D E 0.075135 Sedge, 1 D E 0.068465 Sedge,3 D E 0.061328 Sedge.O D E 0.059872 Woolgrass,3 E 0.05817 Woolgrass,2 E 0.057243 Sedge,2 E 0.055857 * Levels not connected by same letter are significantly different. Phosphorus root to shoot ratio The phosphorus root to shoot (R:S) ratio was highest for woolgrass at TO (0.49). The TO phosphorus R:S ratios for sedge and fall rye were slightly lower at 0.36 and 0.40 respectively. The phosphorus R:S ratio increased for all three species between TO and T1, with the largest increase seen in fall rye (0.97 at T1). The phosphorus R:S ratio then decreased slightly and remained constant for the remainder of the study for all three species. Woolgrass had a lower root to shoot ratio at T3 than the pre-inundation ratio (Figure 22). 58 0 1 2 3 0 1 2 3 0 1 2 3 • 1 1 1 1 Woolgrass Sedge Fall Rye Time within Species Figure 22. Phosphorus root to shoot ratio (±SE) (non transformed data) for all three plant species over time (n=99). There was evidence of a species by time interaction upon the phosphorus R:S ratio (F = 5.4497, p = <0.0001). The ANOVA table for the significant interaction is included in Table 8. The complete A N O V A table is presented in Appendix B. The nutrient root to shoot ratio was log 1 0 transformed to improve normality and homogeneity of variance. The results reported here have been backtransformed. Table 8. ANOVA table for significant interaction between species and time on mean phosphorus R:S ratio Source Nparm DF Sum of Squares *F Ratio Prob > F Species*Time 6 0.35687448 *5.4497 <0001 indicates a significant F-test result The profile plot for the interaction between species and time indicates that the phosphorus root to shoot ratio for fall rye changed differently over time than that of the perennial plant species (Figure 23). There was no significant difference in the phosphorus R:S ratio between any of the species at TO, nor did the ratio change significantly over time for the perennial species. The fall 59 rye phosphorus R:S ratio, however increased rapidly between TO and T1 and then was relatively constant for the remainder of the study. A slight decline noted in the fall rye phosphorus R:S ratio between T1 and T2 was not statistically significant. The estimated difference between the geometric mean R:S ratio for fall rye at T1 was 2.4 times that at TO with a 95% confidence interval ranging from 1.6 to 3.7 times. The estimated difference between the geometric mean phosphorus R:S ratio for fall rye at T1 was 1.8 times that for sedge, with a 95% confidence interval ranging from 1.2 to 2.6 times. Table 9 shows the statistically significant designations of species by time for the phosphorus root to shoot ratio. Figure 23. Profi le plot show ing the interaction of spec ies by time effect on the phosphorus root to shoot ratio (n = 99). 60 Table 9. Least square means of the phosphorus Root to Shoot ratio showing statistically significant designations of species by time Level Least Sq Mean Fall Rye,1 A 0.956933 Fall Rye,3 A 0.789977 Fall Rye,2 A B 0.735512 Woolgrass, 1 B C 0.535769 Sedge, 1 B C 0.529205 Woolgrass, 0 B C 0.484956 Sedge, 3 C 0.440516 Woolgrass,3 C 0.396471 Fall Rye.O C 0.392923 Sedge,2 C 0.392156 Woolgrass,2 C 0.369063 Sedge.O C 0.355865 Teve ls not connected by same letter are significantly different. Nitrogen and phosphorus data for the plant above and belowground plant biomass are included in Appendix A. 3.2.2.3 Calcium, Magnesium and Potassium The total percentages of calcium (Ca), magnesium (Mg) and potassium (K) in above and belowground plant biomass were also measured during this study. Although no further analyses were undertaken for these three macronutrients, the raw data is included in Appendix C for reference purposes. 3.3 Periphyton Data Analysis of periphyton focused mainly on diatom species assemblages, total number of living2 diatom cells per square centimeter of plant material, and total biovolume of diatoms associated with foliar samples at each elevation over time. Mean densities of the samples were highly variable. Periphyton summary data by sample period is included in Appendix D. 3.3.7 Community Composition A qualitative analysis of periphyton samples indicated that the periphyton assemblages associated with all vegetation species at all elevations over time were dominated by diatoms (Bacillariophyta) in terms of total abundance. Twenty-nine diatoms, 18 chlorophytes (green alga), and 4 cyanophytes (blue-green alga) were identified amongst all samples, for a total of 51 taxa. Table 10 contains a list of all periphyton taxa observed. Plates 4 and 5 show typical periphyton assemblages observed during the study. 61 Table 10. List of periphyton taxa, and associated biovolume per cell, observed during the study. Stave Reservoir, 2000 Biovolume/cell Class Species (nm3) Bacillariophyta (Diatoms) Achnanthes sp. 1 50 Achnanthes. minutissima Kutz. 30 Asterionella formosa (Hantz) Grun. 120 Caloneis amphisbaena 430 Caloneis sp. 430 Cyclotella stelligera CI. & Grun. 50 Cymbella sp. 1 400 Cymbella sp. 2 400 Cymbella ventricosa Kutz 1000 Eunotia arcus 50 Eunotia lunaris var. 1 (Ehr.) Grun. 500 Eunotia pectinalis 700 Eunotia sp. 6000 Fragilaria acus (formerly Synedra acus Kutzj 2000 F. ulna (formerly S. ulna (Nitz.) EhrJ 3300 F. vaucheriae (Kutz.) Peters 100 F. construens (Ehr.) Grun. 120 Frustrulia rhomboides (Ehr.) De T 3000 Frustrulia sp. 2000 Gomphonema sp. 1 2000 Gyrosigma 3500 Aulicoseira sp. 90 Navicula sp. 2 700 Nitzschia sp. 1 2000 Pinnularia sp. 1 5000 Rhopalodia 5000 Surirella sp. 2500 Tabellaria fenestrata (Lyngb.) Kutz 3000 T flocculosa (Roth.) Kutz 1000 Chlorophyta (Green alga) Ankistrodesmus Closterium 1000 Cosmarium sp. Cryptosomonas Desmid Euastrum dentriculatum Kephrion microflagellate Mougeotia sp. 1 500 Mougeotia sp. 2 5000 2 Living diatom cells were defined as those which contained a chloroplast. 62 ass B iovo lume/ce l l Spec ies (urn3) Oedegonium sp. Oocystis Rhizoclonium sp. 350 Scenedesmus Spyrogyra sp. Tetrahedron Ulothrix sp. 170 Zygnema sp. C lass Cyanophyta (Blue-green alga) Lyngbya Merismopedia 50 microcystis-like Oscillatoria limnetica Plate 4. Typical periphyton assemblages assoc ia ted with plant material. Photo taken at 600x magnif icat ion. 63 Plate 5. Typical periphyton assemblages assoc ia ted with plant material. Photo taken at 300x magnif icat ion Quantitative analysis of the periphyton community associated with fall rye was not possible due to the advanced state of foliar decay; however, qualitative analyses indicated that the dominant species within the diatom assemblages were relatively constant over time. Frustrulia sp. and Tabellaria sp. remained in the top four diatom species by percent abundance at each elevation throughout all time periods. Eunotia sp. was also frequently found within the top four diatoms by abundance. Quantitative analyses were undertaken for sedge and woolgrass. 3.3.2 Diatom Density The mean diatom densities for woolgrass and sedge were approximately 18,500 cells/cm 2 and 28,450 cells/cm 2, respectively at the first sample period. Mean diatom densities for both plant species increased by approximately 3.5 times between T1 and T2. There was very little change in mean diatom density for either plant species between T2 and T3 (Figure 24). 64 140000H 120000H 1 2 3 1 2 3 • Woolgrass Sedge Time within Species Figure 24. Mean diatom density (cells/cm2) associated with woolgrass and sedge over time (n=54). Diatom density data was log 1 0 transformed prior to conducting A N O V A to improve normality and homogeneity of variance. There was no evidence of interactions between the main effects, plant species, elevation, or time upon mean diatom density, therefore the main effects were examined directly. Of the main effects, only time had a significant effect upon mean diatom density (F = 13.00, p = 0.0001). The A N O V A table for the significant effect is presented in Table 11. Refer to Appendix F for the complete A N O V A table. Table 11. ANOVA table for the significant effect of time upon mean log-diatom density Source S S MS Num DF Num *F Ratio Prob > F Time 2.78463 1.39231 2 *13.00451 0.0001 * Indicates a significant F-test result Time There was evidence that the mean diatom density differed over time (F = 13.00, p = 0.0001), with the mean diatom density at T1 being significantly lower than at T2 and T3. The geometric mean of diatom density at T2 was 1.59 times the density at T1 (95% confidence interval was 1.212 to 2.089 times as much). There was no significant difference in the mean diatom density between T2 and T3 (Figure 25). 65 _ 70000 CM | „ 60000 = 8 50000 8 s *Z 2 40000 re c o- 30000 ° -a 20000 E re % - i 10000 Figure 25. Profile plot showing time effect upon mean diatom density (cells/cm ) associated with woolgrass and sedge. Elevation Although the 80 m samples tended to support a lower mean diatom density than the 78 m and 76 m plant samples, there was no evidence that the mean diatom density differed between the three elevations (F = 5.08, p= 0.08) (Figure 26). It is important to note that the lower mean diatom density observed for the 80 m samples could be in part due to the early removal time of the 80 m samples at T3. Figure 26. Profile plot showing elevation effect upon mean-diatom density (cells/cm ) associated with woolgrass and sedge. 66 Dominant diatom taxa by density Dominant diatom taxa by density associated with sedge and woolgrass were Frustrulia sp. (Including Frustrulia sp. and Frustrulia rhomboides) (30%), Achnanthes minutissima (9%), Tabellaria flocculosa (8%) and T. fenestrata (7%) (Figure 27). Along with Eunotia sp., these genera were often found in the top 5 for abundance for both vegetation types at each site over time and elevation. Aulicoseira sp. was occasionally within the top 5 diatom taxa by abundance, and Achnanthes minutissima became increasingly common between T2 and T3. During the final sample period, A. minutissima was the most abundant species associated with woolgrass at 80 m and 76 m, and was the second most abundant species associated with sedge at 76 m. Cymbella ventricosa 4% Cymbella sp. 3 % Eunotia sp Gomphonema 4% sp. C O / 0 / 0 E. pectinalis 6% Eunotia arcus 6% Aulicoseira sp. 7% Tabellaria fenestrata (Lyngb.) Kutz 7% Frustrulia 22% Achnanthes minutissima Kutz 9 % Frustruli rhomboides (Ehr.) De T T. flocculosa 8% (Roth.) 8% Figure 27. Dominant diatom species by density (cells/cm of plant material) associated with woolgrass and sedge vegetation. The nonmetric multidimensional scaling (NMS) ordination technique was used to examine the relationships between diatom species abundance and variables of elevation, plant species (woolgrass and sedge) and time. Ordination is a family of methods widely used in ecology to describe the relationships between species composition patterns and the underlying environmental gradients which influence these patterns. Sample units are arranged along a synthetic scale (axis) or multiple axes, so that complex relationships are graphically summarized. Similar species and samples are placed close together, and dissimilar species 67 and samples are placed far apart. Ordination helps to select the most important factors from multiple factors; separate strong patterns from weak ones; and/or to reveal unforeseen patterns or processes. It should be noted that NMS is meant to be an exploratory tool and should not be used in hypothesis driven analyses as no p-values are generated to test statistical significance (Palmer 2006), As such, interpretation of the ordination diagrams is relatively subjective. Refer to Appendix G for further information on the NMS ordination technique, data output and interpretation of NMS ordination diagrams. The final stress for a two-dimensional ordination solution describing the relationship between diatom species abundance and variables of elevation, plant species (woolgrass and sedge) and time was 0.137, and the final instability was 0.004 after 38 iterations (seed number = 871). Despite showing minor instability, the ordination was considered to be acceptable for interpretation as the stress was fairly low and the final instability was in the magnitude of 10"3 (McCune and Grace 2002). The cumulative variance represented by each axis based on the r2 between distance in ordination space and distance in the original spaces for axes 1 and 2 was r2 = 0.258 and 0.735, respectively. The two-dimensional ordination of the variables indicated that there was a fairly clear grouping of diatom species density by time, with T1 grouping differently than T2 and T3 (Figure 28). This suggests that the composition of community density was not yet clearly established at T1, but was established by T2, and changed little thereafter. There were no clear patterns of diatom species density by elevation or plant species. Refer to Figures H-1a and H-1b in Appendix H for the ordination diagrams for diatom species by elevation and plant species, respectively. 68 Figure 28. Two-dimensional ordination of diatom density showing groupings by time. Distances between sample units approximate dissimilarity in species composition. Overlay plots of the top five diatom taxa by density were examined to determine if any of the variables of elevation, plant species (woolgrass and sedge) and time appeared to have an effect on diatom species colonization patterns. These taxa were chosen for further examination due to their dominance in terms of overall density, and as such, can be considered indicators of community development on macrophytes in the Stave Reservoir drawdown zone. Results of the overlay plots are described in the following paragraphs. The plots themselves are included in Appendix H. It should be noted that the interpretations of the overlay plots are subjective, and although not completed as part of this thesis, further analysis using ANOVA could be undertaken to determine if the trends and tendencies observed for individual diatom species are statistically significant. The overlay plots for Frustrulia sp., accounting for 22% of all diatoms by density, suggest that there was an increase in density between T2 and T3 (Figure H-2a), and that Frustrulia sp. was found more commonly at the two lower elevations (Figure H-2b). Sedge appeared to support a slightly greater density of Frustrulia sp. in comparison to woolgrass (Figure H-2c). 69 Achnanthes minutissima, accounting for 9% of all diatoms by density, appeared to become increasingly more common between T2 and T3 (Figure H-3a). No clear patterns by elevation (Figure H-3b) or plant species (Figure H-3c) were apparent. Frustrulia rhomboides, accounting for 8% of all diatoms by density appeared to become increasingly common over time (Figure H-4a), and were associated more with sedge than woolgrass (Figure H-4c). The overlay plots also suggest that F. rhomboides had a higher density associated with plants at the 76 and 78 m elevations in comparison to those within the 80 m elevation band (Figure H-4b). Tabellaria flocculosa , accounting for 8% of all diatoms by density, appeared to increase in density over time (Figure H-5a), and was found more abundantly associated with plants at the lower two elevations (Figure H-5b). There were no notable patterns based on plant species (Figure H-5c). The density of T. fenestrata appeared to increase between T1 and T2; however, T2 and T3 seemed to be similar in terms of density (Figure H-6a). The overlay plots also suggest that there was a slight increase in T. fenestrata density associated with plants from the lower two elevation bands (Figure H-6b). No clear patterns related to plant species were detected (Figure H-6c). 3.3.3 Diatom Biovolume Mean diatom biovolume associated with woolgrass and sedge increased approximately 3 - 4 times for both plant species between T1 and T2, and then remained fairly constant for the remaining sample periods. Sedge appeared to support a larger mean biovolume of diatoms (u.m3/cm2 plant material) in comparison to woolgrass, however it should be noted that biovolume results were highly variable, especially at T2 and T3 (Figure 29). Diatom biovolume summary data by sample period is included in Appendix E. 70 5 I 1 2 3 1 2 3 Woolgrass Sedge Time within Species Figure 29. Mean diatom biovolume (u.m3/cm2 plant material) (±SE) (non-transformed data) associated with woolgrass and sedge over time (n=54). Diatom biovolume data was square root transformed to improve normality and homogeneity of variance for ANOVA. There was evidence of an interaction between elevation and time upon mean diatom biovolume (F = 3.53, p = 0.02). The ANOVA table for significant interaction is presented in Table 12. Refer to Appendix F for the complete A N O V A table. Table 12. ANOVA table for the significant interaction of elevation and time upon mean diatom biovolume Source S S MS Num DF Num *F Ratio Prob > F ElevationTime 1.33e+8 3.325e7| 4 *3.5317 0.0211 * Indicates a significant F-test result The profile plot showing the interaction between elevation and time upon mean diatom biovolume indicates that the mean diatom biovolume showed different trends over time for all three elevations. The mean diatom biovolume (square root transformed) associated with plants at all three elevations was the same at T1, and did not change over time for the 80 m samples. The mean diatom biovolume increased steadily over time for the 78 m samples, whereas the 71 mean diatom biovolume associated with the 76 m samples declined between 12 and T3 (Figure 30). 5 0.00E+00 J 1 1 1 2 3 Time Figure 30. Profi le plot showing the interaction of elevation and time upon mean diatom biovolume ( u m 3 / c m 2 plant material) assoc ia ted with woo lgrass and sedge (n= 54). Although the trends over time were different with respect to mean diatom biovolume, there was no evidence that the mean diatom biovolume within each sample date for the 78 m and 76 m samples were different from each other. There was evidence that the mean diatom biovolume differed between the 80 m and 76 m elevation samples at 12. The estimated difference was 3.92E+07 u.m3/cm2, with a 95% confidence interval from 5.72E+04 u.m3/cm2to 1.51 E+08 um 3 /cm 2 . Similarly, there was evidence that the mean diatom biovolume differed between the 80 m and the 78 m samples at T3. The estimated difference was 7.43E+07 u.m3/cm2, with a 95% confidence interval from 6.76E+06 u.m 3/cm 2to 2.14E+08 um 3 /cm 2 (difference between the mean and 95% confidence interval data was back-transformed). Table 13 contains a complete list of the square-root least square means of the diatom biovolume data showing statistically significant designations of elevation by time. 72 Table 13. Least square means of the diatom biovolume data (um3/cm2) *Level Least Sq Mean (u.m3/cm2) 78,3 A 2.336E+08 76,2 A 1.773E+08 78,2 A B 1.487E+08 76,3 A B C 1.426E+08 80,2 B C D 4.979E+07 80,3 B C D 4.438E+07 78,1 C D 3.742E+07 80,1 D 3.188E+07 76,1 D 3.165E+07 * Levels not connected by same letter are significantly different Species There was no evidence of a difference between the mean diatom biovolume associated with sedge and woolgrass (F = 4.7981, p = 0.1599). Dominant diatom taxa by biovolume Frustrulia, Tabellaria, and occasionally Eunotia were the dominant diatom genera based on biovolume estimates. Rhopalodia, due to its large size, was the most dominant genus by biovolume at T3 associated with 80 m sedge samples. Although A. minutissima cells were relatively abundant at T3, the species was not among the top ten dominant species by biovolume due to its small size. Figure 31 shows the dominant diatom taxa by biovolume. 73 Fragilaria ulna 5% E. pectinalis Rhopalodia sp. 2% 3% T. flocculosa (Roth.) Kutz 5% Gomphonema sp. 6% Tabellaria fenestrata (Lyngb.) Kutz 14% Frustrulia sp. 26% Frustrulia rhomboides (Ehr.) De T 15% Eunotia sp. 15% Figure 31. Dominant diatom taxa by biovolume (nm3/cm2 plant material) associated with woolgrass and sedge plant species. NMS was used to examine species groupings by biovolume. The final stress in the analysis for a two-dimensional solution was 0.130, and the final instability was 0.004 after 35 iterations (seed number = 1470). The cumulative variance represented by each axis based on the r2 between distance in ordination space and distance in the original spaces for axes 1 and 2 was r2 = 0.318 and 0.816, respectively. Similar to the analyses for diatom density, ordination diagrams of the variables indicate that there was a fairly clear grouping of diatom species biovolume by time, with T1 grouping differently than T2 and T3 (Figure 32). This suggests that the community composition related to biovolume was not clearly established by T1, but was well developed by T2 and changed little thereafter. There were no clear patterns by elevation or plant species. Refer to Figures 1-1 a and 1-1 b in Appendix I for the ordination diagrams for diatom species biovolume by elevation and plant species, respectively. 74 Diatom biovolume Time Axis 1 Figure 32. Two-dimensional ordination of diatom biovolume showing groupings by time. Distances between sample units approximate dissimilarity in species composition. Overlay plots of the top four diatom taxa by biovolume were examined to determine if any of the variables of elevation, plant species (woolgrass and sedge) and time appeared to have an effect on diatom species colonization patterns. These taxa were chosen for further examination due to their dominance in terms of overall diatom biovolume, and as such, can be considered indicators of community development on macrophytes in the Stave Reservoir drawdown zone. Results of the overlay plots are described in the following paragraphs. The plots themselves are included in Appendix I. Note that the interpretations of the overlay plots are subjective, and although not completed as part of this thesis, further analysis using A N O V A could be undertaken to determine if the trends and tendencies observed for biovolume of individual diatom species are statistically significant. When examining the NMS ordination diagrams for the top four species by biovolume, there appeared to be an increase in Frustrulia sp. biovolume over time (Figure l-2c). Similar to the density data, it seemed as though Frustrulia sp. tended to have a higher biovolume at the 76 m 75 and 78 m elevations in comparison to the 80 m samples (Figure l-2b). Sedge appeared to support a higher biovolume of Frustrulia sp. in comparison to woolgrass (Figure l-2c). The NMS overlay plots for Frustrulia rhomboides, accounting for 15% of all diatoms by biovolume, suggest that the biovolume of Frustrulia rhomboides increased over time (Figure I-3a), and appeared to be higher for sedge than woolgrass (Figure l-3c). There also appeared to be a difference between F. rhomboides biovolume at 76 and 78 m in comparison to the 80 m elevation (Figure l-3b). Eunotia sp. was the third most dominant diatom taxa by biovolume, accounting for 15% of the total diatom biovolume. The NMS overlay plots suggest that the biovolume of Eunotia sp. tended to increase over time (Figure l-4a) and was also higher at the two lower elevations (Figure l-4b). No differences were apparent between species (Figure l-4c). Tabellaria fenestrata was the fourth most dominant diatom species, accounting for 14% of the total diatom biovolume. No differences were noted between the biovolume at 12 and T3; however, a change in T. fenestrata biovolume between T1 and T2 was apparent (Figure l-5a). The overlay plots also suggest that the biovolume of T. fenestrata was higher at the 76 m and 78 m elevations in comparison to the 80 m elevation band (Figure l-5b). There were no clear differences between plant species (Figure l-5c). 3.3.4 Diatom Taxonomic Richness Taxonomic richness is the simplest measure of biodiversity and is simply a count of the number of different taxa noted per unit area. The number of species in a community tends to increase with the complexity of the food web, and with the extent of niche overlap, especially in the lower trophic levels of the food chain (Wetzel 1983). Diatom taxonomic richness ranged from 2 - 2 0 for woolgrass and 6 - 2 1 for sedge (Figure 33). Diatom taxonomic richness data was log 1 0 transformed for A N O V A to improve normality and homogeneity of variance. There was evidence of an interaction between elevation and time (F = 4.0718, p = 0.0117) on mean taxonomic richness. The A N O V A table for the significant interaction is included in Table 14. The complete ANOVA table is presented in Appendix F. 76 H J M i n (Count) 111 Max (Count) Woolgrass Sedge Time within Species Figure 33. Diatom taxonomic richness associated with woolgrass and sedge (n=54). Table 14. ANOVA table for significant interaction between elevation and time upon mean diatom log-taxonomic richness associated with woolgrass and sedge vegetation Indicates a significant F-test result The profile plot for elevation by time indicates that the mean taxonomic richness changed differently over time for the samples from the 80 m elevation band than the samples from the 76 and 78 m elevation bands. Taxonomic richness was statistically the same at T1 for all three elevations, but began to decline over time for the 80 m elevation. Taxonomic richness did not change between T1 and T2 for the 78 and 76 m samples. It did increase slightly, although not significantly, by T3 for the 78 and 76 m samples (Figure 34). The geometric mean diatom taxonomic richness at 76 m was 1.45 times the geometric mean diatom taxonomic richness at 80 m at T3 (95% confidence interval was 1.12 to 1.87 times as much). Note that the geometric mean and 95% confidence interval ratios have been back-transformed. 77 CO CO CD 20 -| c .c CO 18 -o c 16 Q) 14 -o E cu 1- 12 -o 03 10 -c 3 o CT 8 X (7) ro 6 H CO E CD <D 4 2 o _l TO 0 • 76 m • 78 m • 80 m 2 Time Figure 34. Profile plot showing the interaction effect of elevation by time on mean diatom taxonomic richness There was no evidence of any other interactions, nor was there evidence that the mean diatom taxonomic richness differed between plant species (F = 1.4444, p = 0.3524). 3.4 Benthic Organisms 3.4.1 Community Composition 70 benthic taxa including naidid, enchytrid, and lumbriculid worms, nematodes, ostracods, tubificids, water mites, gastropods, aquatic insects, beetles, terrestrial insects and zooplankton were found associated with above and belowground plant and control samples. A total of four juvenile pacific lamprey (Lampetra tridentata) were identified in two woolgrass and one fall rye sample from Site C during the August 9 t h sample period. The lamprey were found at the plant/sediment interface in samples at the 78 and 76 m elevations. The most abundant taxa found on both the aboveground and belowground samples were an assemblage of oligochaete worms (mainly Lumbriculidae and to a lesser extent Enchytraeidae), chironomids (mainly Chironomini, Orthocladiinae, and Tanypodinae) nematodes (Nematoda), and ostracods (Ostracoda). When considering total number of individuals, oligochaetes and chironomids were the most dominant, accounting for 63% of all benthic organisms. A complete list of benthic taxa observed during the study is included in Table 15. Refer to Appendix J for benthos summary data. Figure 35 shows the dominant benthic organisms by abundance. 78 Table 15. List of benthic taxa associated with all foliage and root samples, and barren samples in Stave Reservoir, 2000 Family or Order unless otherwise (subfamily) or Life Taxa indicated' {tribe} Genus Stage2 Code3 Benthic Invertebrates Diptera Chironomidae (Orthocladiinae) L (Orthocladiinae) Corynoneura L s1 (Orthocladiinae) Cricotopus/Orthocladius L s2 (Orthocladiinae) Orthocladius L s3 (Orthocladiinae) Eukiefferiella L s4 (Orthocladiinae) Psectrocladius L s5 (Orthocladiinae) Heterotrissocladius L s6 (Orthocladiinae) unrecogn. Ortho. L s7 {Tanytarsini} Tanytarsus L s8 {Tanytarsini} Rheo/Paratanytarsus L s9 {Tanytarsini} Constempellina L s10 {Chironomini} Chironomus L s11 {Chironomini} Paracladopelma L s12 {Chironomini} Phaenopsectra L s13 {Chironomini} unrec 1st instar Chironomini L s14 (Tanypodinae) Thienemannimyia L s15 (Tanypodinae) Procladius L s16 (Prodiamesinae) Monodiamesa L s17 Chironomidae unrec Chironomidae pupae PU s18 Chironomidae unrec Chironomidae adult AD s19 Ceratopogonidae Bezzia/Probezzia L s20 Sphaeromias L s21 Empididae Chelifera L s22 Ephydridae unrec. Ephydridae L s23 unrec Diptera unrec Diptera adult AD s24 unrec Diptera unrec Diptera larvae • L s25 unrec Diptera unrec Diptera pupae PU s26 Oligochaeta* Naididae s27 Enchytraeidae s28 Tubificidae (C) - s29 Tubificidae Tubificidae eggs E s30 Tubificidae unrec Tubificidae s31 Lumbriculidae Lumbriculus s32 Lumbriculidae Eisenniella s33 unrec Oligochaeta Megadriles-Earth Worms s34 Nematoda*** s35 Ostracoda** A s36 B s37 Arachnoidea* "spider" s38 79 —mm ., Family or Order unless otherwise .indicated1 - (subfamily) or {tribe} .'' Genus !>"• Life Stage2 Taxa Code "Hydracarina" Hygrobatidae s39 Hygrobatidae (D) S40 Lebertiidae s41 Lebertiidae (A) s42 Lebertiidae (B) s43 Unionicolidae s44 Young mites s45 Oribatei (Soil Mites) Eremaeidae Hydrozetes s46 Collembola s47 Gastropoda* Planorbidae Gyraulus s48 Pelecypoda* Lymnaeidae Lymnaea s49 Sphaeriidae Sphaerium s50 Trichoptera Limnephilidae Clistoronia L s51 Hydroptilidae unrec. Hydroptilidae larvae L s52 Leptoceridae Mystacides L s53 Ephemeroptera Ephemerellidae Ephemerella L s54 Plecoptera Capniidae Capnia L s55 Homoptera Aphidae AD s56 Coleoptera Dytiscidae Hydroporus L s57 Staphylinidae Unrec. Staphylinidae larvae L s58 Staphylinidae adult Staphylinidae AD s59 Gyrinidae Gyrinus L s60 unrec. Lepidoptera Lepidoptera larvae unrec. Lepidoptera larvae L s61 Thysanoptera Thripidae s62 Amphipoda Crangonyctidae Synurella s63 Diplopoda* unrec. Millipedes s64 Turbellaria* Planariidae Polycelis s65 Zooplankton and eggs Chydorus s66 Mite eggs s67 Eury/Simo eggs s68 Ephippium eggs s69 Fish Lampetra tridentata (Pacific Lamprey) s70 1. Other classifications are Class*, Subclass**, or Phylum*** 2. L=iarvae, AD = adult, PL) = pupae, E = eggs 3. Taxa code = sequential numbering of benthic taxa (to be referred to when viewing raw data tables in Appendix J) 80 Nematoda Figure 35. Dominant benthic organisms by abundance associated with all three plant species in Stave Reservoir, 2000. Nematodes were found in the greatest overall number of samples (87% of all samples, n=108); Enchytraeidae (oligochaete) were found in 83% of all samples; and the chironomid, Phaenopsectra, was found in 75% of all samples. 3.4.2 Benthos Abundance Mean benthos abundance increased between T1 and 12 for all plant species and control samples. The highest mean density at T1 was associated with sedge (415 individuals or 13,250 animals/m 2, standardized to the exposed surface area of the sample pot), followed by woolgrass (347 individuals or 11,000 animals/m2), fall rye (163 individuals or 5,200 animals/m2) and lastly control (18 individuals or 575 animals/m2). Sedge and woolgrass both peaked at T2 in terms of mean benthos abundance (673 and 497 individuals, or 21,000 and 15,800 animals/m 2, respectively). The mean benthos abundance associated with fall rye continued to increase 81 between T2 (467 individuals or 14,900 animals/m2) and T3 (648 individuals or 20,600 animals/m2). It should be noted that the T2 and T3 results were highly variable (Figure 36). 800 H 1 2 3 1 2 3 1 2 3 1 2 3 l l l I I 1 l 1 w CD CD <n O) >» ro Sed or O) o Sed ro o LL o o O Time within Species Figure 36. Mean benthos abundance (benthic organisms/sample) (±SE) for all three plant types and control samples over time (n = 108). Vegetation species comparison Benthos data was log 1 0 transformed for A N O V A to improve normality and homogeneity of variance. When looking at benthic invertebrate individuals associated with the entire plant (i.e. not standardized to plant biomass), there was no evidence of any interactions between the main effects upon mean benthos abundance. There was, however, strong evidence that the mean benthos abundance differed between plant species (F= 116.3134, p <0.0001) and also over time (F = 12.3369, p O.0001). The A N O V A table for the significant main effects is presented in Table 16. The complete A N O V A table is shown in Appendix K. 82 Table 16. ANOVA table for the significant effects of species and time upon mean benthos log-abundance Source S S MS Num DF Num *F Ratio Prob > F Species 26.2968 8.7656 3 *116.3134 <0001 Time 1.46706 0.73353 2 *12.3369 <0001 * Indicates a significant F-test result Species The profile plot for the effect of plant species on the mean benthos abundance indicates that significantly more benthic organisms were associated with plant material in comparison to control samples. There was no evidence of a difference in the mean abundance of benthic organisms associated with the three plant species (Figure 37). The geometric mean density of benthic organisms associated with the fall rye samples was 2.7 times that of the control samples (95% confidence interval was 2.09 to 3.5 times as much). 500 I 450 H 3 400 8 £ 350 -I n 8 300 = „ 250 H .a < o <o a> CO re 200 150 -100 -50 0 Control Fall Rye Sedge Species Woolgrass Figure 37. Profile plot comparing mean total abundance of benthic organisms associated with control and plant samples (n = 108). Time There was no evidence that the mean abundance of benthic organisms differed between the T2 and T3 samples. The mean abundance of benthic organisms at TI , however, was significantly lower than at both T2 and T3. The geometric mean density of benthic organisms associated with samples at T2 was 1.30 times that of the samples at T1 (95% confidence interval was 1.11 to 1.47 times as much) (Figure 38). 83 Figure 38. Profile plot showing mean abundance of benthic organisms over time (n=108). When looking at the mean benthos density (i.e. number of individuals per gram of plant material), there was evidence of an interaction between time and elevation (F = 4.6055, p = 0.0020) and between species and substrate (F = 5.4340, p = 0.0059). The ANOVA table for the significant interactions is included in Table 17. The complete ANOVA table is presented in Appendix K. Table 17. ANOVA table for the significant interactions upon mean log-density of benthic organisms (log-number of individuals/gram of plant material) Source S S MS Num DF Num *F Ratio Prob > F Species*substrate 1.57744 0.78872 2 *5.4340 0.0059 Time*Elevation 2.67388 0.66847 4 •4.6055 0.0020 * Indicates a significant F-test result Time by elevation interaction Closer examination of the profile plot for the interaction between time and elevation shows that the mean invertebrate density was very similar between all three elevations at T1 and T2; the 76 m and 78 m samples were also similar at T3, however, the 80 m samples tended to support a significantly lower number of invertebrates per gram of plant material at T3. The geometric mean density of benthic organisms/gram of plant material at T3 for the 78 m samples was 1.57 times the density of benthic organisms/gram of plant material for the 80 m samples (95% confidence interval was 1.04 to 2.35 times as much). Differences in mean benthic density at other time periods were not significant. 84 The 76 m samples did not differ significantly from each other over time, however, there was evidence that the mean density of individuals per gram of plant material differed between the 78 m samples at T1 and T3. The geometric mean density of benthic organisms/gram plant material for the 78 m samples at T3 was 1.53 times the density of benthic organisms/gram plant material at T1 (95% confidence interval was 1.02 to 2.30 times as much) (Figure 39). 2 Time Figure 39. Profile plot showing the interaction of elevation by time upon mean benthic organism density Species and substrate interaction (belowground vs. aboveground vegetation) A N O V A was used to compare the capacity of aboveground and belowground biomass of the different plant species to support a benthic community. Abundance data was expressed in terms of log density of organisms per gram of plant material. There was evidence of an interaction between species and substrate (F = 5.4340, p = 0.0059) upon mean log density of organisms per gram of plant material. Figure 40 indicates that fall rye foliar and root biomass supported a similar mean density of organisms per gram of plant material, whereas both sedge and woolgrass had a significantly greater density of benthic organisms associated with the foliar material in comparison to roots per gram of plant material. In addition, sedge foliage supported a significantly greater density of organisms than woolgrass foliage. The estimated geometric mean number of benthic organisms per gram of sedge foliage was 1.62 times that of woolgrass foliage (95% confidence interval was 1.19 to 2.18 times as much). 85 The estimated geometric mean number of individuals per gram of sedge foliage was 1.70 times that of sedge roots (95% confidence interval was 1.26 to 2.31 times as much). The estimated geometric mean number of individuals per gram of woolgrass foliage was 1.37 times that of woolgrass roots (95% confidence interval was 1.01 to 1.85 times as much). Foliage Roots Substrate Figure 40. Profile plot showing the interaction effect of species by substrate on benthic organism abundance per gram of plant material Macro vs. Microbenthos The majority of benthic organisms were classified as microbenthos (< 1 mm in size), accounting for 87% - 89% of all benthos observed on all plant species and strata over the three sample periods. NMS NMS analysis was used to examine the relationship between non-transformed invertebrate species abundance data and variables of sample time, elevation and plant species. The final stress in the analysis for a three-dimensional solution was 0.144, and the final instability was 0.004 after 35 iterations (seed number = 146). The cumulative variance represented by each axis based on the r2 between distance in ordination space and distance in the original spaces for axes 1, 2 and 3 was r2 = 0.380, 0.558 and 0.753, respectively. Ordination diagrams of the variables suggest that there was a strong grouping related to plant species type. The benthic communities associated with perennial species (woolgrass and sedge) tended to group together, whereas the benthos assemblage associated with fall rye was separate, as was the grouping associated with the control samples (Figure 41). 86 There were no clear patterns by time or elevation. Refer to Figures L-1a and L-1b in Appendix L for the ordination diagrams for benthos abundance by time and elevation, respectively. Benthos abundance Axis 1 Figure 41. Ordination for non-transformed benthos abundance data showing a strong grouping by plant species. Distances between sample units approximate dissimilarity in species composition. The ellipses show approximate groupings. Overlay plots were used to examine colonization patterns of the four dominant benthic taxa with respect to variables of time, plant species, and elevation. These taxa were chosen for further examination due to their dominance in terms of overall abundance, and as such, can be considered indicators of benthic community development associated with macrophytes in the Stave Reservoir drawdown zone. Results of the overlay plots are described in the following paragraphs. The plots themselves are included in Appendix L. Note that the interpretations of the overlay plots are subjective, and although not completed as part of this thesis, further analysis using A N O V A could be undertaken to determine if the trends and tendencies observed for abundance of individual benthic taxon are statistically significant. 8 7 Representatives from the subclass Ostracoda, accounting for 22% of all benthic organisms, appeared to be most abundant at T3 (Figure L-2a), and within the 76 m elevation band (Figure L-2b). Ostracods also appeared to be associated more abundantly with fall rye than with either of the perennial plant species (Figure L-2c). Family Enchytraeidae within Class Oligochaeta, accounting for 21% of all benthic organisms in this study, appeared to be marginally more abundant at T1 and T3 in comparison to 12 (Figure L-3a). No strong patterns were noted related to elevation (Figure L-3b); however, the overlay plots suggest that there was a strong preference for perennial vegetation in comparison to fall rye or barren substrate (Figure L-3c). Phaenopsectra, the most plentiful chironomid taxa, (chironomids accounted for 26% of all benthic organisms by abundance) appeared to be most abundant at T3 (Figure L-4a), with no strong patterns for elevation noted (Figure L-4b). Phaenopsectra seemed to show a slight preference to colonize fall rye over the perennial species (Figure L-4c). Nematodes accounted for 8% of the total benthic organism community. They appeared to be most abundantly associated with the 80 m elevation samples early in the experiment (T1) (Figure L-5a). No patterns for elevation were noted (Figure L-5b). The overlay plot for Nematoda colonization by plant species suggests that nematodes were associated mostly with sedge, followed by woolgrass. There was very little noticeable difference in colonization patterns of nematodes on fall rye in comparison to barren substrate (Figure L-5c). 3.4.3 Benthos Taxonomic Richness Benthic taxonomic richness data was highly variable. Results ranged from 2 taxa in one control sample to 25 taxa in a single woolgrass sample. Maximum taxonomic richness associated with fall rye was 20. Maximum taxonomic richness in control pots was 15. No vegetated sample had less than nine benthic taxa associated with it. A N O V A was used to examine the effects of species, time and elevation upon taxonomic richness of benthic communities associated with the three plant species and control samples. No data transformations were necessary. There was evidence of an interaction between elevation and time on the taxonomic richness of the benthic communities associated with the three plant species and control samples (F = 2.72, 88 p = 0.04). There was also strong evidence of an effect of plant species upon the mean number of taxa (F = 28.79, p = 0.0006), The A N O V A table for the significant interaction and main effect is presented in Table 18. The complete A N O V A table is shown in Appendix K. Table 18. ANOVA table for the significant interaction between elevation and time and the main effect, species, upon the mean number of benthic taxa Source S S MS Num DF Num *F Ratio Prob > F Species 753.425 251.142 3 *28.7885 0.0006 Elevation*Time 101.942 25.4855 4 *2.7242 0.0401 * Indicates a signi ficant F-test result The profile plot for the interaction between elevation and time upon mean taxonomic richness shows a similar trend as that for the mean abundance of benthic individuals, with the samples from the 80 m elevation at T3 behaving differently than samples from the lower two elevation bands (Figure 42). The number of benthic taxa associated with samples at the 78 m and 80 m elevations increased significantly between T1 and T2. The number of taxa associated with the 76 m samples also increased between T1 and T2; however the change was not significant. Although there was a drop in the number of taxa at 80 m at T3, none of the samples from T2 and T3 for any elevation were statistically different from each other. The estimated difference in the mean number of taxa for the 78 m samples between T1 and T2 was 4.75 taxa (s.e. = 1.25), with a 95% confidence interval ranging from 0.70 to 8.80 taxa. The estimated difference in the mean number of taxa for the 80 m samples between T1 and T2 was 4.08 taxa (s.e. 1.25), with a 95% confidence interval ranging from 0.03 to 8.14 taxa. 89 • — 7 6 m * - 7 8 m -•—80 m 1 2 3 Time Figure 42. Profile plot showing interaction between elevation and time upon the mean taxonomic richness of the benthic communities associated with all three plant species and control samples (n = 108). Table 19 contains a complete list of the log-least square means of the benthic species richness data showing statistically significant designations of elevation by time. Table 19. Least Square Means of the benthic organism taxa data showing statistically significant designations of elevation by time Level Least Sq Mean 78,3 A 14.250000 78,2 A 14.166667 80,2 A 13.666667 76,3 A B 13.458333 76,2 A B C 11.583333 80,3 A B C 10.500000 80,1 B C 9.583333 78,1 C 9.416667 76,1 C 8.250000 Levels not connected by same letter are significantly different. Species Control samples supported a significantly lower number of taxa than all vegetation species (Figure 43). The estimated difference in the mean number of taxa for the control and sedge samples (the next lowest mean number of taxa) was 4.76 taxa (s.e. 0.82), with a 95% confidence interval ranging from 1.94 to 7.58 taxa. Although woolgrass tended to support the highest mean number of taxa at 14.3, the mean number of benthic taxa was not significantly different between vegetation species. 90 16 14 (0 « « 12 10 8 6 4 2 o «-E w z S Control Fall Rye Sedge S p e c i e s Woolgrass Figure 43. Profile plot showing mean number of taxa associated with each plant species and control samples (n = 108). 3.5 Relationship between diatom biovolume and benthos density The ratio of diatom biovolume to benthos density for the perennial species was examined to determine if there were noticeable linkages between the development of the diatom and benthic communities associated with the aboveground vegetation in Stave Reservoir. Fall rye was excluded from the analysis as no quantitative diatom data was available. The ratio of diatom biovolume to benthos density generally increased over time for both species, with woolgrass having a higher mean ratio at each sample period (Figure 44). A N O V A was used to test the effects of plant species, elevation and time on the diatom biovolume to benthos density ratio. Ratios were log 1 0 transformed to improve normality and homogeneity of variance. 91 >J.0e+7H | 9.0e+6H •o w 8.0e+6H o f 7.0e+6H "§ 6.0e+6H J 5.0e+6H o .9 4.0e+6H E 3.0e+6H | 2.0e+6H ro 1.0e+6H S Oe+OH 1 1 Woolgrass Sedge Time within Species Figure 44. Ratio of diatom biovolume (um3/cm2) to benthos density (organisms/gram plant material) (N= 54). There was no evidence of any significant interaction effects. Of the main factors, only time was found to have a significant effect on the diatom biovolume to benthos density ratio (F = 3.78, p = 0.0373). The A N O V A table for the significant main effect is included in Table 20, and the complete A N O V A table is in Appendix M. The profile plot for the effect of time on the diatom biovolume to benthos density ratio showed that the geometric mean diatom biovolume to benthos density ratio at T3 was 2.2 times that at T1 (95% confidence interval ranging from 1.07 to 4.03 times as much) (Figure 45). There was no significant difference between the diatom to benthos ratios at T1 and T2, nor was there a difference between T2 and T3. Table 20 . ANOVA table for significant effect of Time on the ratio of diatom biovolume to benthos density * Indicates a significant F-test result 92 4.00E+06 T o £ 3.50E+06 -& ~ 3.00E+06 -£ & 2.50E+06 -•5 2 2.00E+06 -I I 1.50E+06 -p | 1.00E+06 -o < •S 5.00E+05 -5 O.OOE+00 -• 1 2 3 Time Figure 45. Profile plot showing the effect of time on the ratio of diatom biovolume to benthos abundance for communities associated with woolgrass and sedge 4.0 Discussion 4.1 Environmental data The daily mean water temperatures recorded in Stave Reservoir of approximately 9 - 1 0 . 5 °C in mid June to a peak of 18 - 21 °C in late August were similar to water temperatures found in other coastal temperate lakes (Stockner and Shortreed 1985), and were relatively consistent with temperatures recorded by BC Hydro in Stave Reservoir between 1996 and 2001 (personal observation; Beer 2004). McAfee (1980) suggested that temperature is the main influence of climate on reservoir biological productivity (i.e. a higher temperature will cause faster growth and onset of maturity and consequently, a higher rate of production). Also, higher water temperatures may have the ability to facilitate greater decomposition of plant material by recycling nutrients, and therefore indirectly promote further growth of the periphyton communities. Light availability is also a major factor regulating the growth and competitive interactions of aquatic macrophytes, and light attenuation rapidly limits the vertical distribution of plant growth. The photic zone is the area within the water column that receives enough light in the 400 - 700 nm waveband (photosynthetically active radiation (PAR)), to support photosynthesis. Irradiance attenuates exponentially from the surface to deeper water; the depth at which light attenuates to 1% of its surface value defines the lower limit of the photic zone, and is known as the compensation depth. At this depth, production is equal to respiration (Wetzel 1983). If Stave Reservoir were at full pool (82.1 m), the deepest samples for the present study would be 6.1 m below the surface. This is well within the compensation depth of 12 -14 m calculated by Beer (2004) for the summer months between 2000 and 2002; therefore, the development of aquatic communities at all sample elevations was expected. The light data collected during the present study supports this conclusion as the light measured at the maximum depth in the littoral zone during all sampling occasions was greater than 1% of that measured at the surface. Furthermore, each location already contained a wetland plant community at each planting elevation; therefore, it was clear that perennial species were able to survive at that depth of inundation given the reservoir operating regime in effect at the time. 94 4.2 Vegetation 4.2.1 Biomass Kistritz et al. (1983) found that the most active period of shoot growth in native wetlands is generally April, May and June. Stave Reservoir is typically at full pool by the Victoria Day long weekend (i.e. third weekend in May) to address recreation requirements, resulting in the loss of at least one month of peak growth for drawdown zone vegetation in comparison to wetland plants in a non-regulated system. Despite the shorter growing season, woolgrass and lenticulate sedge are two species that are able to survive the stresses of the bi-modal annual drawdown cycle of Stave Reservoir. High variability in plant growth rates is expected from year to year depending on environmental conditions and levels of stress that the plants experience (Bernard et al. 1988; Solander 1983a; Moody 2002). It should be noted that the management regime of Stave Reservoir had been altered each spring in the three years prior to completing this research resulting in longer and deeper than usual drawdowns (BC Hydro 1997; Wilson 1998, 1999). These longer periods of drawdown in Stave Reservoir would likely have benefited the existing shoreline plant community by providing an extended period of exposure leading to enhanced growth and possible expansion of the plant community to lower reservoir elevations (W. Carr, Carr Environmental Consultants, pers. comm..). Fall rye is an annual species that has been used in planting programs in British Columbia reservoirs mainly for sediment stabilization (AIM Ecological Consultants and Carr Environmental Consultants 2000), but also for biomass production to improve biological productivity (Department of Fisheries and Oceans 1999). Fall rye is a suitable species for these purposes as it grows quickly and provides an immediate source of organic material and nutrients upon flooding. In addition, although the foliage decomposes quickly, the roots tend to form a mat, effectively stabilizing exposed shoreline sediments during periods of drawdown. The root mats provide additional benefits in that they add organic matter to the substrate, and may also trap and encourage germination of seeds of native plant species during subsequent seeding programs in the same location (W. Carr, Carr Environmental Consultants, pers. comm.). 95 4.2.1.1 Aboveground biomass Woolgrass tended to have a higher mean pre-inundation biomass in relation to sedge in this study, however, there was a high degree of variability in pre-inundation plant sizes and weights, therefore absolute biomass values were considered to be less meaningful than relative decomposition rates. The two perennial species, sedge and woolgrass, showed a much slower rate of decomposition (losing 55% and 30% of their mean pre-inundation biomass, respectively) in comparison to fall rye, which lost 80% of its mean pre-inundation biomass by the first sampling period. In general, higher water temperatures, aerobic conditions, low fiber (lignin) content, and young tissues tend to increase decomposition rates of submerged vegetation (Ogwang 1979). Emergent macrophytes tend to exhibit very slow rates of degradation in comparison to floating leaved and submersed aquatic plants, largely due to the greater quantities of lignin present in the tissues of the emergent vegetation (Sculthorpe 1967; Wetzel 1983). The relative amount of lignin in the cell walls is likely the main reason for the slower decomposition rate seen for the perennial species in relation to the annual fall rye in this study. Lignin is a complex polymer that is laid down in plant cell walls to bind cellulose fibres and give the plant parts rigidity. It provides structural support to emergent plants and is relatively resistant to decomposition (Sculthorpe 1967; Ogwang 1979; Bergon er al. 1986). Lignin is almost solely broken down by fungi, mainly Bacidiomycetes, which have suitable enzymes to break down the aromatic rings and side chains contained in lignin (Trojanowski 1969). The role of bacteria in the decomposition of lignin, on the other hand, is mostly restricted to break down of side chains or the monomers which are first detached from the lignin by the fungi (Healey er al. 1980). Fall rye in this study behaved similarly to vascular submerged macrophytes, which tend to contain very little supportive tissue (Wetzel 1983), and therefore decompose rapidly. Dierberg (1993) found that only 9-14% of initial dry weight of two submerged macrophytes (Vallisneria and Hydrilla) remained after three weeks of inundation, and that only 24% of biomass for a third submerged macrophyte, Potamogeton, remained after 7 weeks of inundation. AIM Ecological Consultants and Carr Environmental Consultants (2000) also found rapid decomposition rates for fall rye in comparison to perennial wetland species during a similar study completed in Arrow Reservoir in 1999. 96 Several other studies have shown similar results for decomposition of emergent vegetation upon inundation. Taxon Aquatic Monitoring Company (1993) found that slough sedge was the most resistant to decomposition in comparison to willow and reed canarygrass in a litter bag experiment in Blue River Reservoir in Oregon. After 100 days, none of the sedge samples had lost less than 50% of their initial dry weight biomass. CARR Environmental Consultants and AIM Ecological Consultants (2002) found that lenticulate sedge retained approximately one third of its pre-inundation biomass in Arrow Reservoir. Wetzel (1983) referred to a series of studies that looked at decomposition rates of various emergent, floating and submergent plants. Koreliakova (1959) (cited by Wetzel 1983) found that 42% of Carex gracilis dry weight was lost after 6 weeks, and 55% was lost after 9 weeks. Godshalk (1977) found that 35% of Scirpus acutus dry weight was lost over a period of 12 weeks during the spring - summer period. Small increases seen in aboveground biomass for both woolgrass and sedge later in the summer could be attributed to biomass increases associated with the attached epiphytic community. This is consistent with findings in Arrow Reservoir (Carr Environmental Consultants and AIM Ecological Consultants 2002), and also in a coastal Carex lyngbyei marsh (Kistritz er al. 1983). The relative loss of bjomass per hectare between vegetation species is important when considering the overall annual input of organic material and associated carbon and nutrients to the detritus pool, and in turn, the effect on biological productivity. This is especially true in an ultra oligotrophic system such as Stave Reservoir (Stockner and Beer 2004), where benthic food webs play such a key role in littoral productivity (Stockner and Antia 1986). Native wetlands are reported to produce anywhere in the order of 5 -10 tons of aboveground biomass per hectare per annum (Goldman and Home 1983; AIM Ecological Consultants er al. 2000). Although not looked at specifically for this study, it is likely that the amount of organic matter produced in the littoral zone of a reservoir that goes through cyclical drawdowns would be less, due to the higher levels of stress potentially affecting plant growth rates and the truncated growing season experienced by the reservoir shoreline vegetation community in comparison to a native wetland (Kistritz er al. 1983). The rapid decomposition of fall rye aboveground biomass relative to sedge or woolgrass suggests that fall rye would have a greater immediate impact on biological productivity in the littoral area by providing a pulse of nutrients and carbon to support the development of the benthic food web. 97 4.2.1.2 Belowground biomass Changes in belowground biomass are more difficult to study than changes in aboveground biomass due to issues associated with isolating root components from mineral substrates for laboratory analysis; potential losses of organic material during initial harvesting of plants from the reservoir (i.e. removing roots from the adjoining root mats); and in determining live vs. dead components. In this study, no clear trends were noted for woolgrass belowground biomass; however a gradual decrease in belowground biomass over time for sedge was apparent. Fall rye tended to have a much smaller pre-inundation root biomass than either of the perennial species, with much of the root material disappearing by the first sampling period. The differences in root biomass between the three species are likely due in part to the fact that fall rye is an annual and as such, its root system is not intended to support the plant during prolonged stressful environmental conditions such as inundation. The perennial species, on the other hand, possess well developed, extensive systems of roots and rhizomes that sustain the plant during periods of inundation (Wetzel 1983; Kistritz et al. 1983). In addition, the roots of the fall rye tended to be very fine in texture relative to the perennial species, which suggests much higher lignin content for belowground biomass of the perennial plants. It was also very difficult to remove all the fall rye root biomass from the growing medium in the laboratory due to the very fine nature of the roots. This may have introduced some error into the fall rye belowground biomass results. Root to shoot ratios The pre-inundation biomass root to shoot ratios for individual plants was highest for sedge with an average value of approximately 4, followed by woolgrass at approximately 2.5 and finally fall rye at approximately 0.5. The biomass root to shoot ratio for both woolgrass and sedge increased upon inundation, peaking at T1 and T2 respectively. The increases are attributed to the senescence and decomposition of aboveground material in concert with little change in belowground biomass. Subsequent declines in the root to shoot ratios could be attributed to foliar colonization by periphyton (Carr Environmental Consultants and AIM Ecological Consultants 2002; Kistritz er al. 1983). 98 The literature indicates that biomass values vary greatly between species and environmental conditions, however there does appear to be consensus that the extensive system of roots and rhizomes associated with emergent plants generally outweighs the aboveground components, representing anywhere between (30-95%) of the total biomass in emergent aquatic vegetation (Wetzel 1983; Kistritz et al. 1983). Muthukumar er al. (2004) found that many plant species adapt to nutrient poor environments through increased allocation of biomass to roots, whereas Wetzel and van der Valk (1998) found that sedges consistently invest greater biomass resources into root production regardless of soil nutrient levels. Wetzel (1983) also reported that the root to shoot biomass ratios usually increase markedly as the season progresses and aboveground vegetation begins to decompose. 4.2.2 Nutrient cycling 4.2.2.1 Nitrogen The mean foliar pre-inundation N content was highest in fall rye (3.04%) in comparison to lenticulate sedge (1.99%) and woolgrass (1.74%). All three species saw a decrease in mean foliar N between TO and T1, which coincided with an increase in N levels in the root material of sedge and fall rye during the same time period, with a less apparent increase for woolgrass. Fall rye also had the highest mean pre-inundation root N content (1.25%) in comparison to sedge (0.63%) and woolgrass (0.76%). When looking at the mean root to shoot ratio, there was a significant increase upon inundation for fall rye from (0.4 to 1.1), but no significant change post-inundation for any of the species. This suggests that translocation of nitrogen from the foliage to roots occurred fairly quickly upon inundation and then remained relatively constant in the roots and foliage for the remainder of the period of inundation. The translocation of nutrients from live shoots to roots in emergent vegetation has been well documented elsewhere (Wetzel 1983; Kistritz et al. 1983; CARR Environmental Consultants and AIM Ecological Consultants Ltd 2002). Kistritz et al. (1983) found that the maximum period of translocation was May to August. During May and June, nutrients were translocated upward, and during July and August, nutrients were moved from the shoots back to the roots. They estimated that of the belowground nutrients shunted upward, 78% of the nitrogen and 66% of the phosphorus was realized as a gain to aboveground biomass, and the remainder of nutrients was lost via 99 aboveground leaching. A similar assumption was made for leaching rates during the downward translocation of nutrients. When nutrients were translocated downward, the increase seen in the roots was in excess of that lost from the aboveground biomass, indicating that the roots were also taking nutrients up from the soil. The rapid shunting of nitrogen from fall rye foliage to roots observed in the present study was not necessarily expected as fall rye is an annual not adapted to flooding. While there are many papers in the literature that describe reactions to flooding in both flood tolerant and intolerant species (e.g. Crawford and Braendle 1996; Drew 1997; Kozlowski 1997; Pezeshki, 1994; Dias-Filho er al. 2000; Crossley 2002; Ernst 1990; Raven 1984; Mitsch and Gosselink 2000; Robe and Griffiths 1998; Steed er al. 2002; Visser er al. 2000), none were found that specifically addressed the potential for nutrient translocation from shoots to roots in flood intolerant species. Although pure speculation, it is possible that the translocation of nitrogen from foliage to roots in fall rye noted in this study occurred as an immediate response to flooding in an attempt by the plant to stay alive; however, the fall rye plants still were unable to survive due to the significant physiological damage, as described in Section 1.2.2 of this thesis, that occurs concurrently in response to flooding. , AIM Ecological Consultants and CARR Environmental Consultants (2000) attributed increasing nitrogen root to shoot ratios over time for perennial species in Arrow Reservoir to a possible increase in periphyton colonization. Further studies completed at Arrow Reservoir which focused strictly on the wetland plant community confirmed that lenticulate sedge was a tight internal nutrient cycler of N in that environment, exhibiting a decrease in live shoot N with an increase in root N upon inundation (CARR Environmental Consultants and AIM Ecological Consultants Ltd 2002). 4.2.2.2 Phosphorus Pre-inundation foliar P values for the present study were significantly higher in fall rye (0.63%) than woolgrass (0.16%) or sedge (0.17%). Upon flooding, the foliar P value for fall rye fell significantly (by approximately 50%) and then remained relatively constant for the rest of the study. No significant changes occurred in foliar P content for the perennial species pre and post-inundation. Similar patterns were noted for root P content, with fall rye having a significantly higher initial P content (0.25%) than either woolgrass (0.08%) or sedge (0.06%). 100 The mean phosphorus root to shoot (R:S) ratio was significantly different between species, with fall rye having the highest ratio followed by sedge and then woolgrass. Also the phosphorus R:S ratio was significantly different between TO and T1, mainly due to the much higher pre-inundation P levels found in both the roots and shoots of fall rye. Although P content dropped drastically in the fall rye foliage after one month of inundation, no increase was noted in the root content at T1, indicating that phosphorus was not translocated from the fall rye foliage to the roots after inundation, and instead was lost to the water column. The results for fall rye were not unexpected as several studies have shown that up to 50% of the foliar phosphorus may be lost rapidly from submerged plants upon inundation (Puriveth 1980; Dierberg 1993). In addition, fall rye, being an annual, likely does not have the capability to store nutrients in its roots to enhance survival during stressful environmental conditions. The lack of detectable phosphorus translocation in the perennial species for this study was somewhat surprising however, as Kistritz et al. (1983) observed downward phosphorus translocation rates of 12.2 mg phosphorus per m 2 per day in a tidal Carex lyngbyei marsh. AIM Ecological Consultants and CARR Environmental Consultants (2000) reported a similar lack of translocation for phosphorus during a submergence study completed in Arrow Reservoir in 1999, however in a subsequent study, CARR Environmental Consultants and AIM Ecological Consultants (2002) found that phosphorus content did not decrease in live shoots, but did decrease in dead shoots in concert with an increase in P levels in the roots for three wetland plant species they studied. Kistritz er al. (1983) noted that there tends to be great variation in nutrient content between the types of plant biomass (i.e. seed heads vs. live vs. dead biomass), and Dierberg (1993) found that phosphorus release from healthy shoots is of relatively minor importance to overall phosphorus release. In the present study, aboveground components were treated as one biomass type, which could have masked any significant effects present in specific plant components of phosphorus translocation. The rapid release of phosphorus from the fall rye observed in this study is of interest in that phosphorus tends to be the limiting nutrient in oligotrophic lacustrine systems (Wetzel 1983). Dierberg (1993) found that 34 to 50% of the initial P concentration rapidly (i.e. approximately 3.5 hours after submergence) leached from decaying submerged macrophytes was soluble reactive phosphorus and was therefore bioavailable. Depending on bioavailability, the 101 phosphorus released from fall rye could be important for development of the benthic food web in the littoral zone. 4.3 Periphyton The periphyton assemblages associated with all vegetation species at all elevations over time were dominated by diatoms (Bacillariophyta) in terms of total abundance and biovolume. Twenty-nine diatoms, 18 chlorophytes (green alga), and 4 cyanophytes (blue-green alga) were identified amongst all samples, for a total of 51 taxa. The five most dominant diatom taxa overall by density in the present study included Frustrulia sp. (22%), Achnanthes minutissima (9%), Frustrulia rhomboides (8%), Tabellaria flocculosa (8%) and T. fenestrata (7%). Several species of the genus Eunotia were also present in smaller proportions. The dominant species associated with vegetation were quite different than those associated with artificial plates reported by Beer (2004). This could be due to differences in the abilities of periphyton species to selectively colonize living substrata over inert substrata in order to capitalize on the associated nutrients, or that plants may be preferentially colonized by certain diatom species due to the higher habitat complexity associated with vegetation in comparison to an artificial plate (Wetzel 1983). In addition, the vegetation samples for the present study were placed in areas with an existing plant community. It is not known if the artificial plates were situated near to vegetated areas that could provide a colonizing diatom source. Alternatively, the observed differences in the community compositions between the live vegetation and artificial plates could be related to differences in sampling procedures or accuracy of species identification. 4.3.1 Density Results of diatom density for woolgrass and sedge from the present study were quite variable. Unfortunately, quantitative analyses for fall rye could not be completed as the state of foliar decay was too advanced to accurately sample the attached algal community. Qualitative observations, however, indicated that the periphyton assemblages were dominated by diatoms during all sample dates, therefore further analyses focused on the diatom community. Beer (2004) also found that diatoms dominated the periphyton species assemblage on artificial substrata in Stave Reservoir, accounting for 72% of the total algal population by abundance. 102 The mean diatom density associated with sedge (28,450 cells/cm2) was higher than that noted for woolgrass (18,500 cells/cm2) at T1; however the difference was not statistically significant. The number of diatoms per cm 2 of plant material did increase significantly between T1 and T2 for both vegetation species. Increases in diatom density noted between T2 and T3 were not significant for either plant species, indicating that the diatom community was fully developed by T2. This is supported by the results of the NMS ordination which indicated that there was a different grouping for samples at T1 in comparison to T2 and T3. The maximum mean diatom densities for woolgrass and sedge were approximately 75,000 and 112,000 diatoms per cm 2 of vegetation, respectively, by the final sample period. The lack of a significant difference between sedge and woolgrass in terms of diatom density over time is not unexpected due to the similarities between woolgrass and sedge in terms of nutrient content, decomposition rates, and reactions to flooding. The sample dates for T1 (i.e. 30 days post-inundation) and T2 (i.e. 63 days post-inundation) appeared to bracket the period of maximum diatom community growth. This is consistent with the findings of Stockner and Armstrong (1971) regarding the development of the diatom community in the littoral zone of a temperate lake in the Experimental Lakes Area in Ontario. They found that the initial period of diatom colonization lasted approximately 29 days, and was characterized by a low growth rate (27 mg organic matter/m2.day). Maximum growth occurred between days 30 to 57 (250 mg organic matter/m2.day), and then reached an asymptote as factors such as crowding, competition for nutrients and light (i.e. shading of the understory by the overlying organisms (Wetzel 1983)), and grazing (Kessler 1981) began to limit further expansion of the diatom community (Figure 46). 103 350000 200 Figure 46 Changes in Total Diatom Density on glass slides at 1 m in Lake 240 in the Experimental Lakes Area, Northwestern Ontario (after Stockner and Armstrong 1971) The slightly lower, although not significantly different, diatom density observed for the 80 m samples at T3 may have been due in part to the earlier retrieval date, or that the 80 m samples at T3 were fully exposed at the time of retrieval. However, given that the diatom community appeared to be fully developed by T2, the lower diatom density for the 80 m samples at T3 could also have been due to photoinhibition, which often occurs in the surface waters of lakes due to a destruction of enzymes under conditions of high light intensity and ultra violet radiation (Wetzel 1983). This results in lower photosynthesis at the surface, and a subsurface photosynthetic maximum. It should be noted, however, that algal response to light intensity is species specific, and many diatom species are able to adapt to changes in light intensity by adjusting their light-saturated photosynthetic rate (Wetzel 1983), and/or by production of UV-absorbing or blocking pigments known as fucoxanthins (Callieri and Stockner 2000). Densities of diatoms in littoral zones of ultra oligotrophic coastal lakes tend to be in the range of 30,000 to 40,000 cells/cm 2, and in oligotrophic lakes they can be anywhere from 78,000 to 104 >1,000,000 cells/cm 2 (Stockner and Shortreed 1976, Shortreed and Stockner 1978). This would suggest that Stave Reservoir is on the border between being oligotrophic and ultra-oligotrophic. Water chemistry samples collected by Beer (2004) and Stockner and Beer (2004) confirm that the reservoir is ultra-oligotrophic. A similar submergence study completed in Arrow Reservoir yielded very low periphyton results in comparison to other oligotrophic systems, with many samples having less than 18,000 algal cells/cm 2 (the majority of samples contained less than 6,000 cells/cm2) (Perrin et al. 2002). Low densities in that study were attributed to large amounts of silt, or possibly due to grazing pressure from benthic invertebrates. 4.3.2 Taxonomic richness Taxonomic richness is a count of the number of species in a given area, and is the simplest measurement of biodiversity. The biodiversity of a community tends to increase with the complexity of the food web, and with the extent of niche overlap, especially in the lower trophic levels of the food chain (Wetzel 1983). According to McCann (2000), if a food web in an ecosystem is more diverse at the lower levels, and has a lot of weak interactions between different species, the community as a whole should be more stable, as the species at the lower level will better be able to resist consumption. Biodiversity can also be an indicator of health of an ecosystem. No significant differences were noted in this study between sedge and woolgrass in the taxonomic richness of the associated diatom communities over time. This is likely because sedge and woolgrass are both perennial species which respond similarly to flooding (i.e. similar decomposition rates and nutrient release patterns). In addition, the foliar samples were taken from live vegetation, rather than decaying plant material for both species. Although it is pure speculation, it is likely that the taxonomic richness of the diatom community that colonized the fall rye would be quite different than that associated with the perennial species, mostly due to the pulse of dissolved organic carbon and phosphorus provided to the bacteria and cyanobacteria from the rapid decomposition of the fall rye foliage, which in turn, provides a food source for the developing periphyton community. 105 The decline in diatom taxonomic richness noted for the 80 m elevation samples at T3, was likely due to a series of factors including potential photoinhibition, exposure to elements (i.e. site C samples were dewatered at time of retrieval) or early retrieval time. 4.3.3 Changes in community composition NMS overlay plots for the top five diatom taxa by density were examined to look for possible changes over time, elevation, or plant species among the dominant diatom species. Note that interpretation of these diagrams was subjective, and no further testing using A N O V A was undertaken to confirm if the noted differences were statistically significant The colonization patterns for Frustrulia rhomboides and Frustrulia sp. suggested that both species increased in density over time, and were more densely distributed at the lower two elevations. In addition, both Frustrulia taxa appeared to be associated slightly more with sedge than woolgrass. Achnanthes minutissima on the other hand, appeared to show no clear preference by depth or plant species, but seemed to increase between 12 and T3. The overlay plots for Tabellaria fenestrata and T. flocculosa suggested that the two species colonized the plant samples differently, with the latter species increasing steadily over time; and the former increasing between T1 and 12, with no apparent differences between 12 and T3. Neither Tabellaria species appeared to have a preference for plant type; however, both species seemed to be concentrated more at the lower two elevations. There are several possible explanations for the different colonization patterns noted for the dominant diatom species in this study related to biological (grazing or competition), chemical (nutrients), or physical (limited light) factors. The fact that the majority of the dominant species appeared to be more concentrated at the lower two elevations could be due to possible photoinhibition of the cells at the 80 m elevation (Wetzel 1983), or because the 80 m samples at Site C were dewatered at time of T3 retrieval. In addition, the 80 m samples were retrieved approximately 2 weeks earlier than the samples at 78 and 76 m. The early retrieval time for the 80 m T3 samples was, however, somewhat accounted for in the NMS ordination, as the environmental variable of degree*days (i.e. the cumulative temperature the plants at each elevation were exposed to over time) was included in the analysis. Another possible explanation for the noted decrease in many dominant diatom species in terms of abundance at the 80 m elevation may be due to more intensive grazing activity by benthic 106 organisms on the vegetation samples installed within this band. Grazing by rotifers and microcrustaceans can greatly influence algal assemblages through selective grazing, mainly due to algal size (Jones et al. 1997, Wetzel 1983). Periphyton communities that persist under intense grazing pressure typically have taxa with high growth or recruitment rates, are buffered from overexploitation due to their growth form (i.e. small, flat or adnate which makes them less accessible to grazers), or are unpalatable or difficult to handle for the grazers (Hann 1991). Less grazing within the 80 m elevation band could also be related to less activity in the presence of predators or alternative food sources being available (i.e. periphyton assemblages associated with plants within the 78 or 76 m elevation bands) (Jones et al. 1997). The greater density of Achnanthes minutissima in the 80 m elevation band could possibly be due to less susceptibility to predation due to the small size of individual cells relative to other diatom taxa. Another possible explanation for the apparent increase in A. minutissima density between T2 and T3, and also in the dominance of A. minutissima associated with plant samples from the 80 m elevation band may be that algae have definite temperature optima, which in concert with light, helps to drive seasonal succession of the periphyton assemblages (Wetzel 1983). The result is that smaller algae with faster turnover rates often dominate algal assemblages at warmer temperatures (Wetzel 1983). 4.3.4 Biovolume Analyses based solely on number of organisms are biased in comparison to studies that look at biomass (or in this case, biovolume) as there can be a great deal of variation in size between species (Wetzel 1983); the overall importance to biological productivity in a specific ecosystem may be either over or underestimated depending on the sizes of the organisms involved. To address this, biovolume for periphyton species was analyzed as a part of this study. Biovolume estimates for diatoms associated with sedge and woolgrass were highly variable, and, similar to diatom density estimates, no statistically significant differences were noted for diatom biovolume estimates between woolgrass and sedge. There was however a significant interaction evident between time and elevation indicating that the diatom biovolume changed differently over time for the plants from the three elevation bands. This is different than the results for diatom density, which indicated that there was a time effect, but no significant difference by elevation. The interaction effect for biovolume is likely due to the relative 107 proportion of small to larger diatoms present over time (for example, Achnanthes minutissima, although one of the most dominant diatoms in terms of density, especially at T2 and T3, did not dominant in volume due to its small size). There was a different group of dominant diatoms by biovolume in comparison to density. Frustrulia sp. (26%) was the dominant diatom taxa by biovolume due to the fact that it was a medium to large-sized diatom and was so dominant in density species composition. Frustrulia rhomboides (15%), Eunotia sp. (15%), Tabellaria fenestrata (14%), and Gomphonema sp. (6%) comprised the remaining top five diatom taxa by biovolume. Eunotia sp, and Gomphonema sp., although accounting for only 4% and 5% of total diatoms by abundance, respectively, dominated biovolume species composition due to their relatively large sizes. The NMS ordination diagrams for overall diatom biovolume supported the ANOVA results indicating that there was no significant difference between sedge and woolgrass in diatom biovolume. The ordination diagrams did suggest that there was a biovolume species composition grouping by time, with the T1 samples grouping differently than the T2 and T3 samples; however no noticeable patterns were observed for elevation. Similar to the diatom density analysis, the difference in biovolume species composition groupings by time can be attributed to initial rapid population growth that tends to slow down over time as biological, chemical and physical limiting factors such as competition, crowding, predation, and grazing begin to affect the community (Stockner and Armstrong 1971). Subjective interpretation of the NMS ordination diagram overlay plots for the four dominant species in terms of biovolume indicate that the patterns for the two Frustrulia taxa were similar to those noted for density (i.e. both species appeared to increase in biovolume over time, were more dominant at the lower two elevations, and had a slightly higher biovolume associated with sedge than woolgrass). The patterns for Eunotia sp. and Tabellaria fenestrata biovolume by depth were similar to those noted for the Frustrulia taxa showing an apparent greater biovolume at the lower two elevations. There were, however, no notable differences in biovolume associated with either perennial vegetation species. There was a notable change for T. fenestrata biovolume between T1 and T2 samples. Eunotia sp. appeared to become more dominant in terms of biovolume over time. No further testing using ANOVA was undertaken to confirm if these noted differences were statistically significant. 108 Refer to Section 4.3.3 for further discussion of possible reasons for changes in the diatom community structure over time and elevation. 4.4 Benthic Community 4.4.1 Abundance The benthic community associated with vegetation in Stave Reservoir was mainly dominated by oligochaete worms (37%) and chironomids (26%), accounting for 63% of all benthic organisms. Researchers in other reservoir systems found even greater dominance of the combined chironomids and oligochaetes, accounting for 89% of the total fauna, by number, and 72% of the total benthic biomass (Kaster and Jacobi (1978)). Other authors found that chironomids alone made up anywhere from 95 to 99% of the benthic animal assemblage associated with plant material in reservoirs (Taxon Aquatic Monitoring Company 1993; Sinclair 1965). Northcote and Atagi (1997), in their literature review of the importance and role of submerged terrestrial vegetation in the flooded littoral zone of reservoirs, found that the invertebrate species assemblages associated with herbaceous plants were dominated mainly by chironomids, with small numbers of other insect larvae, oligochaete worms and crustaceans. Similar results were reported by Kairesalo (1984). The present study showed that vegetation (including all three study species) supported almost 3 times the number of benthic organisms than barren substrate. This is likely due to a number of factors including increased substrate area, abundant food sources associated with the epiphytic alga and detritus (Diehl and Kornijow 1998; Levings 1997), and more complex habitat which can result in lower risk of predation (Hargeby 1990). Taxon Aquatic Monitoring Company (1993) found a low density and standing crop of benthic animals associated with bare sand/sediment in comparison to vegetation plugs in the Blue River Reservoir in Oregon. They also found that the individuals associated with the bare sediment tended to be smaller, which they attributed to a lack of plant derived nutrients and lack of suitable hiding cover for invertebrates. There was a significant difference between the 76 and 78 m samples and the 80 m samples at T3 in the present study/with the 78 m samples supporting approximately 1.6 times the number of organisms than 80 m samples. This result is likely due to a number of reasons including that 109 the 80 m samples at T3 were removed earlier than the 78 and 76 m samples, or that the invertebrates were preferentially attracted to the periphyton food sources available at the lower two sample elevations. Taxon Aquatic Monitoring Company (1993) attributed finding a lower abundance of invertebrates in shallow sampling sites in comparison to deeper sites to a selective avoidance of animals of the warmer and lower oxygen content water. There was no difference between the total numbers of benthic individuals associated with each of the three vegetation species, nor was there a difference when the number of benthic organisms was standardized to plant biomass. The number of organisms associated with vegetation ranged from 163 to 415 individuals at T1, and from 467 to 673 individuals at 12. Fall rye had the highest number of associated individuals at T3 (648 individuals). These results suggest that in general, benthic organisms in this study were attracted to the food sources associated with the plants as opposed to the surface area offered by the plants themselves, as the available area for colonization associated with fall rye decreased dramatically over time in comparison to woolgrass and sedge with no significant difference in the number of associated benthic organisms between plant species. On an areal basis (i.e. standardized to the exposed sediment surface area within the sample pot), the density of benthic organisms ranged from 5,200 - 13,250 animals/m 2 at T1 and 15,800 to 21,000 animals/m 2 at T2. The maximum density seen at T3 (20,600 animals/m2) was associated with fall rye. Perrin er al. (2002) found benthos densities in Arrow Reservoir of up to 43,272 animals/m 2 for aboveground biomass, and up to 64,000 animals/m 2 for belowground biomass. These levels are much higher than those observed in Stave Reservoir, and are also high in comparison to other oligotrophic systems (Perrin et al. 2002). The total benthic organism abundance increased significantly between T1 and T2; however no significant changes were seen between T2 and T3. This is a similar pattern to that observed for diatom abundance associated with woolgrass and sedge, and could be indicative of a relationship between the benthic and diatom communities associated with the vegetation. Of particular interest is the relatively high density of diatoms associated with sedge at T2, which appears to be mirrored closely by the high number of benthic organisms associated with sedge at T2. In both cases, however, the samples were highly variable. Elevation did not have a significant effect on benthos abundance when looking at total number of individuals. The NMS ordination results were consistent with ANOVA in describing changes in benthos community 110 structure over time (i.e. samples from T1 tended to cluster weakly together compared to those from T2 and T3). This also is similar to the ordination results for diatom density and biovolume. Although no significant difference was noted between the total number of benthic individuals or the standardized number of individuals per gram of plant material, when examining differences in the benthic community composition over time using NMS, there was a definite grouping of the two perennial species together, with the fall rye and control samples each forming their own distinct groups. This indicates that there was a difference in the benthic community composition associated with the perennial species in comparison to the annual fall rye and barren substrate. These differences could be due in part to the lack of structural material associated with fall rye to provide habitat for the benthic organisms, or the differences could also be related to the epiphytic food and nutrient sources associated with decaying fall rye in comparison to the perennial species. Unfortunately, no quantitative diatom data was available for fall rye due to the advanced state of decay of the foliar material so no firm conclusions can be drawn about the relative colonization rates of benthos and diatoms associated with fall rye. Although, not looked at specifically during this study, it is expected that the fall rye would have a large number of bacteria associated with the decomposing biomass, which would form the base of a rich benthic food web (Stockner and Antia 1986). Subjective interpretation of the NMS ordination diagram and overlay plots for dominant benthic species in terms of overall abundance, suggest that there was a benthic taxa preference for vegetation type. Representatives from the subclass Ostracoda and the chironomid Phaenopsectra appeared to prefer decaying fall rye to the erect living perennial vegetation. In comparison, representatives from the family Enchytraeidae (Class Oligochaeta) and representatives from the Phylum Nematoda appeared to prefer the perennial plant species over fall rye. Very little difference in colonization patterns of nematodes between fall rye and barren substrate was noted. No further testing using ANOVA was undertaken to confirm if these noted differences were statistically significant. Ostracoda appeared to be most abundant at the 76 m elevation and at T3. Phaenopsectra, the most dominant chironomid taxon, seemed to be more abundant at T3, with no strong patterns for elevation noted..Enchytraeidae appeared to be marginally more abundant at T1 and T3 in comparison to T2, and no strong patterns were noted related to elevation. Nematodes 111 appeared to be most abundantly associated with the 80 m elevation samples early on in the experiment (T1). Again, these noted differences were not tested for statistical significance. Taxon Aquatic Monitoring Company (1993) also found that plant species had an effect on the composition of animals associated with them in the Blue River Reservoir in Oregon. For example, sedge was more attractive to the midge Tanytarsini, whereas the detritus feeder, Phaenopsectra sp. was associated more with willow. Columbia sedge tended to attract more cladocera and copepods. In Arrow Reservoir, benthos tended to favor dead and decaying plant material over submersed living plants. The preference was attributed to the food sources offered by the decaying plant material in the form of periphyton and detritus combined with the nutrients released from the plant material itself, whereas the perennial species provided mainly the epiphytic food source and a substrate for colonization (Perrin er al. 2002). 4.4.1.1 Aboveground vs. Belowground Vegetation There was a significant difference in number of benthic organisms associated with aboveground in comparison to belowground vegetation, with fall rye components supporting approximately the same number of organisms for each plant component. Sedge and woolgrass, however, both supported more benthic organisms per gram of aboveground biomass in comparison to belowground biomass (1.7 times for sedge and 1.4 times for woolgrass). Sedge foliar samples supported 1.6 times the number of benthic organisms than woolgrass foliage. This is in contrast to the findings in Arrow Reservoir, where the belowground benthic organisms (64,000/m2) were found to outnumber the aboveground organisms (44,000/m2) by almost 1.5 times (Perrin et al. 2002). 4.4.2 Taxonomic Richness Invertebrate species richness has been documented to be higher in areas with aquatic macrophytes in comparison to barren areas (Gregg and Rose 1985, Hargeby 1990). Vegetation samples in this study supported approximately 1.7 times the number of benthic taxa than barren sediment. In Arrow Reservoir, the number of taxa associated with vegetation in comparison to non-vegetated areas increased by 2 to 4 times (Perrin et al. 2002). Resh and Grodhaus, (1983) indicate that taxonomic richness in general decreases with decreasing water or habitat quality, which is consistent with the suggestion that the presence of vegetation generally provides 112 higher quality habitat for associated benthic organisms. A drop in number of benthos taxa noted at T3 for the 80 m samples is possibly due to the early retrieval time, or the fact that Site C samples were dewatered at time of retrieval. 4.4.3 Size of animals An interesting observation for the benthic animals observed during this study is that microbenthos (i.e. <1 mm in size) accounted for 87 to 89% of all benthic organisms. This is consistent with results from the submergence study in Arrow Reservoir, where microbenthos accounted for more than 80% of animal densities, and in many cases were more than 95% of those densities (Perrin et al. 2002). Jones er al. (1997) noted that not only do epiphytic invertebrates tend to be more abundant than benthic invertebrates; but they also tend to be smaller. As a result, the relative biomass of epiphytic invertebrates may not be that much different than that of benthic invertebrates. The size of benthic individuals is an important factor when considering availability as a food supply for visual predators (i.e. fish), as smaller animals are less susceptible to predation and are therefore less likely act as a link to higher trophic levels.(Perrin er al. 2002). Also important is the life history of the taxa found within the benthic assemblage. For example, chironomids become highly susceptible to predation by fish upon emergence, whereas nematodes remain in the benthos throughout their life history and are therefore much less available as a food source for fish (C. Perrin, Limnotek Research and Development, pers. comm.). 4.4.4 The relationship between diatoms and benthos Invertebrate grazers can play an important role in exerting top down control of periphyton biomass in lakes, and inversely, herbivore populations appear to be controlled simultaneously from the bottom up by algal availability (Jones et al. 1997). Fairchild (1981) found that numbers of cladocerans and chironomids were highly correlated with diatom numbers, and Fairchild and Lowe (1984) found that chironomid density increased with an increase in algal biomass. Cattaneo and Kalff (1986), and Cattaneo (1990) suggested that a summer minimum in periphyton biomass is a result of grazers. In the present study, no significant differences were noted in the ratio of diatom biovolume to density of benthic organisms by elevation or plant species (note that the analysis was only 113 completed for woolgrass and sedge as quantitative diatom data was not available for fall rye). A significant difference was noted over time, however, with the geometric mean at T3 being just over double that observed at T1. This is likely due to the fact that the diatom community had not yet entered the maximum period of development by T1, but was fully developed by 12 (Stockner and Armstrong 1971). The relatively constant diatom biovolume and benthic invertebrate abundance noted between 12 and T3 may represent a balance between diatom replenishment and losses due to grazing, sloughing, or other mechanical means such as wave action. 5.0 Conclusions The perennial wetland species, woolgrass and sedge, decomposed much slower than the annual agronomic species, fall rye. Fall rye lost approximately 80% of its pre-inundation foliar biomass by the first sample period in comparison to 55% and 30% for sedge and woolgrass, respectively, by the end of the experiment. The difference in loss of biomass is important when considering overall input of biomass and associated nutrients into the waterbody, and may also play an important role in the species composition of associated periphyton and benthic communities. For example, although there was no statistically significant difference in the overall abundance of benthic organisms between the three plant species in this study, NMS analysis suggested that there were differences in the benthic community composition associated with the perennial species in comparison to the annual fall rye, and with the control samples. NMS analysis suggested that the chironomid Phaenopsectra and representatives from the subclass Ostracoda preferred decaying fall rye to the erect living perennial vegetation. In comparison, representatives from the family Enchytraeidae (Class Oligochaeta) and Phylum Nematoda appeared to show a preference for the perennial species over fall rye. The apparent preferences for individual taxa by species were not tested statistically. Fall rye contained the highest pre-inundation %N and %P content of all three plant species in both the shoots and the roots. There was some evidence of nitrogen translocation within all three plant species, with an apparent shunting of N to the roots upon inundation. Nitrogen levels remained fairly constant in subsequent sample periods. There were no noticeable patterns for phosphorus for the perennial species; however, the foliar P level for fall rye declined rapidly upon inundation with no concurrent rise in root P levels. This indicates that foliar P was released from the fall rye during decomposition, much of which was likely bioavailable for the 114 bacteria and cyanobacteria that are the first colonizers in the periphyton assemblage. These organisms begin to photosynthesize and secrete a polysaccharide matrix, or biofilm, which forms the base of the benthic food web. Phosphorus tends to be the limiting nutrient in oligotrophic lacustrine ecosystems, therefore this release of phosphorus from fall rye could provide an immediate and important boost to biological productivity in the littoral zone. The periphyton assemblage, dominated by diatoms, developed on the foliage of the vegetation upon inundation. Diatom density increased dramatically between T1 and T2, and then remained fairly constant between T2 and T3. The community structure appeared to change over time, with many of the dominant species preferring the lower two elevations over the 80 m plant samples. There were many similarities between woolgrass and lenticulate sedge in terms of the quantity and biovolume of epiphytic diatoms associated with each species. Frustrulia sp. and F. rhomboides appeared to show a slight preference for sedge over woolgrass; there were no apparent vegetation preferences for the other dominant diatom taxa. The total number of diatoms was indicative of a system on the borderline between oligotrophy and ultra-oligotrophy. Stockner and Beer (2004) reported that in terms of carbon and nutrient production, Stave Reservoir is ultra-oligotrophic. The presence of vegetation increased the total number of benthic organisms by approximately 3 times, and the total number of taxa by approximately 1.7 times in comparison to barren substrate, indicating that vegetation plays an important role related to development of the benthic community in the littoral ecosystem. Oligochaete worms and chironomids accounted for 63% of all benthic organisms observed in this study. Planting elevation appeared to have a less important effect upon the metrics of diatom density and biovolume and benthos abundance in comparison to plant species and time. The only situations when the elevation effect was significant, was when it interacted with the main effect of time. When they occurred, the interactions could have been due to photoinhibition or higher grazing rates within,the 80 m elevation band, but were most likely due to the early retrieval of the 80 m samples at T3. The ratio of diatom biovolume to benthos abundance increased significantly between T1 and T2, indicating that the diatom biovolume increased more quickly than the benthos abundance during the period of rapid diatom community growth. The relatively constant diatom biovolume 115 to benthos abundance ratio noted between T2 and T3 may represent a balance between diatom replenishment and loss due to grazing, sloughing, or other mechanical means such as wave action. 5.1 Summary and recommendations for reservoir management This research demonstrated that the presence of shoreline vegetation increased biological productivity in comparison to unvegetated areas in the Stave Reservoir drawdown zone during the period of flooding. It was clear that upon inundation, the vegetation supported rapid development of the periphytic and benthic communities that comprise the benthic food web. In contrast to the native perennial vegetation, fall rye decomposed very rapidly, releasing a large proportion of its total phosphorus reserve from both the foliage and roots shortly after inundation. This is significant in an oligotrophic system such as Stave Reservoir, as phosphorus tends to be the limiting nutrient for biological productivity. This pulse of biomass and phosphorus provides an immediate boost to biological productivity in the drawdown zone. Although the presence of native vegetation was shown to be important for biological productivity within the drawdown zone, the importance to overall reservoir biological productivity is questionable for several reasons: • The native vegetation lost relatively little biomass over the period of flooding, resulting in little organic input into the surrounding water column; • The native plants translocated nitrogen from the foliage to the roots and phosphorus levels in both plant components did not change over time, indicating that very little of the nutrient reserves were released from the vegetation to the water column; • Any organic material and nutrients that were released through leaching and decomposition were likely quickly taken up by the periphyton and benthic assemblage and tightly cycled within the microbial loop; • The spring growing season in the Stave Reservoir drawdown zone is fairly short due to the operating regime that generally sees water levels approaching full pool by the Victoria Day long weekend in May. This could result in less biomass production per plant than would be seen in a more favorable growing environment; and 116 • The morphology of the Stave Reservoir shoreline is such that there are few areas with suitable gradient where a vegetation community can become established. In addition, the depth of inundation can have a dramatic effect on plant survival; therefore development of the vegetation community is limited to the upper portion of the drawdown zone. Fall rye planting programs have been used effectively to control dust in the drawdown zones of other reservoirs in British Columbia (e.g. Arrow, Williston, and Carpenter Reservoirs). The present research shows that fall rye also benefits biological productivity in the immediate area upon flooding through release of a pulse of phosphorus and organic matter, and is a good interim mitigation measure to address potential losses in biological productivity in the littoral zone due to reservoir operations. Similar to the native vegetation, it is expected that the phosphorus would be taken up rapidly and intensively cycled within the periphyton and benthic communities, and would therefore provide more benefit to biological productivity in the immediate drawdown zone rather than to the overall reservoir. Perhaps equally or more important than the release of phosphorus from the foliage, fall rye has been shown in other reservoirs to encourage expansion of the native plant community by stabilizing substrates, trapping native seeds in the remaining fall rye belowground biomass, accruing carbon in the sediment, and depending on the seeding method, driving native vegetation seeds that may be sitting on top of the substrate into the sediment thereby increasing the chance of germination (W. Carr, Carr Environmental Consultants, pers. comm.). This expansion of the native vegetation community is a more sustainable long term approach to increasing biological productivity in the drawdown zone. A second possible mitigation measure to address potential loss of littoral zone productivity due to reservoir operations would be to apply fertilizer directly to the native plants in the drawdown zone. This could result in increased growth rates for both above and belowground biomass, and incorporation of increased amounts of carbon and nutrients in the substrate. This is in turn could potentially encourage expansion of the native plant community. The objectives of any mitigation program designed to address losses in biological productivity due to reservoir operations would need to be clearly defined. If the goal is to increase biological productivity mainly ; in the drawdown zone, a fall rye planting program or native wetland 117 fertization program would be appropriate. However, if the objective is to target overall reservoir biological productivity, other mitigation measures, such as reservoir stabilization or fertilization, should be explored. Phosphorus dynamics associated with the below-ground biomass component of fall rye and native vegetation should be examined more closely to determine the fate of the nutrients and impact on biological productivity. As well bioavailability of phosphorus released from fall rye foliage should be assessed further. More quantitative studies on the diatom community associated with fall rye and also on the microorganisms involved in decomposition of macrophytes, such as bacteria, fungi and the rest of the heterotrophic community, should be undertaken to better understand their roles in nutrient cycling and biological productivity. 118 6.0 References AIM Ecological Consultants Inc. and Carr Environmental Consultants. 2000. Summary report on vegetation and soil analyses for the 1999 pilot study Revelstoke Reach - Upper Arrow Reservoir. Report prepared for BC Hydro. AIM Ecological Consultants Ltd., Eco-logic Ltd, and CARR Environmental Consultants. 2000. Vegetation benefits to fish: a literature review. Report prepared for BC Hydro. Allan, J.D. 1995. Stream Ecology: Structure and function in running waters. Chapman and Hall. Allen, J.A., S.R Pezeshki, and J.L. Chambers. 1995. Interaction of flooding and salinity stress on bald cypress (Taxodium distichum). Tree Physiology\Q: 307-3 13. Ashley, K.I., and P.A. Slaney. 1997. Accelerating recovery of stream, river, and pond productivity by low-level nutrient replacement. In: Fish Habitat Rehabilitation Procedures. Watershed Restoration Program, British Columbia Ministry of Environment, Lands and Parks, Watershed Restoration Technical Circular No. 9. Ashley, K.I., Thompson, L.C., Lasenby, D.C., McEachern, L., Smokorowski, K.E., and D. Sebastian. 1997. Restoration of an interior lake ecosystem: The Kootenay Lake Experiment. Water Quality Research Journal of Canada (32):295-323. Barlocher, F. 1980, Leaf-eating invertebrates as competitors of aquatic Hyphomycetes. Oecologia 47: 303-306. Baxter, G.M. 1977. Environmental effects of dams and impoundments. Ann Rev. Syst. Ecol. 8:255-283. BC Hydro. 1997. Stave Falls Powerplant Replacement Project Stave Lake Reservoir Drawdown, March/April, 1997. Environmental Management Plan. BC Hydro. 1999. Stave River Water Use Plan. Draft September 1999. BC Hydro. 2000. Making the Connection. The BC Hydro Electric System and How it Operates. 2nd revision. BC Hydro. 2003. Stave River Water Use Plan (Stave Falls and Ruskin Projects). Revised for Acceptance by the Comptroller of Water Rights. Beer, J.A. 2004. Littoral Zone Primary Production in a Coastal Reservoir Ecosystem. M.Sc. Thesis. University of British Columbia. Bergey, E.A. 1995. Local effects of a sedentary grazer on stream algae. Freshwat. Biol. 33:401-409. Bergon, M., J.L, Harper, and C.R. Townsend. 1986. Ecology: individuals, populations, and communities. Sinauer Associates. Inc. Publishers. Sunderland, Mass. 119 Bernard, J.M and G. Hankinson. 1979. Seasonal changes in standing crop, primary production, and nutrient levels in a Carex rostrata wetland. Oikos, 32:328-336. Bernard, J.M., D. Solander, and J. Kvet. 1988. Production and nutrient dynamics in Carex wetlands. Aquatic Botany, 30:125-147. Biggs, B.J.F. and R.A. Smith. 2002. Taxonomic richness of stream benthic algae: Effects of flood disturbance and nutrients Limnol. Oceanogr, 47(4): 1175-1186 Bldm, C.W.P.M. and L.A.C.J. Voesenek. 1996. Flooding: The survival strategies in plants. Trends in Ecology and Evolution 11:290-295. Bloomqvist, P. and P. Olsen. 1981. Vaxtplanktonkompendium, Inst. Limnologie, Uppsala Universitet, Uppsala, Sweden, 186p. Bowker, D.W., M.T. Wareham and M.A. Learner, 1983. The selection and ingestion of epilithic algae by Nais elinguis (oligochaeta: Naididae). Hydrobiologia 98:171-178, Boyd, C .E . 1970. Losses of mineral nutrients during decomposition of Typha latifolia. Arch. Hydrobiol. 66:511-517. Brinson, M.M, A . E . Lugo and S. Brown. 1981. Primary productivity, decomposition and consumer activity in freshwater wetlands. Ann. Res. Ecol. Syst., 12:123-161. Callieri, C and J .G . Stockner. 2000. Picocyanobacteria success in oligotrophic lakes: fact or fiction. J . Limnology, 59(1): 72-76. Canter-Lund, H. and J.W.G. Lund. 1995. Freshwater Algae - Their Microscopic World Explored. BioPress Ltd. Bristol UK. 360p. Carpenter, S.R. and D.M. Lodge. 1986. Effects of submersed macrophytes on ecosystem processes. Aquatic Botany 26: 341-370. C A R R Environmental Consultants and AIM Ecological Consultants Ltd. 2002. Synthesis of Vegetation and Soil Studies for Revelstoke Reach - Upper Arrow Reservoir. Prepared for BC Hydro. 22 pages. Carr, W.W., A . E . Brotherston and A.I. Moody. 1993. B.C. Hydro Upper Arrow Dust Control Program - Revegetation and Special Studies. Program Summary and Recommendations 1990 -1993. Report prepared for B.C. Hydro. Cattaneo, A. 1983. Grazing on epiphytes. Limnology and Oceanography 28:124-132. Cattaneo, A. 1990. The effect of fetch on periphyton variation. Hydrobiologia 206:1-10. Cattaneo, A., and J . Kalff. 1980. The relative contribution of aquatic macrophytes and their epiphytes to the production of macrophyte beds. Limnology and Oceanography 25: 280-289 120 Cattaneo, A., and J. Kalff. 1986. The effect of grazer site manipulation on periphyton communities. Oecologia 69:612-617. Chambers, P. A., R.E. DeWreede, E.A. Irlandi, and H. Vandermeulen (1999). "Management issues in aquatic macrophyte ecology: a Canadian perspective." Can. J. Bot 77: 471-487. Clarke, K.R. 1993. Non-parametric multivariate analyses of changes in community structure. Australian Journal of Ecology 18:117-143. Cowell, B.C. and P.L. Hudson. 1968. Some environmental factors influencing benthic invertebrates in two Missouri River reservoirs. In: Reservoir Fishery Resources Symposium. University of Georgia Press, Athens. Southern Division, American Fisheries Society Special Publication, pp 541-555. Crawford, R.M.M. and R. Braendle. Oxygen deprivation stress in a changing environment. Journal of Experimental Botany, Oxford, v.47, p.145-159, 1996. Crossley, M.N. 2002. The Effects of Water Flow, pH and Nutrition on the Growth of the Native Aquatic Plant, Aponogeton elongates. A thesis submitted for the degree of Master of Philosophy at the University of Queensland, Gatton. Crow, J.H. and K.B. Macdonald. 1978. Wetland Values: Secondary Production. In "Wetland Functions and Values: The State of Our Understanding. Proceedings of the National Symposium on Wetlands. P.E. Greeson, J.R. Clark, and J .E . Clark, eds. American Water Resources Association, Minneapolis, Minn. Cyr, H. and J.A. Downing. 1988. Empirical relationships of phytomacrofaunal abundance to plant biomass and macrophyte bed characteristics. Can. J. Fish. Aquat. Sci. 45:976-984. Dias-Filho, B. Moacyr and C.J.R. de Carvalho. 2000. Physiological and morphological responses of Brachiaria spp. to flooding. Pesq. agropec. bras., Oct. 2000, vol.35, no. 10, p. 1959-1966. Department of Fisheries and Oceans, 1999. Authorization No. 1999-SF-1. Schedule 2: Stave Reservoir Compensation Program - Shoreline and Drawdown Zone Vegetation Enhancement Plan. Dibble, E.D., K.J. Killgore and S.L. Harrel. 1996. Assessment of Fish-Plant Interactions. American Fisheries Society Symposium 16:357-372. Diehl, S. and R. Kornijow. 1997. "Influence of Submerged Macrophytes on Trophic Interactions Among Fish and Macroinvertebrates", in: The Structuring Role of Submerged Macrophytes in Lakes. E. Jeppesen, M. Sondergaard, M. Sondergaard, and K. Christoffersen, eds. Springer-Verlag, New York, NY. Dierberg, F.E. 1993. Decomposition of desiccated submersed aquatic vegetation and bioavailability of released phosphorus. Lake and Reserv. Manage. 8(1):31-36. Dodds, W.K. 1991. Community interactions between the filamentous alga Cladophora glomerata (L.) Keutzing, its epiphytes, and epiphyte grazers. Oecologia 85:572-580. 121 Drew, M.C. 1997. Oxygen deficiency and root metabolism: injury and acclimation under hypoxia and anoxia. Annual Review of Plant Physiology and Plant Molecular Biology, Palo Alto, v.48, p.223-250. Edmondson, W.T. 1959. Freshwater biology. 2nd ed. John Wiley & Sons, New York. 1248 pp. Eichenberger, E. and A. Schlatter. 1978. Effect of herbivorous insects on the production of benthic algal vegetation in outdoor channels. Verh. Int. Verein Limnol. 20:1806-1810. Ernst, W.H.O. 1990. Ecophysiology of plants in waterlogged and flooded environments. Aquatic Botany 38: 73-90. Fairchild, G.W. 1981. Movement and distribution of Sida crystalline and other littoral microcrustacea. Ecology 62:1341 -1352. Fairchild, G.W. and R.L. Lowe. 1984. Artificial substrates which release nutrients: effects on periphyton and invertebrate succession. Hydrobiologia 114:29-37 Feminella, J.W. and C P . Hawkins. 1995. Interactions between stream herbivores and periphyton: a quantitative analysis of past experiments. J. North Am. Benth. Soc. 14:465-509. Ford, D. 1990. Reservoir transport processes, pp. 15-41 in: Reservoir limnology: Ecological perspectives. K.W. Thornton, B.L. Kimmel and F.E. Pyne (eds.). Wiley-lnterscience, Toronto 246 p. Fraley, J . , B. Marotz, J . Decker-Hess, W. Beattie, and R. Zubik. 1989. Mitigation, Compensation, and Future Protection for Fish Populations Affected by Hydropower Development in the Upper Columbia System, Montana, U.S.A. Regulated Rivers: Research and Management, Vol 3: 3-18. Furey P.C., A. Mazumder, and R.N. Nordin. 2002. Sediment Characteristics and benthic communities under reservoir drawdown. Presented at the North American Benthological Society Annual Meeting, Pittsburg, PA. in Lentic Ecology. Gallepp, G. W. 1979. Chironomid influence on phosphorus release in sediment-water microcosms. Ecology 60:547-556. Gallepp, G.W., J.F. Kitchell, and S.M. Bartell. 1978. Phosphorus released from lake muds affected by chironomids. Verh. Internal Verein. Limnol. 20:458-474. Gardner, W.S., T.F. Nalepa, M.A. Quigley, and J.M. Malczyk. 1981. Release of phosphorus by certain benthic invertebrates. Can. J. Fish. Aquat. Sci. 38:978-981. Gardner, W.S., T.F. Nalepa, D.R. Slavens, and G.A. Laird. 1983. Patterns and rates of nitrogen release by benthic Chironomidae and Oligochaeta. Can. J. Fish. Aquat. Sci. 40:259-266. Gelwick F.P. and W.J. Matthews. 1992. Effects of an algivorous minnow on temperate stream ecosystem properties. Ecology 73, 1630 -1645 122 Godshalk, G.L. 1977. Decomposition of aquatic plants in lakes. Ph.D. Dissertation, Michigan State University. 139 pp. + 340 figs. Gosselink, J .G . and C.J . Kirby. 1974. Decomposition of salt marsh grass, Spartina alterniflora Loisel. Limnol. Oceanogr. 19:825-832. Graneli, W. 1979. The influence of Chironomus plumosus larvae on the exchange of dissolved substances between sediment and water. Hydrobiologia 66:149-159 Gregg, W.W. and F.L. Rose. 1985. Influences of aquatic macrophytes on invertebrate community structure, guild structure, and microdistribution in streams. Hydrobiologia 128: 45-56. Gregory, S.V. 1983. Plant-herbivore interactions in stream systems, pp. 157-190 in J.R. Barnes and G.W. Minshall (eds.) Stream Ecology. Plenum Press, New York. Gressens, S .E . and R.L. Lowe. 1994. Periphyton patch preference in grazing chironomid larvae. J. North Am. Benth. Soc. 13:89-99. Hann, B.J. 1991. Invertebrate grazer-periphyton interactions in a eutrophic marsh pond. Freshwater Biology 26:87-96. Hargeby, A. 1990. Macrophyte associated invertebrates and the effect of habitat permanence. Oikos 57: 33-346. Hart, D.D. 1981. Foraging and resource patchiness: field experiments with a grazing stream insect. Oikos 37:46-52. Healy, J.B., Jr., L.Y. Young and M. Reinhard. 1980. Methanogenic decomposition of ferulic acid, a model lignin derivative. Appl. Environ. Microbiol. 39:436-444. Hicks, C.R. and K.V. Turner Jr. 1999. Fundamental concepts in the design of experiments - 5 t h ed. Oxford University Press, Inc. New York, NY. Hirst, S.M. 1991. Impacts of the operation of existing hydroelectric developments on fishery resources in British Columbia. Volume II. Inland Fisheries. Can. Manuscr. Rep. Fish. Aquat. Sci. 2093: 200 p. Hunt, P.C. and J.W. Jones. 1972. The effect of water level fluctuations on a littoral fauna. J. Fish Biol. 4:385-394. Hynes, H.B.N. 1970. Further adaptations of benthic invertebrates. Pages 161-182 in: The ecology of running waters. University of Toronto Press. Ontario, Canada. Izumi, H., A. Hattori, C P . McRoy. 1980. Nitrate and nitrite in interstitial waters of eelgrass beds in relation to the rhizosphere. J. Exp. Mar. Biol. Ecol. 47:191-201. Jackson M.B. 1985. Ethylene and responses of plants to soil waterlogging and submergence. Annu. Rev. Plant Physiol. 36:145-174. 123 Jeppesen, E., M. Sondergaard, M. Sondergaard, and K. Christoffersen. 1997. Preface in: "The Structuring Role of Submerged Macrophytes in Lakes". E. Jeppesen, M. Sondergaard, M. Sondergaard, and K. Christoffersen, eds. Springer-Verlag, New York, NY. Johnston, NT . , J.S. MacDonald, K.J. Hall, and P.J. Tschaplinski. 1997. A Preliminary Study of the Role of Sockeye Salmon (Oncorhynchus nerka) Carcasses as Carbon and Nitrogen Sources for Benthic Insects and Fishes in the "Early Stuart" Stock Spawning Streams, 1050 km from the Ocean. British Columbia Ministry of Environment, Lands and Parks Fisheries Project Report No. RD55. Jones, J.I., B. Moss, and J.O. Young. 1997. Interactions between Periphyton, Nonmolluscan Invertebrates, and Fish in Standing Freshwaters. In: "The Structuring Role of Submerged Macrophytes in Lakes". E. Jeppesen, M. Sondergaard, M. Sondergaard, and K. Christoffersen, eds. Springer-Verlag, New York, NY. Kairesalo, T. 1984. The seasonal succession of epiphytic communities within an Equisetum fluviatile L. stand in Lake Paajarvi, Southern Finland. Internationale Rev. Ges. Hydrobiol. 69:475-505. Karouna, N.K. and R.L. Fuller. 1992. Influence of four grazers on periphyton communities associated with clay tiles and leaves. Hydrobiologia 245:53-64 Kaster, J.L. and G.Z. Jacobi. 1978. Benthic macroinvertebrates of a fluctuating reservoir. Freshwater Biology 8, 283-290. Kawase, M. 1981. Effect of Ethylene on Aerenchyma Development. American Journal of Botany, Vol. 68, No. 5 pp. 651-658 Kessler, D.H. 1981. Grazing rate determination of Corynoneura scutellata Winnertz (Chironomidae: Diptera). Hydrobiologia 80:63-66. Kistritz, R.U., K.J. Hall, and I. Yesaki. 1983. Productivity, Detritus Flux, and Nutrient Cycling in a Carex lyngbyei Tidal Marsh. Estuaries 6(3): 227-236. Kohler, S.L. 1984. Search mechanism of a stream grazer in patchy environments: the role of food abundance. Oecologia 62:209-218. Kornijow, R., R.D. Gulati, and T. Ozimek. 1995. Food preference of some freshwater macroinvertebrates: comparing fresh and decomposed vascular plants and a filamentous alga. Freshwat. Biol. 33:205-212. Kornijow, R., R.D. Gulati, and E. van Donk. 1990. Hydrophyte-macroinvertebrate interactions in Zwemhulst, a lake undergoing biomanipulation. Hydrobiologia 200/201:467-474 Kornijow, R., T. Kairesalo. 1994. Elodea canadensis sustains rich environment for macroinvertebrates. Verh. Int. Verein. Limnol. 25:4098-4111. Kozlowski, T.T. 1997. Responses of woody plants to flooding and salinity. Tree Physiology, v.1, 1997. Heron Publishing site. URL: http://www.heronpublishing.com/tp/monograph/kozlowski.pdf 124 Krull, J.M. 1970. Aquatic plant-macroinvertebrate associations and waterfowl. Journal of Wildlife Management 43:707-714. Lamberti, G.A., J.W. Feminella and V.H. Resh. 1987. Herbivory and intraspecific competition in a stream caddisfly population. Oecologia 73:75-81. Lamberti, G.A., S.V. Gregory, L.R. Ashkenas, A.D. Steinman and C D . Mclntire. 1989. Productive capacity of periphyton as a determinant of plant-herbivore interactions in streams. Ecology 70:1840-1856. Lamberti, G.A. and J.W. Moore. 1984. "Aquatic Insects as primary consumers" in The Ecology of Aguatic Insects. V.H. Resh and D.M. Rosenberg, eds. Praeger. New York, N.Y. Lamberti, G.A. and V.H. Resh. 1983. Stream periphyton and insect herbivores: an experimental study of grazing by a caddisfly population. Ecology 64: 1124-1135. Lavkulich, L.M. 1977. Methods Manual, Pedology Laboratory. Department of Soil Science, University of British Columbia. Lavkulich, L.M. 1978. Methods Manual, Pedology Laboratory. Department of Soil Science, University of British Columbia. Levings, C D . 1997. Synthesis and Conclusions In: Levings, C D . and D.J.H. Nishimura. 1997. Created and Restored Marshes in the Lower Fraser River, British Columbia: Summary of Their Functioning as Fish Habitat. Water Qual. Res. J. Can. Vol 32, No. 3, 599-618. Likens, G .E . and O.L. Loucks. 1978. Analysis of five North American lake ecosystems. III. Sources, loading and fate of nitrogen and phosphorus. Verh. Internal Ver. Theor. Angew. Limnol. 20:568-573. Lodge, D.M. 1991. Herbivory on freshwater macrophytes. Aquat. Bot. 41:195-224. McAfee, M.E. 1980. Effects of water drawdown on the fauna in small cold water reservoirs. Water Resources Bulletin American Water Resources Association. Vol. 16, No. 4:690-696. McCann, K.S. 2000. The diversity-stability debate. Nature 405: 228-233. McCarron, J.K., K.W. McLeod, and W.H. Conner. 1998. Flood and Salinity Stress of Wetland Woody Species, Buttonbush (Cephalanthus occidentalis)and Swamp Tupelo (Nyssa sylvatica var. biflora). WetlandsW{2): 165-1 75. McCormick, P.V. 1994. Evaluating the multiple mechanisms underlying herbivore-algal interactions in streams. Hydrobiologia 291:47-59 McCormick, P. V., and R. J . Stevenson. 1991. Grazer control of nutrient availability in the periphyton. Oecologia 86:287-291. McCune, B. and J.B. Grace. 2002. Analysis of Ecological Communities. MjM Software Design. Gleneden Beach, Oregon. 125 McCune, B. and M.J. Mefford. 1999. PC-ORD for Windows. Multivariate analysis of Ecological Data Version 4.14. McKeague, J.A. 1987. Manual of Soil Sampling and Methods of Analysis. Agriculture Canada Research Branch. McKee, K.L., I .A. Mendelson, and D.M. Burdick. 1989. Effect of long term flooding on root metabolic response in five freshwater marsh plant species. Canadian Journal of Botany 67:3446-3452 McLaughlin, D.B. and H.J. Harris. 1990. Aquatic insect emergence in two Great Lakes marshes. Wetlands Ecology and Management. 1:111-121. McRoy, C P . , R.J. Barsdate, and M. Nebert. 1972. Phosphorus cycling in an eelgrass (Zostera marina L.) ecosystem. Limnol. Oceanogr. 17:58-67. Malone, M. and I. Ridge. 1983. Ethylene-induced growth and proton excretions in the aquatic plant Nymphoides peltata. Planta 157:71-73 Marotz, B.L., D. Gustafson, and C. Althen. 1994. Model Development to Establish Integrated Operational Rule Curves for Hungry Horse and Libby Reservoirs Montana. Draft report prepared for Bonneville Power Administration. Mason, C F . and R.J. Bryant. 1975. Periphyton production and grazing by chironomids in Alderfen broad, Norfolk. Freshwater Biol. 5:271-277. Merritt, R.W. & K.W.K Cummins, eds. 1984. An introduction to the aquatic insects of North America, 2 ed. Kendall/Hunt, Dubuque, Iowa. 722 pp. Merritt, R.W. and K.W. Cummins. 1996. Trophic relations of macroinvertebrates. In: Methods in Stream Ecology. F.R. Hauer and G. Lamberti, eds. Academic Press. 674 pp. Merritt, R.W., K.W. Cummins, and T.M. Burton. 1984. "The Role of Aquatic Insects in the processing and cycling of nutrients" in The Ecology of Aquatic Insects. V H . Resh and D.M. Rosenberg, eds. Praeger. New York, N.Y. Mitsch, W.J. and J .G . Gosselink. 2000. Wetlands. Third Edition. John Wiley & Sons, Inc. New York, NY. Moody, A.I. 2002. Long term monitoring of vegetation expansion and trials in the dust control treatment areas of Revelstoke Reach - Upper Arrow Reservoir. Prepared for BC Hydro. 29 pages. Muthukumar, T., K. Udaiyan and P. Shanmughavel. 2004. Mycorrhiza in s e d g e s - a n overview. Mycorrhiza 14:65-77. Northcote, T .G . and D.Y. Atagi. 1997. Ecological Interactions in the Flooded Littoral Zone of Reservoirs: The Importance and Role of Submerged Terrestrial Vegetation with Special Reference to Fish, Fish Habitat and Fisheries in the Nechako Reservoir of British Columbia, 126 Canada. British Columbia Ministry of Environment, Lands and Parks Skeena Fisheries Report SK-111. Ogwang, B.H. 1979. Some Plant-mediated processes in the maritime wetlands of southwestern British Columbia. Unpublished Ph.D. Thesis. Department of Plant Science. University of British Columbia. Olson, M.H., G .G. Mittelbach, and C W Osenberg. 1995. Competition between predator and prey: resource-base mechanisms and implications for stage-structured dynamics. Ecology 76:1758-1771. Osenberg, C W , G .G. Mittelbach, and P . C Wainwright. 1992. Two-stage life histories in fish: the interaction between juvenile competition and adult performance. Ecology 73:255-267. Ottosen, L.D.M., N. Risgaard-Petersen, and L.P. Nielsen. 1999. Direct and indirect measurements of nitrification and denitrification in the rhizosphere of aquatic macrophytes. Aquat. Microb. Ecol. Vol. 19: 81-91. Pace, M.L., G.B. McManus and S .E .G . Findlay. 1990. Plankton community structure determines the fate of bacterial production in a temperate lake. Limnol. Oceanogr. 35: 795-808. Page, N.A. 2003. Community and Regional Scale Patterns of Native and Exotic Plant Species in Sand Beaches of Vancouver Island, British Columbia, p. 79, Resource Management and Environmental Studies. University of British Columbia. Palmer, M.J., A.P. Covich, S. Lake, P. Biro, J .J . Brooks, J . Cole, C. Dahm, J . Gibert, W. Goedkoop, K. Martens, J . Verhoeven, and W.J. Van de Bund. 2000. Linkages between Aquatic Sediment Biota and Life above Sediments as Potential Drivers of Biodiversity and ecological Processes. Bioscience 50, No. 12: 1062-1075. Parkinson, J.A. and S.E. Allen. 1975. A wet oxidation procedure for the determination of nitrogen and mineral nutrients in biological material. Comm. In Soil Science and Plant Analysis 6(1): 1-11. Patterson, C G . and C H . Fernando. 1969. The effect of winter drainage on reservoir benthic fauna. Can. J. Zool. 47, 589-595. Peltier, W.H. and E.B. Welch. 1970. Factors affecting growth of rooted aquatic plants in a reservoir. Weed Science 18: 7-9. Pennak, R.W. 1978. Fresh water invertebrates of the United States, 2nd ed. John Wiley & Sons, New York. 803 pp. Perrin, C.J . , RL&L Environmental Services Ltd. and J .G . Stockner. 2001. Biofilm, invertebrate and fish communities associated with vegetation strata in the drawdown zone of the Arrow Lake Reservoir. Draft final report. Prepared by Limnotek Research and Development Inc., Vancouver, BC, for BC Hydro, Burnaby BC. 88 p. 127 Perrin, C.J . , Golder Associates (RL&L Ltd.) and J .G . Stockner. 2002. Biofilm, invertebrate and fish communities associated with vegetation strata in the drawdown zone of the Arrow Lake Reservoir. Final report. Prepared for BC Hydro, Burnaby BC. 79 p. Perrin, C.J . 1997. Fertilization and monitoring of Wahleach Reservoir in 1996. Limnotek Research and Development Inc., Vancouver BC. Prepared for BC Hydro. Persson, L. 1988. Asymmetries in competitive and predatory interactions in fish populations. In: Ebenmann, B. and Persson. L., eds. Size-structured populations - ecology and evolution. Heidelberg: Springer-Verlag: 1988:203-218. Persson, L. and L.B. Crowder. 1990. Fish-Habitat Interactions Mediated via Ontogenetic Niche Shifts. Pp. 3-23 In: "The Structuring Role of Submerged Macrophytes in Lakes". E. Jeppesen, M. Sondergaard, M. Sondergaard, and K. Christoffersen, eds. Springer-Verlag, New York, NY. Pezeshki, S.R. 1994. Plant response to flooding. In: Wilkinson, R.E. (Ed.). Plant-environment interactions. New York : M. Dekker. p289-321. Poff, N.L. and J.V. Ward. 1992. Heterogeneous currents and algal resources mediate in situ foraging activity of a mobile stream grazer. Oikos 65:465-478. Pomeroy, L.R. 1974. The ocean's food web: a changing paradigm. Bioscience 24:499-504 Porter, K.G., H. Paerl, R. Hodson, M. Pace, J . Priscu, B. Riemann, D. Scavia, and J. Stockner. 1988. Microbial Interactions in Lake Food Webs. In: "Complex Interactions in Lake Communities", S.R. Carpenter, ed. Springer-Verlag, New York, NY. Power, M.E. 1990. Effects offish in river food webs. Science 250:811-814. Prescott, G.W. 1978. How to know the freshwater algae. Third Edition. W C B McGraw-Hill. New York, NY. Puriveth, P. 1980. Decomposition of emergent macrophytes in a Wisconsin marsh. Hydrobiologia 72:231-242. Rasmussen, J.B. 1988. Littoral zoobenthic biomass in lakes, and its relationships to physical, chemical and trophic factors. Can. J. Fish. Aquat. Sci. 45:1438-1447 Raven, J.A. 1984. Energetics and transport in aquatic plants. Alan R. Liss, Inc. New York Resh V.H. and G. Grodhaus. 1983. Aquatic insects in urban environments. In Urban Entomology: Interdisciplinary Perspectives, eds. G.W. Frankie and C.S. Koehler, pp. 247-76. Praeger Pubs., New York. Robe, W.E . and H. Griffiths. 1998. Adaptations for an amphibious life: changes in leaf morphology, growth rate, carbon and nitrogen investment, and reproduction during adjustment to emersion by the freshwater macrophyte Littorella uniflora. New Phytol. (1998) 140, 9 - 23. Robinson, G.G.C. , S .E . Gurney and L.G. Goldborough. 1997. The Primary Productivity of Benthic and Planktonic Algae in a Prairie Wetland Under Controlled Water-Level Regimes. Wetlands, Vol 17, No. 2. pp 182-194. 128 Roseff, S. Jr. and J.M. Bernard. 1979. Seasonal changes in carbohydrate levels in tissues of Carex lacustris. Can. J. Bot. 57:2140-2144. Rosemond, A.D., P.J. Mulholland and J.W. Elwood. 1993. Top-down and bottom-up control of stream periphyton: effects of nutrients and herbivores. Ecology 74:1264-1280. Sand-Jensen, K., C. Prahl, and H. Stockholm. 1982. Oxygen release from roots of submerged aquatic macrophytes. Oikos 38: 349-354. SAS Institute Inc. 2005. JMP IN Statistical Discovery Software. Version 5.1.2. Sculthorpe, C. D. 1967. The Biology of Aquatic Vascular Plants. St. Martin's Press, New York. 610 pp. Scholander, P.F., L. van Dam, and S. I. Scholander. 1955. Gas exchange in the roots of mangroves. American Journal of Botany 42:92-98. Schwarz, C.J . 2005. Stat-403/Stat-650 Intermediate Sampling and Experimental Design and Analysis 2005. Department of Statistics and Actuarial Science, Simon Fraser University, Burnaby B.C. Shortreed, K.S and J .G . Stockner. 1978. Response of attached algae to whole-lake fertilization experiments in five British Columbia lakes. Fish. Mar. Tech. Rep. No. 802, 25p. Sieburth, J.McN., V. Smetacek and J. Lenz. 1978. Pelagic ecosystem structure: heterotrophic compartments of the plankton and their relationship to plankton size fractions. Limnol. Oceanogr. 23: 1256-1263. Sinclair, D.C. 1965. The effects of water level changes on the limnology of two British Columbia coastal lakes with particular reference to the bottom fauna. M.Sc. thesis, Department of Zoology, University of British Columbia, Vancouver, B.C. Canada, 84 pages. Soballe D.M., B .L Kimmel, R.H. Kennedy, and R.F. Gaugush. 1992. Reservoirs, pp 421-474 in C T . Hackney, S.M. Adams, and W.H. Martin eds. Biodiversity of the southeastern United States, Aquatic Communities. Wiley, New York, NY. Solander, D. 1983a. Biomass and Shoot Production of Carex rostrata and Equisetum fluviatile in unfertilized and fertilized subarctic lakes. Aquat. Bot. 15:349-366. Solander, D. 1983b. Biomass, production and nutrient content of macrophytes in a natural and a fertilized subarctic lake. Abstracts. Ph.D. Thesis, University of Uppsala, Sweden. 17 pp. Steed, J .E . , L.E. DeWald and T .E . Kolb. 2002. Physiological and Growth Responses of Riparian Sedge Transplants to Groundwater Depth. Int. J. Plant Sci. 163(6):925-936. Stockner, J .G . 1991. Autotrophic Picoplankton in Freshwater Ecosystems: The View from the Summit. Int. Revue ges. Hydrobiol. 76:4 pp 483-492 129 Stockner, J .G . 2001. Microbial Ecology and multiple stressors. A "key" role for microorganisms in nutrient and carbon fluxes in aquatic ecosystems: an emerging paradigm. Unpublished paper. Stockner, J .G . and N.J. Antia. 1986. Algal Picoplankton from Marine and Freshwater Ecosystems: A Multidisciplinary Perspective. Can. J. Fish. Aquat. Sci. 43: 2472-2503 Stockner, J .G . and F.A.J. Armstrong. 1971. Periphyton of the Experimental Lakes Area, Northwestern Ontario. J . Fish. Res. Board Can. 28:215-229. Stockner, J .G . and J. Beer. 2004. The limnology of Stave/Hayward reservoirs: with a perspective on carbon production. Prepared for BC Hydro Water Use Plans by Eco-logic Ltd. And University of British Columbia, Institute for Resources, Environment and Sustainability, Vancouver. 33 pp. + App. Stockner, J .G . , A. Langston, D. Sebastion and G. Wilson. 2005. The Limnology of Williston Reservoir: British Columbia's Largest Lacustrine Ecosystem. Water Qual. Res. J . Canada, 40: (1) 28-50. Stockner, J .G . and K.G. Porter. 1988. Microbial Food Webs in Freshwater Planktonic Ecosystems. In: "Complex Interactions in Lake Communities", S.R. Carpenter, ed. Springer-Verlag, New York, NY. Stockner, J .G . , E. Rydin, and P. Hyenstrand. 2000. Cultural oligotrophication: Causes and consequences for fisheries resources. Fisheries, 25: (5) 7-14. Stockner, J .G . and K.S. Shortreed. 1976. Enhancement of autotrophic production by nutrient addition in a coastal stream on Vancouver Island. J . Fish. Res. Bd. Canada 35: 28-34. Stockner, J .G . and K.S. Shortreed. 1989. Algal picoplankton production and contribution to food-webs in oligotrophic British Columbia Lakes. Hydrobiologia, 173: 151-166. Stockner, J .G . and E.A. Maclsaac. 1996. British Columbia Lake Enrichment Programme: Two Decades of Habitat Enhancement for Sockeye Salmon. Regulated Rivers Research and Management, Vol. 12: 547-561) Suren, A.M, and P.S. Lake. 1989. Edibility of fresh and decomposing macrophytes to three species of freshwater invertebrate herbivores. Hydrobiologia 178:165-178 Tarnocai, C. 1979. Canadian Wetland Registry. In: Proceedings of a Workshop on Canadian Wetlands Environment. C.D.A. Rubec and F.C. Pottett, eds. Canada Land Directorate, Ecological Land Classification Series, No. 12. pp. 9-38. Taxon Aquatic Monitoring Company. 1993. The role of in-reservoir-produced terrestrial vegetation in the aquatic ecosystem. Blue River Reservoir. Report prepared for Willamette National Forest Oregon. Technicon Autoanalyzer II Methodology: Individual/Simultaneous Determination of Nitrogen and Phosphorus in BD Acid Digests. Industrial Method No. 334-74A. 130 Thornton, K.W. 1990. Perspectives on reservoir limnology, Pages 1-14 in: Reservoir limnology: Ecological perspectives. K.W. Thornton, B.L. Kimmel and F.E. Pyne (eds.). Wiley-lnterscience, Toronto 246 p. Tokeshi, M. 1986. Resource utilization, overlap and temporal community dynamics. Journal of Animal Ecology. 55:491-506. Tran, T. Sen and R.R. Simard. 1993. Mehlich III - Extractable Elements. In: Soil Sampling and Methods of Analysis. M.R. Carter, ed. Canadian Society of Soil Science. Lewis Publishers. Trojanowski, J . 1969. Biological degradation of lignin. Int. Biodetn. Bull. 5:119-124. Underwood, G.J .C. 1991. Growth enhancement of the macrophyte Ceratophyllum demersum in the presence of the snail Planorbis planorbis: the effect of grazing and chemical conditioning. Freshwater Biology 26: 325-334 Vaughn, C .C. 1986. The role of periphyton abundance and quality in the microdistribution of a stream grazer, Helicopsyche borealis (Trichoptera: Helicopsychidae). Freshwat. Biol. 16:485-493. Visser, E.J.W., G.M. Bogemann, H.M. Van De Steeg, R. Pierik and G.W.P.M. Blom. 2000. Flooding tolerance of Carex species in relation to field distribution and aerenchyma formation. New Phytol. 148: 93-103. Voesenek, L.A.C.J. 1990. Adaptation of Rumex in Flooding Gradients. Thesis, Catholic University of Nijmegen, Nijmegen, Netherlands. 159 pp. Voigts, D.K. 1976. Aquatic invertebrate abundance in relation to changing marsh vegetation. American Midland Naturalist 95:313-322. Walshe, BM. 1951. The feeding habits of certain Chironomid larvae (subfamily Tendipedinae): Proceedings of the Zoological Society of London. 121: 63-79. Weller, M.W. 1994. Freshwater marshes, 3 r d ed. University of Minnesota Press, Minneapolis. 192 pp. Wetzel, R.G. 1983. Limnology. 2 n d edition. Philadelphia: Saunders College Publishing. Wetzel, R.G. 1990. Reservoir Ecosystems: Conclusions and Speculations. Pp 227-238 In: K.W. Thornton, B.L. Kimmel and F.E. Payne. Reservoir Limnology: Ecological Perspectives. John Wiley and Sons, Inc. Toronto. Wetzel, R.G. 1990b. Land-water interfaces: metabolic and limnological regulators. Verh. Int. Verein. Limnol. 24:6-24 Wetzel, R.G. 1993. Microcommunities and microgradients: linking nutrient regeneration, microbial mutualism, and high sustained aquatic primary production. Netherlands J. Aquat. Ecol. 27:3-9 131 Wetzel, R.G. 1996. Benthic Algae and nutrient cycling in lentic freshwater ecosystems. In: Stevenson, R.J, M.L. Bothwell, and R.L. Lowe eds. Algal ecology: freshwater benthic ecosystems. New York: Academic Press. 1996:641-667 Wetzel, R.G. and B.A. Manny. 1972. Secretion of dissolved organic carbon and nitrogen by aquatic macrophytes. Verh. Internat. Verein. Limnol. 18: 229-240. Wetzel, R.G. and M. Sondergaard. 1997. Role of Submerged Macrophytes for the Microbial community and Dynamics of Dissolved Organic Carbon in Aquatic Ecosystems. In: "The Structuring Role of Submerged Macrophytes in Lakes". E. Jeppesen, M. Sondergaard, M. Sondergaard, and K. Christoffersen, eds. Springer-Verlag, New York, NY. Wetzel, R.G. and G. van der Valk. 1998. Effects of nutrient and soil moisture on competition between Carex stricta, Phalaris arundinacea, and Typha latifolia. Plant Ecol 138:179-190. Wilson, G., R. Land, K. Ashley, T. Berkhout, F. Pick, M. McCusker, G. Scholten, and D. Sebastion. 2000. The Alouette Reservoir Fertilization Experiment: Year Two (1999) Report: The first summer of fertilization. Fisheries Project Report No. RD 85. Wilson, S.J. 1998. Stave Falls Powerplant Replacement Project Stave Lake Reservoir Drawdown, March/April, 1998. Environmental Management Plan. Prepared for BC Hydro Power Supply - Maintenance, Engineering and Projects Wilson, S.J. 1999. Stave Falls Powerplant Replacement Project: Planned Drawdown of Stave Reservoir, March/April, 1999. Environmental Management Plan. Prepared for BC Hydro Power Supply Engineering. 132 Appendix A. Plant Biomass, Nitrogen and Phosphorus Data Table A-1 Aboveground Plant Biomass, Nitrogen and Phosphorus Data Site Time Elevation Species Total N Total P * Total LOI Corrected (m) (foliage) (foliage) Foliage (%) Foliage (%) (%) Weight (g) Weight (g) Pre 0 Woolgrass 1.83 0.17 63.18 94.6 59.77 Pre 0 Woolgrass 1.62 0.16 34.37 94.9 32.62 Pre 0 Woolgrass 1.78 0.16 12.41 94.7 11.75 Pre 0 Woolgrass 1.57 0.16 25.08 94.2 23.63 Pre 0 Woolgrass 1.92 0.17 36.68 90.3 33.12 Pre 0 Sedge 1.69 0.12 25.52 94.6 24.13 Pre 0 Sedge 1.99 0.17 22.06 93.5 20.62 Pre 0 Sedge 1.96 0.17 15.31 94.9 14.53 Pre 0 Sedge 3.02 0.23 10.81 92.7 10.02 Pre 0 Sedge 1.76 0.14 24.63 95.0 23.38 Pre 0 Sedge 1.52 0.20 19.09 94.6 18.06 Pre 0 Fall Rye 3.22 0.70 54.68 88.8 48.55 Pre 0 Fall Rye 2.98 0.64 52.02 90.1 46.87 Pre 0 Fall Rye 3.17 0.75 55.67 90.9 50.60 Pre 0 Fall Rye 2.80 0.62 57.02 90.6 51.66 Pre 0 Fall Rye 2.79 0.56 61.48 88.6 54.47 Pre 0 Fall Rye 3.12 0.63 58.94 85.2 50.22 Pre 0 Fall Rye 3.22 0.53 60.49 90.9 54.98 Site A 1 80 m Woolgrass 1.41 0.14 19.62 89.6 17.58 Site B 1 80 m Woolgrass 1.37 0.13 13.82 89.6 12.38 S i t e C 1 80 m Woolgrass 1.34 0.14 17.28 90.6 15.65 Site A 1 80 m Sedge 1.31 0.11 11.87 83.7 9.94 Site B 1 80 m Sedge 1.22 0.15 10.26 81.8 8.39 S i t e C 1 80 m Sedge 1.27 0.11 8.47 89.6 7.59 Site A 1 80 m Fall Rye 1.70 0.14 22.05 59.6 13.14 Site B 1 80 m Fall Rye 1.52 0.16 13.64 68.0 9.27 S i t e C 1 80 m Fall Rye 1.06 0.11 23.58 51.5 12,14 Site A 2 80 m Woolgrass 1.46 0.16 18.27 74.2 13.55 Site B 2 80 m Woolgrass 1.51 0.17 14.14 82.7 11.69 S i t e C 2 80 m Woolgrass 1.42 0.16 39.40 87.0 34.28 Site A 2 80 m Sedge 1.52 0.12 11.88 71.0 8.43 Site B 2 80 m Sedge 1.39 0.15 8.40 78.4 6.58 S i t e C 2 80 m Sedge 1.45 0.12 6.51 69.7 4.54 Site A 2 80 m Fall Rye 1.36 0.12 21.12 65.8 13.90 Site B 2 80 m Fall Rye 1.05 0.13 12.69 59.9 7.60 S i t e C 2 80 m Fall Rye 0.94 0.10 21.29 67.9 14.46 Site A 3 80 m Woolgrass 1.38 0.16 32.25 83.7 26.99 Site B 3 80 m Woolgrass 1.66 0.19 22.18 87.5 19.41 S i t e C 3 80 m Woolgrass 1.26 0.14 36.07 86.9 31.35 Site A 3 80 m Sedge 2.32 0.29 9.21 84.1 7.74 Site B 3 80 m Sedge 1.49 0.15 14.31 81.6 11.68 133 Site Time Elevation Species Total N Total P * Total LOI Corrected (m) (foliage) (foliage) Foliage (%) Foliage (%) (%) Weight (g) Weight (g) S i t e C 3 80 m Sedge 1.33 0.14 21.15 72.3 15.29 Site A 3 80 m Fall Rye 1.09 0.10 10.26 67.9 6.97 Site B 3 80 m Fall Rye 1.21 0.12 31.69 68.3 21.64 S i t e C 3 80 m Fall Rye 1.21 0.11 12.03 83.4 10.03 Site A 1 78 m Woolgrass 1.88 0.20 17.20 90.6 15.59 Site B 1 78 m Woolgrass 1.11 0.11 15.23 87.9 13.38 S i t e C 1 78 m Woolgrass 1.38 0.13 21.24 90.4 19.20 Site A 1 78 m Sedge 2.65 0.25 5.33 85.5 4.56 Site B 1 78 m Sedge 1.37 0.14 11.39 88.0 10.02 S i t e C 1 78 m Sedge 1.46 0.13 11.39 88.1 10.03 Site A 1 78 m Fall Rye 1.70 0.15 25.53 73.4 18.74 Site B 1 78 m Fall Rye 1.02 0.12 17.40 48.0 8.35 S i t e C 1 78 m Fall Rye 1.21 0.12 32.82 59.7 19.59 Site A 2 78 m Woolgrass 1.61 0.21 62.45 81.7 51.02 Site B 2 78 m Woolgrass 1.43 0.17 44.86 82.4 36.96 S i t e C 2 78 m Woolgrass 1.22 0.13 26.02 74.9 19.49 Site A 2 78 m Sedge 1.05 0.10 10.41 71.4 7.43 Site B 2 78 m Sedge 1.68 0.16 3.69 42.9 1.58 S i t e C 2 78 m Sedge 2.72 0.22 5.77 71.8 4.15 Site A 2 78 m Fall Rye 1.29 0.14 23.20 73.0 16.93 Site B 2 78 m Fall Rye 0.89 0.10 41.37 35.3 14.60 Site C 2 78 m Fall Rye 0.85 0.14 18.29 52.3 9.56 Site A 3 78 m Woolgrass 1.67 0.16 26.99 87.2 23.54 Site B 3 78 m Woolgrass 1.43 0.16 22.26 82.3 18.32 S i t e C 3 78 m Woolgrass 1.08 0.11 29.79 67.6 20.14 Site A 3 78 m Sedge 1.33 0.11 8.41 72.6 6.10 S i teB 3 78 m Sedge 1.39 0.12 11.47 57.7 6.62 S i t e C 3 78 m Sedge 1.12 0.11 10.66 55.8 5.95 Site A 3 78 m Fall Rye 1.05 0.13 9.98 74.2 7.41 Site B 3 78 m Fall Rye 1.54 0.14 19.75 60.3 11.91 S i t e C 3 78 m Fall Rye 1.82 0.15 10.92 65.5 7.15 Site A 1 76 m Woolgrass 1.37 0.14 16.55 61.4 10.16 Site B 1 76 m Woolgrass 1.30 0.13 15.85 83.2 13.19 S i t e C 1 76 m Woolgrass 1.65 0.17 15.80 94.0 14.85 Site A 1 76 m Sedge 1.27 0.09 21.70 85.3 18.51 Site B 1 76 m Sedge 1.11 0.10 11.09 86.8 9.63 S i t e C 1 76 m Sedge 1.68 0.15 10.69 90.6 9.68 Site A 1 76 m Fall Rye 1.60 0.12 14.76 73.2 10.80 Site B 1 76 m Fall Rye 1.13 0.11 9.86 45.3 4.47 S i t e C 1 76 m Fall Rye 1.28 0.12 23.60 49.0 11.56 Site A 2 76 m Woolgrass 1.67 0.16 25.82 88.7 22.90 Site B 2 76 m Woolgrass 1.08 0.12 31.52 86.8 27.36 Site C 2 76 m Woolgrass 1.46 0.14 18.69 87.6 16.37 Site A 2 76 m Sedge 1.46 0.11 8.93 78.3 6.99 Site B 2 76 m Sedge 2.37 0.22 8.56 67.4 5.77 134 Site Time Elevation (m) Species Total N (foliage) (%) Total P (foliage) (%) * Total Foliage Weight (g) LOI (%) Corrected Foliage Weight (g) S i t e C 2 76 m Sedge 1.22 0.11 6.81 65.4 4.46 Site A 2 76 m Fall Rye 1.31 0.12 9.58 80.3 7.69 Site B 2 76 m Fall Rye 1.24 0.12 8.43 49.5 4.17 S i t e C 2 76 m Fall Rye 1.32 0.12 24.93 62.8 15.66 Site A 3 76 m Woolgrass 1.17 0.15 44.16 86.9 38.38 Site B 3 76 m Woolgrass 1.21 0.14 16.57 78.0 12.92 S i t e C 3 76 m Woolgrass 1.37 0.12 8.78 82.1 7.21 Site A 3 76 m Sedge 2.47 0.21 5.26 78.0 4.10 Site B 3 76 m Sedge 1.01 0.09 7.40 59.7 4.42 S i t e C 3 76 m Sedge 1.08 0.11 14.98 60.6 9.08 Site A 3 76 m Fall Rye 2.15 0.23 10.66 84.4 9.00 Site B 3 76 m Fall Rye 1.04 0.14 18.35 50.7 9.30 S i t e C 3 76 m Fall Rye 1.96 0.19 12.73 61.6 7.84 * Total Weight = sum of measured weights *2 to account for subsampling Table A-2 Belowground Plant Biomass, Nitrogen and Phosphorus Data Site Time Elevation (m) Species Total N (roots) (%) Total P (roots) (%) * Total Root Weight (g) A S H % Corrected Root Weight (g) Pre 0 Woolgrass 0.88 0.09 157.60 11.8 139.00 Pre 0 Woolgrass 0.74 0.07 66.09 8.2 60.68 Pre 0 Woolgrass 0.71 0.08 31.59 21.9 24.67 Pre 0 Woolgrass 0.62 0.07 88.06 10.6 78.73 Pre 0 Woolgrass 0.87 0.09 127.45 8.3 116.87 Pre 0 Sedge 0.59 0.05 131.59 16.4 110.01 Pre 0 Sedge 0.63 0.06 82.54 8.6 75.44 Pre 0 - Sedge 0.64 0.06 84.52 10.1 75.98 Pre 0 Sedge 0.70 0.06 45.29 11.1 40.26 Pre 0 Sedge 0.73 0.06 112.65 11.9 99.24 Pre 0 Sedge 0.46 0.07 67.72 15.4 57.29 Pre 0 Fall Rye 1.13 0.22 38.45 32.2 26.07 Pre 0 Fall Rye 1.27 0.24 30.01 28.7 21.39 Pre 0 Fall Rye 1.73 0.38 19.96 26.7 14.63 Pre 0 Fall Rye 1.56 0.29 26.68 27.4 19.37 Pre 0 Fall Rye 0.89 0.22 84.97 49.4 42.99 Pre 0 Fall Rye 1.20 0.22 32.12 34.8 20.94 Pre 0 Fall Rye 1.00 0.20 74.12 44.9 40.84 Site A 1 80 m Woolgrass 0.76 0.06 80.85 14.9 68.80 Site B 1 80 m Woolgrass 0.74 0.08 53.66 13.4 46.47 S i t e C 1 80 m Woolgrass 0.85 0.08 91.49 14.3 78.41 Site A 1 80 m Sedge 0.73 0.08 54.10 6.1 50.80 Site B 1 80 m Sedge 0.55 0.06 63.86 24.9 47.96 Site C 1 80 m Sedge 0.84 0.05 46.72 18.4 38.13 Site A 1 80 m Fall Rye 1.40 0.10 4.85 7.0 4.51 135 Site Time Elevation (m) Species Total N (roots) (%) Total P (roots) (%) * Total Root Weight (g) A S H % Corrected Root Weight (g) Site B 1 80 m Fall Rye 1.51 0.14 7.35 14.5 6.29 S i t e C 1 80 m Fall Rye 1.53 0.11 8.56 7.8 7.90 Site A 2 80 m Woolgrass 0.88 0.06 56.53 24.8 42.51 Site B 2 80 m Woolgrass 0.71 0.05 101.96 23.3 78.20 S i t e C 2 80 m Woolgrass 0.67 0.06 79.96 18.2 65.41 Site A 2 80 m Sedge 0.70 0.04 84.66 38.6 51.98 Site B 2 80 m Sedge 0.75 0.07 65.96 42.4 37.99 S i t e C 2 80 m Sedge 0.88 0.06 79.24 21.5 62.21 Site A 2 80 m Fall Rye 1.33 0.08 7.03 11.3 6.24 Site B 2 80 m Fall Rye 1.53 0.12 1.13 8.3 1.04 S i t e C 2 80 m Fall Rye 1.15 0.08 15.11 5.6 14.26 Site A 3 80 m Woolgrass 0.67 0.07 225.60 21.8 176.42 Site B 3 80 m Woolgrass 0.78 0.08 72.09 9.1 65.53 Site C 3 80 m Woolgrass 0.65 0.06 83.35 20.9 65.93 Site A 3 80 m Sedge 0.79 0.09 62.14 18.8 50.46 Site B 3 80 m Sedge 0.79 0.07 106.89 15.0 90.86 Site C 3 80 m Sedge 0.77 0.06 47.71 29.5 33.63 Site A 3 80 m Fall Rye 1.18 0.14 12.14 11.5 10.74 Site B 3 80 m Fall Rye 1.29 0.10 5.46 4.4 5.22 S i t e C 3 80 m Fall Rye 1.11 0.05 6.29 9.0 5.73 Site A 1 78 m Woolgrass 0.86 0.10 67.47 28.6 48.17 Site B 1 78 m Woolgrass 0.71 0.06 40.23 13.1 34.96 S i t e C 1 78 m Woolgrass 0.83 0.07 184.34 9.1 184.34 Site A 1 78 m Sedge 0.69 0.07 40.46 7.7 37.35 Site B 1 78 m Sedge 0.76 0.06 63.54 10.1 57.12 S i t e C 1 78 m Sedge 0.77 0.07 46.62 27.5 33.80 Site A 1 78 m Fall Rye 1.71 0.14 10.13 3.8 9.75 Site B 1 78 m Fall Rye 1.56 0.12 5.95 9.2 5.40 Site C 1 78 m Fall Rye 1.07 0.10 14.50 14.0 12.47 Site A 2 78 m Woolgrass 0.66 0.04 108.92 7.7 100.53 Site B 2 78 m Woolgrass 0.69 0.05 208.06 19.0 168.53 S i t e C 2 78 m Woolgrass 0.67 0.06 116.78 11.1 103.81 Site A 2 78 m Sedge 0.71 0.04 33.49 43.9 18.79 Site B 2 78 m Sedge 0.81 0.05 74.73 18.3 61.06 S i t e C 2 78 m Sedge 0.87 0.07 53.55 11.2 47.55 Site A 2 78 m Fall Rye 1.30 0.08 7.12 13.2 6.18 Site B 2 78 m Fall Rye 1.17 0.08 8.59 5.8 8.09 S i t e C 2 78 m Fall Rye 1.44 0.09 5.22 6.0 4.91 Site A 3 78 m Woolgrass 0.77 0.06 72.35 11.4 64.10 Site B 3 78 m Woolgrass 0.81 0.04 86.63 15.5 73.21 S i t e C 3 78 m Woolgrass 0.91 0.05 103.27 29.1 73.22 Site A 3 78 m Sedge 0.85 0.06 20.13 10.6 17.99 Site B 3 78 m Sedge 0.79 0.07 35.97 13.1 31.26 S i t e C 3 78 m Sedge 0.66 0.06 43.62 16.8 36.29 Site A 3 78 m Fall Rye 1.32 0.09 7.88 9.2 7.15 136 Site Time Elevation (m) Species Total N (roots) (%) Total P (roots) (%) * Total Root Weight (g) A S H % Corrected Root Weight (g) Site B 3 78 m Fall Rye 1.12 0.15 11.42 6.4 10.69 S i t e C 3 78 m Fall Rye 1.72 0.15 13.58 5.7 12.81 Site A 1 76 m Woolgrass 0.90 0.07 90.27 7.8 83.23 Site B 1 76 m Woolgrass 0.84 0.08 94.44 15.5 94.44 S i t e C 1 76 m Woolgrass 0.86 0.08 187.66 17.7 154.44 Site A 1 76 m Sedge 0.85 0.07 33.07 21.9 25.83 Site B 1 76 m Sedge 0.88 0.09 77.56 13.9 66.78 S i t e C 1 76 m Sedge 0.84 0.07 105.53 15.8 88.85 Site A 1 76 m Fall Rye 1.49 0.12 7.25 8.8 6.62 Site B 1 76 m Fall Rye 1.59 0.15 5.73 20.5 4.55 Site C 1 76 m Fall Rye 1.44 0.12 10.79 20.9 8.53 Site A 2 76 m Woolgrass 0.86 0.07 132.98 7.3 123.27 Site B 2 76 m Woolgrass 0.70 0.05 120.79 41.6 70.54 S i t e C 2 76 m Woolgrass 0.71 0.08 87.03 29.9 61.01 Site A 2 76 m Sedge 0.79 0.04 41.86 4.7 39.89 Site B 2 76 m Sedge 0.85 0.07 71.95 8.6 65.76 S i t e C 2 76 m Sedge 0.63 0.07 102.20 7.0 95.05 Site A 2 76 m Fall Rye 1.26 0.07 2.30 11.5 2.04 Site B 2 76 m Fall Rye 1.51 0.10 5.52 13.1 4.79 S i t e C 2 76 m Fall Rye 1.37 0.10 6.78 27.0 4.95 Site A 3 76 m Woolgrass 0.71 0.04 26.42 18.1 21.64 Site B 3 76 m Woolgrass 0.87 0.07 41.88 7.9 38.58 S i t e C 3 76 m Woolgrass 0.95 0.06 22.88 12.4 20.04 Site A 3 76 m Sedge 0.71 0.07 22.39 7.5 20.71 Site B 3 76 m Sedge 0.63 0.04 43.11 4.2 41.30 S i t e C 3 76 m Sedge 0.58 0.04 76.17 14.9 64.82 Site A 3 76 m Fall Rye 1.10 0.12 10.09 14.2 8.66 Site B 3 76 m Fall Rye 1.23 0.13 28.16 5.2 26.70 S i t e C 3 76 m Fall Rye 1.44 0.13 15.12 7.5 13.99 * Total Weight = sum of measured weights *2 to accoun t for subsampling 137 Appendix B. Complete ANOVA tables for Plant Analyses Table B-1. Complete ANOVA table for foliar N (%) arcsine transformed) content. Post inundation data only. Source SS MS Num DF Num *F Ratio Prob > F Site&Random 0.00142 0.00071 2 Elevation 0.00008 0.00004 2 0.3387 0.7314 Species 0.00114 0.00057 2 *29.9852 0.0039 Site*Elevation&Random 0.00045 0.00011 4 0.7311 0.5955 Site*Species&Random 0.00008 0.00002 4 0.1226 0.9703 Elevation*Species 0.00056 0.00014 4 0.8995 0.5073 Site*Elevation*Species&Random 0.00123 0.00015 8 0.5912 0.7785 Time 0.00003 0.00001 2 0.0505 0.9508 Elevation*Time 0.00029 0.00007 4 0.2732 0.8933 Species*Time 0.00109 0.00027 4 1.0435 0.3984 Elevation*Species*Time 0.00167 0.00021 8 0.8014 0.6053 * Indicates a significant F-test result Table B-2. Complete ANOVA table for root N (%) arcsine transformed) content. Post inundation data only. Source SS MS Num DF Num *F Ratio Prob > F Site&Random 1.36e-6 6.8e-7 2 Elevation 0.00006 0.00003 2 2.0015 0.2498 Species 0.01536 0.00768 2 *1065.902 <0001 Site*Elevation&Random 0.00006 0.00001 4 0.2887 0.8773 Site*Species&Random 0.00003 7.2e-6 4 0.1460 0.9597 Elevation*Species 0.00006 0.00002 4 0.3212 0.8562 Site*Elevation*Species&Random 0.00039 0.00005 8 1.1601 0.3493 Time 0.00026 0.00013 2 3.0185 0.0614 ElevationTime 0.00022 0.00005 4 1.2900 0.2922 Species*Time 0.00025 0.00006 4 1.4777 0.2293 Elevation*Species*Time 0.00027 0.00003 8 0.8014 0.6054 * Indicates a significant F-test resu t Table B-3. Complete ANOVA table for Foliar P (% arcsine transformed) content. Post inundation data only. Source SS MS Num DF Num *F Ratio Prob > F Site&Random 0.00837 0.00419 2 5.2094 0.4254 Elevation 0.001 0.0005 2 0.2916 0.7617 Species 0.00925 0.00463 2 *7.6066 0.0433 Site*Elevation&Random 0.00684 0.00171 4 1.1289 0.4079 Site*Species&Random 0.00243 0.00061 4 0.4015 0.8028 Elevation*Species 0.0079 0.00197 4 1.3035 0.3464 Site*Elevation*Species&Random 0.01212 0.00151 8 0.5466 0.8134 Time 0.00338 0.00169 2 0.6106 0.5486 ElevationTime 0.01072 0.00268 4 0.9666 0.4377 Species*Time 0.00581 0.00145 4 0.5236 0.7190 Elevation*Species*Time 0.03487 0.00436 8 1.5726 0.1676 * Indicates a significant F-test result Table B-4. Complete ANOVA table for Root P (% arcsine transformed) content. Post inundation data only. Source SS MS Num DF Num *F Ratio Prob > F Site&Random 0.00245 0.00123 2 0.6662 0.5536 Elevation 0.00097 0.00049 2 0.3323 0.7353 Species 0.11744 0.05872 2 *84.6993 0.0005 Site*Elevation&Random 0.00586 0.00146 4 *4.6001 0.0319 Site*Species&Random 0.00277 0.00069 4 2.1771 0.1621 Elevation*Species 0.00266 0.00067 4 2.0910 0.1739 Site*Elevation*Species&Random 0.00255 0.00032 8 0.3250 0.9510 Time 0.0204 0.0102 2 *10.4094 0.0003 Elevation*Time 0.00361 0.0009 4 0.9209 0.4625 Spec iesT ime 0.00472 0.00118 4 1.2045 0.3258 Elevation*Species*Time 0.01041 0.0013 8 1.3276 0.2616 * Indicates a significant F-test resu t Table B-5. Complete ANOVA table for Foliar N (% arcsine transformed) content. Pre and post- inundation data included. Source Nparm DF Sum of Squares *F Ratio Prob > F species 2 2 0.00095220 2.4193 0.0950 time 3 3 0.01375307 *23.2952 <0001 species*time 6 6 0.00823963 *6.9782 <0001 * Indicates a significant F-test result Table B-6. Complete ANOVA table for Root N (% arcsine transformed) content. Pre and post- inundation data included. Source Nparm DF Sum of Squares *F Ratio Prob > F species 2 2 0.01858236 *200.0152 < 0001 time 3 3 0.00058900 •4.2265 0.0077 species*time 6 6 0.00036560 1.3117 0.2605 * Indicates a significant F-tes t result Table B-7. Complete ANOVA table for N% Root to Shoot ratio. Pre and post- inundation data included. Source Nparm DF Sum of Squares *F Ratio Prob > F Species 2 2 1.2516927 •64.1374 <0001 Time 3 3 0.8679150 •29.6483 <0001 Species*Time 6 6 0.3320085 •5.6708 <0001 * Indicates a significant F-test result Table B-8. Complete ANOVA table for Foliar P (% arcsine transformed) content. Pre and post- inundation data included. Source Nparm DF Sum of Squares *F Ratio Prob > F species 2 2 0.00149987 •35.8944 <0001 time 3 3 0.00392581 •62.6343 <0001 species*time 6 6 0.00570926 •45.5441 <0001 * Indicates a significant F-test result Table B-9. Complete ANOVA table for Root P (% arcsine transformed) content. Pre and post- inundation data included. Source Nparm DF Sum of Squares *F Ratio Prob > F species 2 2 0.00306714 •147.4960 < 0001 time 3 3 0.00082280 •26.3785 <0001 species*time 6 6 0.00095123 •15.2480 <0001 * Indicates a significant F-tes t result Table B-10. Complete ANOVA table for P root to shoot ratio (arcsine transformed). Pre and post- inundation data included. Source Nparm DF Sum of Squares *F Ratio Prob > F Species 2 2 0.86373880 •39.5699 < 0001 Time 3 3 0.48325666 •14.7594 < 0001 Species*Time 6 6 0.35687448 •5.4497 <0001 * Indicates a significant F-test result Appendix C. Raw Data for Calcium, Magnesium and Potassium in Above and Belowground Plant Biomass Table C-1. Aboveground Plant Calcium, Magnesium and Potassium Raw Data Site Time Elevation (m) Species Total C a (foliage) (%) Total Mg (foliage) (%) Total K (foliage) (%) LOI (%) pre 0 Woolgrass 0.28 0.060 1.32 94.6 pre 0 Woolgrass 0.27 0.054 1.19 94.9 pre 0 Woolgrass 0.27 0.058 1.27 94.7 pre 0 Woolgrass 0.31 0.069 1.04 94.2 pre 0 Woolgrass 0.31 0.054 0.88 90.3 pre 0 Sedge 0.29 0.07 1.18 94.55 pre 0 Sedge 0.34 0.06 1.30 93.50 pre 0 Sedge 0.42 0.06 1.31 94.90 pre 0 Sedge 0.68 0.10 1.84 92.70 pre 0 Sedge 0.34 0.07 1.15 94.95 pre 0 Sedge 0.32 0.06 1.31 94.60 pre 0 Fall rye 0.57 0.239 3.22 88.8 pre 0 Fall rye 0.57 0.188 2.60 90.1 pre 0 Fall rye 0.77 0.228 2.52 90.9 pre 0 Fall rye 0.64 0.212 2.50 90.6 pre 0 Fall rye 0.57 0.275 2.79 88.6 pre 0 Fall rye 0.50 0.219 3.08 85.2 pre 0 Fall rye 0.50 0.261 2.76 90.9 C 1 76 Woolgrass 0.34 0.084 1.23 94.0 C 1 78 Woolgrass 0.46 0.069 0.80 90.4 C 1 80 Woolgrass 0.34 0.077 0.88 90.6 C 1 76 Sedge 0.53 0.069 0.84 90.6 C 1 78 Sedge 0.50 0.065 0.85 88.1 C 1 80 Sedge 0.38 0.058 0.81 89.6 C 1 76 Fall rye 0.38 0.026 0.15 49.0 C 1 78 Fall rye 0.38 0.049 0.15 59.7 C 1 80 Fall rye 0.41 0.041 0.15 51.5 B 1 76 Woolgrass 0.42 0.077 0.88 83.2 B 1 78 Woolgrass 0.27 0.061 0.80 87.9 B 1 80 Woolgrass 0.50 0.069 0.91 89.6 B 1 76 Sedge 0.46 0.061 0.84 86.8 B 1 78 Sedge 0.38 0.061 0.91 88.0 B 1 80 Sedge 0.46 0.061 0.76 81.8 B 1 76 Fall rye 0.38 0.091 0.19 45.3 B 1 78 Fall rye 0.49 0.091 0.19 48.0 Site Time Elevation (m) Species Total Ca (foliage) (%) Total Mg (foliage) (%) Total K (foliage) (%) LOI (%) B 1 80 Fall rye 0.42 0.038 0.15 68.0 A 1 76 Woolgrass 0.34 0.061 0.92 61.4 A 1 78 Woolgrass 0.31 0.085 1.34 90.6 A 1 80 Woolgrass 0.42 0.069 0.80 89.6 A 1 76 Sedge 0.46 0.073 0.73 85.3 A 1 78 Sedge 0.63 0.120 1.45 85.5 A 1 80 Sedge 0.54 0.069 0.81 83.7 A 1 76 Fall rye 0.46 0.141 0.19 73.2 A 1 78 Fall rye 0.39 0.142 0.20 73.4 A 1 80 Fall rye 0.45 0.053 0.15 59.6 C 2 76 Woolgrass 0.38 0.073 0.88 87.6 C 2 78 Woolgrass 0.46 0.080 0.80 74.9 C 2 80 Woolgrass 0.31 0.069 1.19 87.0 C 2 76 Sedge 0.46 0.057 0.38 65.4 C 2 78 Sedge 0.91 0.143 1.06 71.8 C 2 80 Sedge 0.50 0.088 0.69 69.7 C 2 76 Fall rye 0.55 0.090 0.12 62.8 C 2 78 Fall rye 0.50 0.120 0.16 52.3 C 2 80 Fall rye 0.43 0.062 0.08 67.9 B 2 76 Woolgrass 0.44 0.072 0.72 86.8 B 2 78 Woolgrass 0.67 0.103 0.83 82.4 B 2 80 Woolgrass 0.64 0.076 0.95 82.7 B 2 76 Sedge 1.30 0.191 0.53 67.4 B 2 78 Sedge 1.07 0.229 0.91 42.9 B 2 80 Sedge 0.64 0.095 0.60 78.4 B 2 76 Fall rye 0.47 0.116 0.12 49.5 B 2 78 Fall rye 0.77 0.170 0.23 35.3 B 2 80 Fall rye 0.43 0.086 0.12 59.9 A 2 76 Woolgrass 0.52 0.079 0.79 88.7 A 2 78 Woolgrass 0.55 0.114 1.06 81.7 A 2 80 Woolgrass 0.63 0.059 0.67 74.2 A 2 76 Sedge 0.67 0.114 0.28 78.3 A 2 78 Sedge 0.58 0.117 0.16 71.4 A 2 80 Sedge 0.62 0.074 0.51 71.0 A 2 76 Fall rye 0.75 0.135 0.12 80.3 A 2 78 Fall rye 0.47 0.144 0.16 73.0 A 2 80 Fall rye 0.51 0.047 0.09 65.8 C 3 76 Woolgrass 0.63 0.078 0.47 82.1 C 3 78 Woolgrass 0.67 0.075 0.87 87.2 Site Time Elevation (m) Species Total C a (foliage) (%) Total Mg (foliage) (%) Total K (foliage) (%) LOI (%) C 3 80 Woolgrass 0.20 0.047 0.55 86.9 C 3 76 Sedge 0.39 0.081 0.12 60.6 c 3 78 Sedge 0.47 0.101 0.23 55.8 c 3 80 Sedge 0.47 0.082 0.55 72.3 c 3 76 Fall rye 0.50 0.073 0.08 61.6 c 3 78 Fall rye 0.42 0.104 0.12 65.5 c 3 80 Fall rye 0.27 0.039 0.05 83.4 B 3 76 Woolgrass 0.55 0.106 0.59 78.0 B 3 78 Woolgrass 0.62 0.116 0.46 67.6 B 3 80 Woolgrass 0.63 0.091 0.91 87.5 B 3 76 Sedge 0.46 0.108 0.12 59.7 B 3 78 Sedge 0.77 0.116 0.15 57.7 B 3 80 Sedge 0.55 0.090 0.67 81.6 B 3 76 Fall rye 0.61 0.142 0.15 50.7 B 3 78 Fall rye 0.46 0.131 0.12 60.3 B 3 80 Fall rye 0.47 0.074 0.12 68.3 A 3 76 Woolgrass 0.47 0.098 0.70 86.9 A 3 78 Woolgrass 0.74 0.116 0.66 82.3 A 3 80 Woolgrass 0.63 0.079 0.71 83.7 A 3 76 Sedge 1.08 0.255 0.23 78.0 A 3 7 8 Sedge 0.55 0.118 0.24 72.6 A 3 80 Sedge 0.51 0.094 1.42 84.1 A 3 76 Fall rye 0.92 0.246 0.22 84.4 A 3 78 Fall rye 0.43 0.132 0.11 74.2 A 3 80 Fall rye 0.27 0.047 0.06 67.9 Table C-2. Belowground Plant Calcium, Magnesium and Potassium Raw Data Site Time Elevation (m) Species Total C a (roots) (%) Total Mg (roots) (%) Total K (roots) (%) Ash (%) Pre 0 Woolgrass 0.23 0.054 0.35 11.8 Pre 0 Woolgrass 0.23 0.046 0.23 8.2 Pre 0 Woolgrass 0.30 0.057 0.19 21.9 Pre 0 Woolgrass 0.35 0.050 0.19 10.6 Pre 0 Woolgrass 0.23 0.046 0.23 8.3 Pre 0 Sedge 0.26 0.053 0.19 16.4 Pre 0 Sedge 0.23 0.053 0.19 8.6 Pre 0 Sedge 0.23 0.042 0.19 10.1 Pre 0 Sedge 0.23 0.046 0.15 11.1 Pre 0 Sedge 0.23 0.050 0.15 11.9 Pre 0 Sedge 0.19 0.061 0.27 15.4 Pre 0 Fall Rye 1.02 0.227 0.45 32.2 Pre 0 Fall Rye 1.05 0.218 0.53 28.7 Pre 0 Fall Rye 1.13 0.275 0.60 26.7 Pre 0 Fall Rye 1.33 0.281 0.35 27.4 Pre 0 Fall Rye 1.12 0.274 0.42 49.4 Pre 0 Fall Rye 1.43 0.291 0.46 34.8 Pre 0 Fall Rye 1.20 0.375 0.35 44.9 A 1 80 Woolgrass 0.26 0.03 0.74 19.0 A 1 80 Sedge 0.41 0.06 0.79 41.6 A 1 80 Fall rye 0.69 0.12 0.38 8.6 A 1 78 Woolgrass 0.32 0.06 0.92 17.7 A 1 78 Sedge 0.35 0.08 0.80 42.4 A 1 78 Fall rye 0.63 0.09 0.35 8.3 A 1 76 Woolgrass 0.26 0.03 0.54 7.8 A 1 76 Sedge 0.28 0.05 0.67 18.3 A 1 76 Fall rye 0.50 0.06 0.26 5.8 B 1 80 Woolgrass 0.42 0.05 0.63 28.6 B 1 80 Sedge 0.37 0.06 0.71 29.9 B 1 80 Fall rye 0.76 0.10 0.39 7.0 B 1 78 Woolgrass 0.32 0.05 0.59 24.8 B 1 78 Sedge 0.30 0.04 0.60 21.5 B 1 78 Fall rye 0.71 0.10 0.28 5.6 B 1 76 Woolgrass 0.32 0.04 0.60 15.5 B 1 76 Sedge 0.24 0.05 0.80 11.2 B 1 76 Fall rye 0.83 0.11 0.31 6.0 C 1 80 Woolgrass 0.30 0.04 0.81 14.9 Site Time Elevation (m) Species Total C a (roots) (%) Total Mg (roots) (%) Total K (roots) (%) Ash (%) C 1 80 Sedge 0.32 0.04 0.34 18.1 C 1 80 Fall rye 0.64 0.11 0.34 7.5 c 1 78 Woolgrass 0.22 0.04 0.81 9.1 c 1 78 Sedge 0.22 0.05 1.24 18.8 c 1 78 Fall rye 0.74 0.10 0.46 11.5 c 1 76 Woolgrass 0.27 0.04 0.86 18.2 c 1 76 Sedge 0.18 0.03 0.50 10.6 c 1 76 Fall rye 0.58 0.09 0.40 9.2 A 2 80 Woolgrass 0.30 0.03 0.42 14.2 A 2 80 Sedge 0.30 0.03 0.44 15.0 A 2 80 Fall rye 0.50 0.06 0.16 4.4 A 2 78 Woolgrass 0.26 0.03 0.61 28.6 A 2 78 Sedge 0.22 0.03 0.30 13.1 A 2 78 Fall rye 0.50 0.07 0.24 6.4 A 2 76 Woolgrass 0.26 0.04 0.72 12.4 A 2 76 Sedge 0.16 0.03 0.30 4.2 A 2 76 Fall rye 0.65 0.08 0.30 5.2 B 2 80 Woolgrass 0.31 0.05 0.61 29.1 B 2 80 Sedge 0.41 0.05 0.75 29.5 B 2 80 Fall rye 0.84 0.13 0.52 9.0 B 2 78 Woolgrass 0.29 0.05 0.71 13.1 B 2 78 Sedge 0.30 0.04 0.63 16.8 B 2 78 Fall rye 0.65 0.07 0.24 5.7 B 2 76 Woolgrass 0.22 0.03 0.47 11.1 B 2 76 Sedge 0.28 0.04 0.59 14.9 B 2 76 Fall rye 0.56 0.06 0.26 7.5 C 2 80 Woolgrass 0.30 0.03 0.51 20.9 C 2 80 Sedge 0.22 0.02 0.26 6.1 C 2 80 Fall rye 0.50 0.07 0.30 7.0 C 2 78 Woolgrass 0.41 0.06 0.61 21.8 C 2 78 Sedge 0.18 0.03 0.62 7.7 C 2 78 Fall rye 0.48 0.06 0.20 3.8 C 2 76 Woolgrass 0.38 0.05 0.57 23.3 C 2 76 Sedge 0.45 0.04 0.53 21.9 C 2 76 Fall rye 0.62 0.09 0.32 8.8 A 3 80 Woolgrass 0.37 0.08 0.51 27.0 A 3 80 Sedge 0.30 0.06 0.45 24.9 A 3 80 Fall rye 0.78 0.11 0.50 14.5 B 3 80 Woolgrass 0.34 0.03 0.48 11.4 Site Time Elevation (m) Species Total C a (roots) (%) Total Mg (roots) (%) Total K (roots) (%) Ash (%) B 3 80 Sedge 0.36 0.06 0.44 18.4 B 3 80 Fall rye 0.70 0.09 0.30 7.8 C 3 80 Woolgrass 0.23 0.03 0.44 14.3 C 3 80 Sedge 0.20 0.04 0.53 7.3 C 3 .80 Fall rye 0.34 0.04 0.18 4.7 A 3 78 Woolgrass 0.28 0.05 0.67 7.9 A 3 78 Sedge 0.28 0.05 0.40 10.1 A 3 78 Fall rye 0.64 0.08 0.32 9.2 A 3 76 Woolgrass 0.18 0.02 0.22 7.7 A 3 76 Sedge 0.22 0.03 0.22 13.9 A 3 76 Fall rye 0.91 0.15 0.63 20.5 B 3 78 Woolgrass 0.27 0.04 0.45 11.5 B 3 78 Sedge 0.35 0.07 0.43 27.5 B 3 78 Fall rye 0.86 0.16 0.54 14.0 B 3 76 Woolgrass 0.26 0.04 0.46 13.1 B 3 76 Sedge 0.26 0.06 0.38 15.8 B 3 76 Fall rye 0.79 0.13 0.67 20.9 C 3 78 Woolgrass 0.41 0.07 0.81 13.1 C 3 78 Sedge 0.41 0.10 0.72 38.6 C 3 78 Fall rye 0.78 0.16 0.48 11.3 C 3 76 Woolgrass 0.34 0.03 0.30 13.4 C 3 76 Sedge 0.26 0.07 0.39 43.9 C 3 76 Fall rye 0.66 0.12 0.48 13.2 Appendix D. Periphyton Density Summary Data By Sample Period Table D-1. Average periphyton density (cells/cm2) estimates associated with woolgrass and sedge foliage at each elevation on July 6, 2000. Periphyton Taxa Woolgrass 80 m Woolgrass 78 m Woolgrass 76 m Sedge 80 m Sedge 78 m Sedge 76 m Class Bacillariophyta Achnanthes sp. 1 3236.79 1086.79 782.44 0.00 189.43 0.00 Achnanthes. minutissima Kutz. 0.00 0.00 0.00 0.00 0.00 0.00 Asterionella formosa (Hantz) Grun. 0.00 0.00 153.44 0.00 80.64 350.84 Caloneis amphisbaena 27.93 0.00 103.09 0.00 318.49 167.53 Caloneis sp. 0.00 145.27 0.00 0.00 0.00 0.00 Cyclotella stelligera CI. & Grun. 517.90 0.00 287.20 1050.60 2131.60 509.81 Cymbella sp. 1 27.93 0.00 0.00 0.00 0.00 0.00 Cymbella sp. 2 1323.52 48.42 0.00 1057.60 986.68 470.27 Cymbella ventricosa Kutz 562.82 485.56 1035.34 741.67 1048.06 1795.22 Eunotia arcus 544.13 355.77 450.48 1624.19 3676.66 1602.00 Eunotia lunaris var. 1 (Ehr.) Grun. 577.80 0.00 954.32 525.30 1175.09 940.53 Eunotia pectinalis 979.25 1077.70 738.30 1165.11 1906.69 2279.46 Eunotia sp. 0.00 242.12 484.78 518.30 1486.26 852.77 F. acus (formerly Synedra acus Kutz) 310.58 572.08 0.00 0.00 80.64 287.52 F. ulna (formerly S. ulna (Nitz.) Ehr.) 677.91 340.29 1035.96 468.51 1266.65 916.76 F. vaucheriae (Kutz.) Peters 145.04 0.00 0.00 0.00 0.00 0.00 Fragilaria construens (Ehr.) Grun. 0.00 242.12 143.60 0.00 0.00 494.60 Frustrulia rhomboides (Ehr.) De T 3819.72 1168.74 778.81 2628.94 2166.86 1579.46 Frustrulia sp. 4543.58 2982.38 1624.35 4285.85 5305.20 2871.51 Gomphonema sp. 1 791.31 291.87 165.67 931.53 640.04 1410.80 Gyrosigma 245.15 48.42 0.00 0.00 0.00 103.54 Aulicoseira sp. 1592.52 1791.67 2093.27 1098.11 5020.36 1227.50 Navicula sp. 2 946.58 328.64 400.13 0.00 318.49 191.31 Nitzschia sp. 1 145.04 43.26 0.00 133.08 0.00 0.00 Pinnularia sp. 1 416.43 148.51 0.00 252.15 239.89 103.54 Rhopalodia 57.54 0.00 0.00 518.30 238.87 103.54 Surirella sp. 187.61 48.42 0.00 133.08 80.64 0.00 Periphyton Taxa Woolgrass 80 m Woolgrass 78 m Woolgrass 76 m Sedge 80 m Sedge 78 m Sedge 76 m T. flocculosa (Roth.) Kutz 2155.89 1902.24 2631.50 867.74 5617.67 5437.48 Tabellaria fenestrata (Lyngb.) Kutz 1391.63 1915.81 1169.72 684.88 4530.54 4161.43 Class Chlorophyta Ankistrodesmus 57.54 0.00 0.00 0.00 0.00 0.00 Closterium 0.00 0.00 0.00 0.00 0.00 0.00 Cosmarium sp. 384.79 0.00 71.80 126.08 79.62 71.88 cryptosomonas 0.00 113.66 0.00 0.00 0.00 71.88 Desmid sp. 145.04 0.00 0.00 0.00 0.00 0.00 Euastrum dentriculatum (desmid) 0.00 0.00 0.00 0.00 0.00 0.00 kephrion 72.52 56.83 0.00 0.00 0.00 0.00 microflagellate 0.00 0.00 153.44 133.08 350.71 71.88 Mougeotia sp. 1 435.12 0.00 0.00 0.00 806.40 0.00 Mougeotia sp. 2 0.00 0.00 0.00 0.00 0.00 0.00 Oedegonium sp. 0.00 56.83 0.00 0.00 0.00 0.00 oocystis 0.00 129.79 0.00 0.00 189.43 0.00 Rhizoclonium sp. 0.00 0.00 0.00 0.00 0.00 0.00 scenedesmus 0.00 0.00 0.00 0.00 348.68 191.31 Spyrogyra sp. 251.34 0.00 0.00 0.00 0.00 0.00 tetrahedron 0.00 48.42 0.00 0.00 0.00 0.00 Ulothrix sp. 4351.16 3460.97 187.74 5043.00 0.00 0.00 Zygnema sp. 572.06 0.00 0.00 0.00 0.00 0:00 Class Cyanophyta Lyngbya 558.53 0.00 0.00 0.00 0.00 0.00 merismopedia 0.00 0.00 0.00 0.00 80.64 0.00 microcystis-like 0.00 0.00 0.00 0.00 0.00 0.00 Oscillatoria limnetica 9446.27 4988.80 3019.61 17685.58 21544.58 3188.22 Others 0.00 0.00 0.00 0.00 0.00 0.00 Table D-2. Average periphyton density (cells/cm2) estimates by associated with woolgrass and sedge foliage at each elevation on August 9, 2000. Periphyton Taxa Woolgras s 80 m Woolgrass 78 m Woolgrass 76 m Sedge 80 m Sedge 78 m Sedge 76 m Class Bacillariophyta Achnanthes minutissima Kutz. 989.62 539.72 3548.71 1838.46 2106.12 6839.29 Achnanthes sp. 1 0.00 0.00 0.00 0.00 0.00 0.00 Asterionella formosa (Hantz) Grun. 0.00 609.61 0.00 0.00 0.00 0.00 Caloneis amphisbaena 0.00 1312.21 0.00 382.90 283.20 645.72 Caloneis sp. 0.00 249.79 0.00 0.00 0.00 159.58 Cyclotella stelligera CI. & Grun. 0.00 179.91 0.00 765.80 2106.12 797.04 Cymbella sp. 1 0.00 0.00 0.00 0.00 0.00 0.00 Cymbella sp. 2 526.36 0.00 0.00 7445.12 3008.89 3483.30 Cymbella ventricosa Kutz 891.36 4406.48 3972.71 2505.27 3175.99 2399.38 Eunotia arcus 1576.70 1718.79 7043.51 5667.48 10796.41 2598.64 Eunotia lunaris var. 1 (Ehr.) Grun. 263.18 999.16 3042.76 996.61 2757.15 2280.33 Eunotia pectinalis 2988.47 4133.57 9578.00 3917.61 10292.94 6305.98 Eunotia sp. 0.00 1968.58 6085.51 1412.34 2139.74 3379.93 F. construens (Ehr.) Grun. 203.65 2264.61 2998.25 0.00 1069.87 0.00 F. ulna (formerly S. ulna (Nitz.) Ehr.) 0.00 3983.40 6441.59 930.44 1939.01 2009.12 F. vaucheriae (Kutz.) Peters 0.00 0.00 0.00 0.00 0.00 0.00 Fragilaria acus (formerly Synedra acus Kutz) 890.17 1132.30 2480.59 0.00 0.00 1451.02 Frustrulia rhomboides (Ehr.) De T 2359.09 2381.25 9770.06 3435.71 14841.54 8537.52 Frustrulia sp. 4990.89 14870.64 13592.90 14050.77 43871.43 18365.63 Gomphonema sp. 1 785.97 2031.84 9083.77 2812.13 6414.91 4408.50 Gyrosigma 0.00 882.51 0.00 0.00 0.00 645.72 Aulicoseira sp. 670.48 22616.33 4190.50 1534.28 14742.82 1675.15 Navicula sp. 2 989.62 2874.21 0.00 0.00 534.94 0.00 Nitzschia sp. 1 261.99 1811.79 370.10 0.00 0.00 0.00 Pinnularia sp. 1 1149.78 249.79 0.00 689.76 0.00 398.52 Rhopalodia 0.00 359.82 1850.51 1227.43 818.14 916.94 Surirella sp. 0.00 882.51 0.00 382.90 0.00 645.72 T. flocculosa (Roth.) Kutz 1149.78 5765.98 15265.37 3917.61 7370.85 8977.43 Periphyton Taxa Woolgras s 80 m Woolgrass 78 m Woolgrass 76 m Sedge 80 m Sedge 78 m Sedge 76 m Tabellaria fenestrata (Lyngb.) Kutz 523.98 7277.43 9601.43 2844.95 10932.05 13936.53 Class Chlorophyta Ankistrodesmus 0.00 0.00 0.00 0.00 0.00 0.00 Closterium 0.00 0.00 0.00 0.00 0.00 0.00 Cosmarium sp. 0.00 882.51 370.10 0.00 0.00 844.98 cryptosomonas 0.00 0.00 0.00 0.00 0.00 0.00 Desmid 0.00 0.00 0.00 0.00 0.00 0.00 Euastrum dentriculatum 0.00 0.00 0.00 0.00 0.00 0.00 kephrion 0.00 0.00 0.00 0.00 0.00 0.00 microflagellate 263.18 882.51 370.10 798.62 702.04 0.00 Mougeotia sp. 1 1353.43 7128.21 0.00 0.00 0.00 20336.53 Mougeotia sp. 2 0.00 0.00 0.00 0.00 0.00 0.00 Oedegonium sp. 0.00 0.00 0.00 0.00 0.00 0.00 oocystis 261.99 1242.33 0.00 207.86 0.00 1291.44 Rhizoclonium sp. 0.00 0.00 0.00 0.00 0.00 0.00 scenedesmus 0.00 249.79 0.00 0.00 0.00 199.26 Spyrogyra sp. 0.00 719.63 0.00 0.00 0.00 0.00 tetrahedron 261.99 0.00 0.00 0.00 0.00 0.00 Ulothrix sp. 0.00 0.00 8760.51 0.00 0.00 0.00 Zygnema sp. 0.00 7493.70 0.00 0.00 0.00 0.00 Class Cyanophyta Lyngbya 0.00 0.00 0.00 0.00 0.00 0.00 merismopedia 0.00 0.00 0.00 0.00 0.00 0.00 microcystis-like 0.00 0.00 0.00 0.00 0.00 0.00 Oscillatoria limnetica 25041.61 50732.05 81388.24 18138.60 62986.93 19811.03 Others 0.00 0.00 0.00 0.00 0.00 0.00 Table D-3. Average periphyton density (cells/cm2) estimates associated with woolgrass and sedge foliage on August 28, 2000 for 80 m samples and September 8, 2000 for 76 m and 78 m samples. Periphyton Taxa Woolgrass Woolgrass Woolgrass Sedge Sedge Sedge 80 m 78 m 76 m 80 m 78 m 76 m Class Bacillariophyta Achnanthes minutissima Kutz. 2544.16 9173.75 36171.59 2415.79 17385.15 15961.92 Achnanthes sp. 1 0.00 0.00 0.00 0.00 201.30 0.00 Asterionella formosa (Hantz) Grun. 0.00 0.00 0.00 0.00 0.00 0.00 Caloneis amphisbaena 0.00 268.34 496.85 662.58 1043.01 506.46 Caloneis sp. 0.00 1384.39 0.00 0.00 0.00 0.00 Cyclotella stelligera CI. & Grun. 0.00 3112.51 822.13 0.00 1843.06 2347.73 Cymbella sp. 1 0.00 0.00 0.00 0.00 0.00 0.00 Cymbella sp. 2 0.00 0.00 0.00 0.00 7049.16 5541.49 Cymbella ventricosa Kutz 0.00 7189.67 3825.84 1366.19 7046.58 2799.61 Eunotia arcus 0.00 3455.31 8454.39 4600.91 13280.82 9322.64 Eunotia lunaris var. 1 (Ehr.) Grun. 0.00 268.34 1410.60 1987.74 201.30 876.08 Eunotia pectinalis 2182.47 3649.08 3905.81 1246.98 3332.89 9770.05 Eunotia sp. 0.00 10872.39 796.66 1444.36 8724.85 9495.86 F. acus (formerly Synedra acus Kutz) 1086.53 193.16 102.77 662.58 645.56 0.00 F. ulna (formerly S. ulna (Nitz.) Ehr.) 0.00 3412.30 2395.82 584.40 1928.94 351.43 F. vaucheriae (Kutz.) Peters 0.00 0.00 0.00 0.00 0.00 0.00 Fragilaria construens (Ehr.) Grun. 364.41 268.34 205.53 0.00 805.20 488.26 Frustrulia rhomboides (Ehr.) De T 2166.37 10881.45 3060.67 0.00 17418.34 13326.04 Frustrulia sp. 2530.77 24961.12 12851.52 5769.72 55041.61 28136.15 Gomphonema sp. 1 722.12 4195.97 5185.52 4213.89 5593.25 5195.07 Gyrosigma 0.00 482.85 456.87 0.00 0.00 0.00 Aulicoseira sp. 0.00 9444.88 3278.07 0.00 12326.11 4067.99 Navicula sp. 2 728.81 6739.21 1978.72 0.00 645.56 0.00 Nitzschia sp. 1 728.81 1234.14 197.04 584.40 444.26 244.13 Pinnularia sp. 1 364.41 805.02 197.04 662.58 802.63 0.00 Rhopalodia 357.72 289.69 867.93 3908.92 198.73 244.13 Surirella sp. 360.44 0.00 456.87 0.00 1045.58 262.32 T : flocculosa (Roth.) Kutz 2186.44 6030.72 3760.40 3234.73 9638.36 10553.87 Periphyton Taxa Woolgrass 80 m Woolgrass 78 m Woolgrass 76 m Sedge 80 m Sedge 78 m Sedge 76 m Tabellaria fenestrata (Lyngb.) Kutz 1808.65 5763.42 2467.09 2415.79 7679.26 10269.64 Class Chlorophyta Ankistrodesmus 0.00 0.00 0.00 0.00 0.00 0.00 Closterium 0.00 0.00 0.00 0.00 0.00 0.00 Cosmarium sp. 364.41 579.48 0.00 1950.59 444.26 506.46 cryptosomonas 0.00 0.00 0.00 0.00 0.00 524.65 Desmid sp. 0.00 0.00 0.00 0.00 0.00 0.00 Euastrum dentriculatum (desmid) 0.00 0.00 0.00 0.00 0.00 0.00 kephrion 0.00 0.00 0.00 0.00 0.00 0.00 microflagellate 0.00 268.34 653.92 662.58 201.30 0.00 Mougeotia sp. 1 0.00 0.00 0.00 3506.43 397.45 0.00 Mougeotia sp. 2 0.00 0.00 0.00 0.00 0.00 0.00 Oedegonium sp. 0.00 0.00 0.00 0.00 0.00 0.00 oocystis 0.00 2006.46 102.77 0.00 888.51 0.00 Rhizoclonium sp. 0.00 0.00 0.00 0.00 0.00 0.00 scenedesmus 0.00 0.00 0.00 0.00 0.00 0.00 Spyrogyra sp. 0.00 805.02 0.00 0.00 0.00 0.00 tetrahedron 0.00 461.50 0.00 0.00 794.90 0.00 Ulothrix sp. 0.00 0.00 0.00 0.00 2818.20 0.00 Zygnema sp. 0.00 0.00 0.00 0.00 0.00 0.00 Class Cyanophyta Oscillatoria limnetica 0.00 0.00 0.00 0.00 0.00 0.00 merismopedia 360.44 0.00 0.00 0.00 0.00 0.00 Lyngbya 0.00 0.00 0.00 0.00 0.00 0.00 microcystis-like 17438.02 41108.19 25983.50 25014.04 17698.66 28302.29 Others 0.00 0.00 0.00 0.00 0.00 0.00 to Appendix E. Periphyton Biovolume Summary Data By Sample Period Table E-1. Average periphyton biovolume (Dm3/cm2) estimates associated with woolgrass and sedge foliage at each elevation on July 6, 2000. Periphyton Taxa biovolume/ Woolgrass Woolgrass Woolgrass Sedge Sedge Sedge cell (um3) 80 m 78 m 76 m 80 m 78 m 76 m (if known) Class Bacillariophyta Achnanthes sp. 1 50 97103.80 32603.56 23473.07 0.00 5682.85 0.00 Achnanthes. minutissima Kutz. 30 0.00 0.00 0.00 0.00 0.00 0.00 Asterionella formosa (Hantz) Grun. 120 0.00 0.00 18412.88 0.00 9676.78 42100.92 Caloneis amphisbaena 430 12008.49 0.00 44328.26 0.00 136952.83 72039.64 Caloneis sp. 430 0.00 62466.19 0.00 0.00 0.00 0.00 Cyclotella stelligera CI. & Grun. 50 25894.90 0.00 14359.94 52530.07 106579.75 25490.67 Cymbella sp. 1 400 11170.69 0.00 0.00 0.00 0.00 0.00 Cymbella sp. 2 400 529406.77 19369.36 0.00 423040.58 394672.73 188106.99 Cymbella ventricosa Kutz 1000 562823.07 485560.58 1035339.58 741665.28 1048059.72 1795215.00 Eunotia arcus 50 27206.69 17788.70 22524.03 81209.54 183832.95 80099.79 Eunotia lunaris var. 1 (Ehr.) Grun. 500 288899.06 0.00 477159.37 262650.33 587547.06 470267.49 Eunotia pectinalis 700 685474.61 754389.73 516809.69 815578.74 1334683.74 1595619.82 Eunotia sp. 6000 0.00 1452702.13 2908654.11 3109803.28 8917573.31 5116607.58 F. acus (formerly Synedra acus Kutz) 2000 621150.94 1144169.86 0.00 0.00 161279.63 575044.33 F. ulna (formerly S. ulna (Nitz.) Ehr.) 3300 2237107.93 1122958.20 3418667.00 1546098.73 4179954.67 3025314.63 F. vaucheriae (Kutz.) Peters 100 14503.85 0.00 0.00 0.00 0.00 0.00 Fragilaria construens (Ehr.) Grun. 120 0.00 29054.04 17231.93 0.00 0.00 59352.25 Frustrulia rhomboides (Ehr.) De T 3000 11459157.52 3506218.13 2336429.90 7886818.94 6500569.04 4738379.42 Frustrulia sp. 2000 9087162.76 5964753.57 3248699.23 8571698.32 10610393.84 5743022.04 Gomphonema sp. 1 2000 1582618.30 583733.92 331335.04 1863052.75 1280076.26 2821604.91 Gyrosigma 3500 858031.66 169481.92 0.00 0.00 0.00 362389.89 Aulicoseira sp. 90 143326.89 161249.94 188394.53 98829.47 451832.37 110474.59 Navicula sp. 2 700 662608.10 230048.96 280090.30 0.00 . 222946.46 133914.91 Nitzschia sp. 1 2000 290077.04 86524.35 0.00 266150.39 0.00 0.00 Pinnularia sp. 1 5000 2082131.86 742570.47 0.00 1260750.77 1199436.44 517699.84 Rhopalodia 5000 287721.07 0.00 0.00 2591502.73 1194356.04 517699.84 Surirella sp. 2500 469019.22 121058.51 0.00 332687.99 201599.54 0.00 Periphyton Taxa biovolume/ Woolgrass Woolgrass Woolgrass Sedge Sedge Sedge cell (urn 3) 80 m 78 m 76 m 80 m 78 m 76 m (if known) T. flocculosa (Roth.) Kutz 1000 2155890.84 1902244.11 2631500.51 867740.36 5617667.10 5437480.35 Tabellaria fenestrata (Lyngb.) Kutz 3000 4174887.18 5747431.33 3509153.47 2054638.14 13591618.93 12484287.62 Class Chlorophyta Ankistrodesmus 0.00 0.00 0.00 0.00 0.00 0.00 Closterium 1000 0.00 0.00 0.00 0.00 0.00 0.00 Cosmarium sp. 0.00 0.00 0.00 0.00 0.00 0.00 cryptosomonas 0.00 0.00 0.00 0.00 0.00 0.00 Desmid sp. 0.00 0.00 0.00 0.00 0.00 0.00 Euastrum dentriculatum (desmid) 0.00 0.00 0.00 0.00 0.00 0.00 kephrion 0.00 0.00 0.00 0.00 0.00 0.00 microflagellate 0.00 0.00 0.00 0.00 0.00 0.00 Mougeotia sp. 1 500 217557.78 0.00 0.00 0.00 403199.08 0.00 Mougeotia sp. 2 5000 0.00 0.00 0.00 0.00 0.00 0.00 Oedegonium sp. 0.00 0.00 0.00 0.00 0.00 0.00 oocystis 0.00 0.00 0.00 0.00 0.00 0.00 Rhizoclonium sp. 350 0.00 0.00 0.00 0.00 0.00 0.00 scenedesmus 0.00 0.00 0.00 0.00 0.00 0.00 Spyrogyra sp. 0.00 0.00 0.00 0.00 0.00 0.00 tetrahedron 0.00 0.00 0.00 0.00 0.00 0.00 Ulothrix sp. 170 739696.46 588365.59 31915.06 857310.52 0.00 0.00 Zygnema sp. 0.00 0.00 0.00 0.00 0.00 0.00 Class Cyanophyta Lyngbya 0.00 0.00 0.00 0.00 0.00 0.00 merismopedia 50 0.00 0.00 0.00 0.00 4031.99 0.00 microcystis-like 0.00 0.00 0.00 0.00 0.00 0.00 Oscillatoria limnetica 0.00 0.00 0.00 0.00 0.00 0.00 Others 0.00 0.00 0.00 0.00 0.00 0.00 Table E-2. Average periphyton biovolume (um3/cm2) estimates associated with woolgrass and sedge foliage at each elevation on August 9, 2000. Periphyton Taxa biovolume/ Woolgrass Woolgrass Woolgrass Sedge Sedge Sedge cell (urn3) 80 m 78 m 76 m 80 m 78 m 76 m (if known) Class Bacillariophyta Achnanthes minutissima Kutz. 30 29688.46 16191.71 106461.16 55153.90 63183.53 205178.79 Achnanthes sp. 1 50 0.00 0.00 0.00 0.00 0.00 0.00 Asterionella formosa (Hantz) Grun. 120 0.00 73152.69 0.00 0.00 0.00 0.00 Caloneis amphisbaena 430 0.00 564250.83 0.00 164647.73 121776.05 277660.37 Caloneis sp. 430 0.00 107409.68 0.00 0.00 0.00 68618.75 Cyclotella stelligera CI. & Grun. 50 0.00 8995.39 0.00 38290.17 105305.88 39851.92 Cymbella sp. 1 400 0.00 0.00 0.00 0.00 0.00 0.00 Cymbella sp. 2 400 210544.05 0.00 0.00 2978048.38 1203554.82 1393319.66 Cymbella ventricosa Kutz 1000 891364.87 4406475.30 3972709.60 2505271.11 3175990.09 2399377.04 Eunotia arcus 50 78834.87 85939.57 352175.39 283374.02 539820.49 129931.83 Eunotia lunaris var. 1 (Ehr.) Grun. 500 131590.03 499579.90 1521377.71 498307.48 1378575.49 1140166.85 Eunotia pectinalis 700 2091925.58 2893497.34 6704602.58 2742325.54 7205056.30 4414186.75 Eunotia sp. 6000 0.00 11811487.77 36513064.93 8474020.79 12838468.72 20279553.47 F. construens (Ehr.) Grun. 120 24437.93 271752.80 359790.57 0.00 128384.69 0.00 F. ulna (formerly S. ulna (Nitz.) Ehr.) 3300 0.00 13145213.54 21257249.14 3070450.00 6398748.38 6630095.18 F. vaucheriae (Kutz.) Peters 100 0.00 0.00 0.00 0.00 0.00 0.00 F. acus (formerly Synedra acus Kutz) 2000 1780346.82 2264606.70 4961185.82 0.00 0.00 2902044.16 Frustrulia rhomboides (Ehr.) De T 3000 7077266.53 7143759.28 29310194.93 10307131.53 44524626.29 25612553.19 Frustrulia sp. 2000 9981778.83 29741283.98 27185808.85 28101538.40 87742859.92 36731250.07 Gomphonema sp. 1 2000 1571931.59 4063685.57 18167531.14 5624255.48 12829816.50 8816993.58 Gyrosigma 3500 0.00 3088796.89 0.00 0.00 0.00 2260026.30 Aulicoseira sp. 90 60343.10 2035469.58 377145.37 138085.48 1326854.15 150763.23 Navicula sp. 2 700 692730.63 2011948.69 0.00 0.00 374455.34 0.00 Nitzschia sp. 1 2000 523977.20 3623582.27 740202.41 0.00 0.00 0.00 Pinnularia sp. 1 5000 5748895.46 1248949.75 0.00 3448791.63 0.00 1992595.95 Rhopalodia 5000 0.00 1799078.87 9252530.15 6137132.55 4090681.56 4584678.73 Surirella sp. 2500 0.00 2206283.50 0.00 957254.25 0.00 1614304.50 T. flocculosa (Roth.) Kutz 1000 1149779.09 5765979.76 15265370.54 3917607.91 7370846.62 8977426.08 Periphyton Taxa biovolume/ Woolgrass Woolgrass Woolgrass Sedge Sedge Sedge cell ((.im3) 80 m 78 m 76 m 80 m 78 m 76 m (if known) Tabellaria fenestrata (Lyngb.) Kutz 3000 1571931.59 21832295.16 28804282.23 8534843.66 32796146.21 41809599.02 Class Chlorophyta Ankistrodesmus 0.00 0.00 0.00 0.00 0.00 0.00 Closterium 1000 0.00 0.00 0.00 0.00 0.00 0.00 Cosmarium sp. 0.00 0.00 0.00 0.00 0.00 0.00 cryptosomonas 0.00 0.00 0.00 0.00 0.00 0.00 Desmid 0.00 0.00 0.00 0.00 0.00 0.00 Euastrum dentriculatum 0.00 0.00 0.00 0.00 0.00 0.00 kephrion 0.00 0.00 0.00 0.00 0.00 0.00 microflagellate 0.00 0.00 0.00 0.00 0.00 0.00 Mougeotia sp. 1 500 676714.24 3564105.67 0.00 0.00 0.00 10168266.37 Mougeotia sp. 2 5000 0.00 0.00 0.00 0.00 0.00 0.00 Oedegonium sp. 0.00 0.00 0.00 0.00 0.00 0.00 oocystis 0.00 0.00 0.00 0.00 0.00 0.00 Rhizoclonium sp. 350 0.00 0.00 0.00 0.00 0.00 0.00 scenedesmus 0.00 0.00 0.00 0.00 0.00 0.00 Spyrogyra sp. 0.00 0.00 0.00 0.00 0.00 0.00 tetrahedron 0.00 0.00 0.00 0.00 0.00 0.00 Ulothrix sp. 170 0.00 0.00 1489287.01 0.00 0.00 0.00 Zygnema sp. 0.00 . 0.00 0.00 0.00 0.00 0.00 Class Cyanophyta Lyngbya 0.00 0.00 0.00 0.00 0.00 0.00 merismopedia 50 0.00 0.00 0.00 0.00 0.00 0.00 microcystis-like 0.00 0.00 0.00 0.00 0.00 0.00 Oscillatoria limnetica 0.00 0.00 0.00 0.00 0.00 0.00 Others 0.00 0.00 0.00 0.00 0.00 0.00 Table E-3. Average periphyton biovolume (um3/cm2) estimates associated with woolgrass and sedge foliage on August 28, 2000 for 80 m samples and September 8, 2000 for 76 m and 78 m samples. biovolume/ Woolgrass Woolgrass Woolgrass Sedge Sedge Sedge Periphyton Taxa cell (urn 3) 80 m 78 m 76 m 80 m 78 m 76 m (if known) Class Bacillariophyta Achnanthes minutissima Kutz. 30 76324.78 275212.36 1085147.75 72473.80 521554.53 478857.72 Achnanthes sp. 1 50 0.00 0.00 0.00 0.00 10065.00 0.00 Asterionella formosa (Hantz) Grun. 120 0.00 0.00 0.00 0.00 0.00 0.00 Caloneis amphisbaena 430 0.00 115385.94 213647.39 284909.55 448492.62 217776.27 Caloneis sp. 430 0.00 595287.86 0.00 0.00 0.00 0.00 Cyclotella stelligera CI. & Grun. 50 0.00 155625.57 41106.31 0.00 92152.82 117386.72 Cymbella sp. 1 400 0.00 0.00 0.00 0.00 0.00 0.00 Cymbella sp. 2 400 0.00 0.00 0.00 0.00 2819663.39 2216597.86 Cymbella ventricosa Kutz 1000 0.00 7189674.81 3825843.46 1366187.34 7046583.59 2799613.32 Eunotia arcus 50 0.00 172765.45 422719.27 230045.65 664041.07 466132.12 Eunotia lunaris var. 1 (Ehr.) Grun. 500 0.00 134169.69 705298.54 993870.53 100650.02 438041.91 Eunotia pectinalis 700 1527732.50 2554359.23 2734066.69 872889.26 2333025.24 6839032.63 Eunotia sp. 6000 0.00 65234343.08 4779986.91 8666180.32 52349075.75 56975139.39 F. acus (formerly Synedra acus Kutz) 2000 2173059.77 386318.54 205531.56 1325160.71 1291111.58 0.00 F. ulna (formerly S. ulna (Nitz.) Ehr.) 3300 0.00 11260585.71 7906201.44 1928534.20 6365511.01 1159735.04 F. vaucheriae (Kutz.) Peters 100 0.00 0.00 0.00 0.00 0.00 0.00 Fragilaria construens (Ehr.) Grun. 120 43728.88 32200.73 24663.79 0.00 96624.02 58591.67 Frustrulia rhomboides (Ehr.) De T 3000 6499102.88 32644358.60 9182024.30 0.00 52255007.22 39978130.02 Frustrulia sp. 2000 5061549.94 49922241.65 25703032.68 11539443.31 110083217.52 56272305.94 Gomphonema sp. 1 2000 1444245.09 8391940.29 10371034.00 8427774.96 11186505.59 10390132.91 Gyrosigma 3500 0.00 1689958.81 1599049.71 0.00 0.00 0.00 Aulicoseira sp. 90 0.00 850039.59 295025.90 0.00 1109349.83 366119.26 Navicula sp. 2 700 510170.28 4717447.37 1385105.03 0.00 451889.05 0.00 Nitzschia sp. 1 2000 1457629.36 2468271.45 394088.61 1168808.60 888511.51 488263.91 Pinnularia sp. 1 5000 1822036.71 4025090.81 985221.53 3312901.76 4013126.25 0.00 Rhopalodia 5000 1788576.01 1448430.53 4339672.37 19544575.85 993625.76 1220659.78 Surirella sp. 2500 901095.73 0.00 1142178.36 0.00 2613952.44 655811.20 biovolume/ Woolgrass Woolgrass Woolgrass Sedge Sedge Sedge Periphyton Taxa cell (urn3) 80 m 78 m 76 m 80 m 78 m 76 m (if known) T. flocculosa (Roth.) Kutz 1000 2186444.05 6030719.21 3760398.63 3234725.71 9638360.06 10553865.64 Tabellaria fenestrata (Lyngb.) Kutz 3000 5425957.28 17290255.87 7401265.83 7247379.78 23037794.95 30808932.43 Class Chlorophyta Ankistrodesmus 0.00 0.00 0.00 0.00 0.00 0.00 Closterium 1000 0.00 0.00 0.00 0.00 0.00 0.00 Cosmarium sp. 0.00 0.00 0.00 0.00 0.00 0.00 cryptosomonas 0.00 0.00 0.00 0.00 0.00 0.00 Desmid sp. 0.00 0.00 0.00 0.00 0.00 0.00 Euastrum dentriculatum (desmid) 0.00 0.00 0.00 0.00 0.00 0.00 kephrion 0.00 0.00 0.00 0.00 0.00 0.00 microflagellate 0.00 0.00 0.00 0.00 0.00 0.00 Mougeotia sp. 1 500 0.00 0.00 0.00 1753212.91 198725.15 0.00 Mougeotia sp. 2 5000 0.00 0.00 0.00 0.00 0.00 0.00 Oedegonium sp. 0.00 0.00 0.00 0.00 0.00 0.00 oocystis 0.00 0.00 0.00 0.00 0.00 0.00 Rhizoclonium sp. 350 0.00 0.00 0.00 0.00 0.00 0.00 scenedesmus 0.00 0.00 0.00 0.00 0.00 0.00 Spyrogyra sp. 0.00 0.00 0.00 0.00 0.00 0.00 tetrahedron 0.00 0.00 0.00 0.00 0.00 0.00 Ulothrix sp. 170 0.00 0.00 0.00 0.00 479094.08 0.00 Zygnema sp. 0.00 0.00 0.00 0.00 0.00 0.00 Class Cyanophyta Oscillatoria limnetica 0.00 0.00 0.00 0.00 0.00 0.00 merismopedia 50 18021.91 0.00 0.00 0.00 0.00 0.00 Lyngbya 0.00 0.00 0.00 0.00 0.00 0.00 microcystis-like 0.00 0.00 0.00 0.00 0.00 0.00 Others 0.00 0.00 0.00 0.00 0.00 0.00 1^ 1 oo Appendix F. Complete ANOVA tables for analyses related to periphyton Table F-1. Complete ANOVA table for diatom density (log10(x) transformed data). Source SS MS Num DF Num *F Ratio Prob > F Site&Random 0.01964 0.00982 2 0.0316 0.9691 Species 0.48256 0.48256 1 3.0623 0.2222 Elevation 2.49183 1.24592 2 5.0799 0.0798 Site*Species&Random 0.31516 0.15758 2 1.7017 0.2919 Site*Elevation&Random 0.98105 0.24526 4 2.6485 0.1842 Species*Elevation 0.03018 0.01509 2 0.1630 0.8550 Site*Species*Elevation&Random 0.37041 0.0926 4 0.8649 0.4991 Time 2.78463 1.39231 2 *13.0045 0.0001 Species*Time 0.09222 0.04611 2 0.4307 0.6550 ElevationTime 1.07673 0.26918 4 2.5142 0.0682 Species*Elevation*Time 0.61713 0.15428 4 1.4410 0.2511 * Indicates a significant F-test result Table F-2. Complete ANOVA table for diatom biovolume (square root transformed data). Source SS MS Num DF Num *F Ratio Prob > F Site&Random 1.226e7 6127529 2 0.2369 0.8103 Species 6.282e7 6.282e7 1 4.7981 0.1599 Elevation 2.284e8 1.142e8 2 3.8882 0.1154 Site*Species&Random 2.618e7 1.309e7 2 0.7887 0.5144 Site*Elevation&Random 1.175e8 2.937e7 4 1.7694 0.2970 Species*Elevation 62453.9 31227 2 0.0019 0.9981 Site*Species*Elevation&Random 6.64e+7 1.66e+7 4 1.7633 0.1692 Time 3.362e8 1.681e8 2 *17.8581 <0001 Species*Time 1.142e7 5708169 2 0.6063 0.5535 Elevation*Time 1.33e+8 3.325e7 4 *3.5317 0.0211 Species*Elevation*Time 4.587e7 1.147e7 4 1.2181 0.3292 * Indicates a significant F-test result Table F-3. Complete ANOVA table for diatom species richness (Iog10(x) transformed data). Source SS MS Num DF Num *F Ratio Prob > F Site&Random 0.0052 0.0026 2 0.3298 0.8226 Species 0.00978 0.00978 1 1.4444 0.3524 Elevation 0.40941 0.2047 2 *12.2064 0.0198 Site*Species&Random 0.01354 0.00677 2 0.4325 0.6760 Site*Elevation&Random 0.06708 0.01677 4 1.0716 0.4741 Species*Elevation 0.01783 0.00891 2 0.5696 0.6058 Site*Species*Elevation&Random 0.0626 0.01565 4 0.9172 0.4701 Time 0.0539 0.02695 2 1.5794 0.2268 Species*Time 0.01715 0.00858 2 0.5026 0.6112 Elevation*Time 0.27791 0.06948 4 •4.0718 0.0117 Species*Elevation*Time 0.10603 0.02651 4 1.5535 0.2188 * Indicates a significant F-test result Appendix G: The Nonmetric Multidimensional Scaling Ordination Technique 1.0 Background Ordination is a family of methods widely used in ecology to describe the relationships between species composition patterns and the underlying environmental gradients which influence these patterns. Sample units are arranged along a synthetic scale (axis) or multiple axes, so that complex relationships are graphically summarized. Similar species and samples are placed close together, and dissimilar species and samples are placed far apart. Ordination helps to select the most important factors from multiple factors; separate strong patterns from weak ones; and/or to reveal unforeseen patterns or processes. Ordination techniques are generally classified into indirect and direct gradient analysis procedures. Indirect gradient analysis utilizes only the species by sample matrix (i.e. any information about the environment is used only as an interpretative tool after the indirect gradient analysis has been performed); whereas, direct gradient analysis utilizes external environmental data in addition to the species data, and allows the researcher to test the null hypothesis that species composition is unrelated to measured environmental variables (Palmer 2006). Commonly used indirect gradient analysis ordination techniques include the distance-based Bray-Curtis ordination, Principal Coordinates Analysis, and Nonmetric Multidimensional Scaling; and the Eigenanalysis-based Principal Components Analysis, Correspondence Analysis, Detrended Correspondence Analysis. Commonly used direct gradient analysis techniques include Redundancy Analysis, Canonical Correspondence Analysis, and Detrended Canonical Correspondence Analysis (Palmer 2006). Further detail on the Nonmetric Multidimensional Scaling (NMS), the technique used for this thesis, is provided below. 160 2.0 Nonmetric Multidimensional Scaling NMS is an ordination technique well suited to data that are non-normal, and is generally the most effective ordination methodology for ecological data (McCune and Grace 2002). NMS conducts an iterative search for the best positions of n data points on k axes (dimensions) that minimizes stress (i.e. the relationship between the similarity in species composition and the closeness in ordination space of the reduced /c-dimensional configuration). To achieve the lowest possible stress, samples are repeatedly "rearranged" within the space defined by the requested number of dimensions. After each set of movements, NMS checks how well the distances between objects can be reproduced by each new configuration. The process is repeated until stress appears to reach a minimum. Figure G-1 shows the output plot of stress vs. iteration number produced by PC-ORD, the software used for NMS analyses within this thesis (McCune and Mefford 1999). Note that PC-ORD reports stress as a percentage, whereas many other software packages present stress in terms of a numerical value (C. Perrin Limnotek Research and Development, pers. comm.). Stress levels reported in this thesis have therefore been divided by 100 to conform with standard reporting terminology. Stress will typically decrease as a function of the number of dimensions chosen. The final configuration of points represents the ordination solution. PC-ORD, the software used to perform NMS for this thesis outputs a recommended number of dimensions based on the lowest stress solution available. Stress levels between 0.05 - 0.10 indicate that the ordination is good and there are no real risks of drawing false inferences from the ordination diagrams; 0.10 - 0.20 indicate that the diagrams are fairly reliable for interpretation, although too much reliance should not be placed on the details of the diagram; stress levels > 0.20 are considered to be poor in terms of diagram interpretation (Clarke 1993). The solutions for most ecological community data sets tend to have stress levels between 0.10 - 0.20 (McCune and Grace 2002). After determining the synthetic variables or axes, the next step in NMS is to attempt to relate the synthetic axes to other variables, such as measured environmental variables. One way to do this is using overlays on the ordination diagrams, as discussed below. 161 39.1010475 31.5331028 25.2264822 Stress (%) 18.9198616 12.6132411 k -k -k k -k k 6.3066205 0.0000000 5 10 15 20 25 30 35 I t e r a t i o n number Figure G-1. PC-ORD output plot of stress (expressed as a percentage) vs. iteration number. The final solution is achieved when stress levels no longer decrease appreciably. In this ordination, the lowest stress level was reached after 35 iterations. The plot also indicates that the solution is relatively stable as the curve settles on an even stress level and does not fluctuate erratically. 2.1 Ordination diagrams An ordination diagram is a two-dimensional diagram of one synthetic axis against another. The distance between the points in the ordination space is proportional to the underlying distance measure. In NMS, the order of the axes is arbitrary (i.e. the first axis is not necessarily more important than the second axis, etc.); therefore, one of the first steps is view ordination diagrams of the objects in the various two-dimensional planes, and search for visual patterns. Note that the diagram can be rotated if necessary to provide further clarity. Similar samples are placed closer together in the diagram than dissimilar samples. The final orientation of axes in the plane or space is generally the result of a subjective decision by the researcher to choose the axes orientations that can most easily be interpreted. Figure G-2 shows the three possible two-axis configurations for a three-dimensional ordination solution using benthos abundance 162 data from this thesis. For this example, the axis 3 vs. axis 1 (Figure G - 2b ) configuration appeared to show the strongest patterns and was therefore chosen for interpretation purposes. Note that in most ordination methods, the axes are uncorrelated by definition; therefore, the ordination diagrams should not be used to look for regression-type patterns. Benthos abundance CN • • (a) • • • T m T T T T • • T • T • • * T • • w • — • * • * • * T • • • • • # • • * i i tit m • • 9 T • • t T T • • • Ii • II Species • Woolgrass • Sedge • Fall rye • Control Axis 1 163 Benthos abundance Species • Woolgrass • Sedge a Fall rye • Control Axis 1 164 Benthos abundance CO CO (C) • # ISi W m • 4 # • -• • • Species • Woolgrass • Sedge • Fall rye • Control Axis 2 Figure G-2. A two dimensional ordination, showing axis 2 vs. axis 1 (Fig. G-2a), axis 3 vs. axis 1 (Fig. G-2b), and axis 3 vs. axis 2 (Fig. G-2c), of benthos abundance in species space. Distances between sample units approximate dissimilarity in species composition. Overlay plots are used to look at the distributions of individual species in relation to the factors of interest; in this case of this thesis, plant species, elevation, and time. Species distributions are clearly shown by scaled symbol sizes to reflect the presence or abundance of species. When individual variables are overlaid, each point on the ordination is substituted by a symbol representing the magnitude of the variable. Figure G-3 shows an example of the ordination diagram output for Nematoda in terms of abundance over time. The diagram clearly shows that the species of interest is most abundant at T1 in comparison to T2, and is generally more abundant, with the exception of a few samples, at T1 in comparison to T3. This is apparent due to the larger symbols associated with the samples from T1 and T3, relative to T2. The analysis 165 can be repeated for the same invertebrate species to discern any patterns for that particular species related to elevation or plant species. Nematoda Time • 1 • 2 • 3 • • CO CO X < Axis 1 Figure G-3. Overlay plot with sizes of symbols proportional to the magnitude of the variable, in this case, abundance of Nematoda in relation to time from this thesis. 2.2 Proportion of variance represented NMS calculates an r2 value for each of the synthetic axes in the final ordination solution. The r2 value provides an assessment of how well the distances between points in the ordination diagram represent distances in the original unreduced space, and of how the variance explained is distributed among the axes. Incremental and cumulative values can be calculated for each ordination axis. Figure G-4 shows the P C - O R D output for r2 values for the benthos abundance data from this thesis. 166 R Squared Axi s Increment Cumulative 1 .380 .380 2 .178 .558 3 . 195 .753 Figure G-4. PC-ORD output for coefficients of determination (r2 values) for the correlations between ordination distances and distances in the original n-dimensional space, in this case for the invertebrate abundance data from this thesis 3.0 Conclusion Indirect gradient ordination techniques, such as NMS, are meant to be exploratory tools, and should not be used in hypothesis driven analysis. No null hypotheses can be rejected, nor are p-values generated to test statistical significance (Palmer 2006). It is important to keep in mind that no matter which ordination technique is adopted, the purpose of ordination is to assist a researcher to find pattern in data sets that are otherwise too complicated to interpret. A good ordination technique will be able to identify the most important dimensions in a data set, and ignore the "noise", in order to show these patterns. 4.0 References Clarke, K.R. 1993. Non-parametric multivariate analyses of changes in community structure. Australian Journal of Ecology 18:117-143. McCune, B. and J.B. Grace. 2002. Analysis of Ecological Communities. MjM Software Design. Gleneden Beach, Oregon. McCune, B. and M.J. Mefford. 1999. PC-ORD for Windows. Multivariate analysis of Ecological Data Version 4.14. Palmer, M. 2006. "Ordination Methods for Ecologists" Oklahoma State University Department of Ecology webpage: http://ordination.okstate.edu 167 Appendix H: NMS ordination diagrams for periphyton density analysis Diatom Elevation (m) • 76 • 78 80 Axis 1 Diatom density Species • Woolarass • Sedge Axis 1 Figure H-1. Two-dimensional ordination of diatom density by elevation (m) (Fig. H-1a) and plant species (Fig. H-1b). Distances between sample units approximate dissimilarity in species composition. 168 Diatom density Axis 1 Diatom density Elevation (m) • 76 • 78 80 Axis 1 169 Diatom density Species • Woolarass • Sedge Axis 1 Figure H-2. NMS ordination diagrams for Frustrulia sp., accounting for 22% of all diatoms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of Frustrulia sp. The different symbols in the ordination diagrams represent different sample times (Figure H-2a), elevation (m) (Figure H-2b) or plant species (Figure H-2c). 170 Diatom density Axis 1 Diatom density (b) Elevation (m) • 76 • 78 80 f Stress = 0.137 Axis 1 Diatom density Species • Woolarass • Sedae Axis 1 Figure H-3. NMS ordination diagrams for Achnanthes minutissima, accounting for 9% of all diatoms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of A. minutissima. The different symbols in the ordination diagrams represent different sample times (Figure H-3a), elevation (m) (Figure H-3b) or plant species (Figure H-3c). 172 Diatom density Axis 1 Diatom density Elevation (m) • 76 • 78 80 Axis 1 173 Diatom density Species • Woolqrass • Sedge Axis 1 Figure H-4. NMS ordination diagrams for Frustrulia rhomboides, accounting for 8% of all diatoms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of F. rhomboides. The different symbols in the ordination diagrams represent different sample times (Figure H-4a), elevation (m) (Figure H-4b) or plant species (Figure H-4c). 174 Diatom density (a) , • • Stress = 0.137 Time • 1 • 2 3 Axis 1 Diatom density (b) Elevation (m) • 76 • 78 80 Stress = 0.137 Axis 1 Diatom density Species • Woolqrass • Sedge . • • t • Stress = 0.137 Axis 1 Figure H-5. NMS ordination diagrams for Tabellaria flocculosa, accounting for 8% of all diatoms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of T. flocculosa. The different symbols in the ordination diagrams represent different sample times (Figure H-5a), elevation (m) (Figure H-5b) or plant species (Figure H-5c). 176 Diatom density Time Axis 1 Diatom density Elevation (m) • 76 • 78 • 80 Axis 1 Diatom density Species • Woolgrass • Sedae Axis 1 Figure H-6. NMS ordination diagrams for Tabellaria fenestrata, accounting for 7% of all diatoms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of T. fenestrata. The different symbols in the ordination diagrams represent different sample times (Figure H-6a), elevation (m) (Figure H-6b) or plant species (Figure H-6c). 178 Appendix I: NMS ordination diagrams for diatom biovolume analysis Diatom biovolume CM U3 • • (a) s » * • a • Stress = 0.130 Elevation (m) • 76 • 78 * 80 Axis 1 Diatom biovolume Species • Woolgrass • Sedge Axis 1 Figure 1-1. Two-dimensional ordination of diatom biovolume. Note that there are no clear groupings by elevation (m) (Fig. 1-1 a) or plant species (Fig. 1-1 b). Distances between sample units approximate dissimilarity in species composition. 179 Diatom biovolume (a) • p . a • « • J> * • • • • • .* m • t • • • • Stress = 0.130 Axis 1 Diatom biovolume (b) • • • • • • • • • • « • • • Stress = 0.130 Time • 1 • 2 3 Elevation (m) • 76 • 78 80 Axis 1 Diatom biovolume Species • Woolqrass • Sedqe Axis 1 Figure I-2. NMS ordination diagrams for Frustrulia sp., accounting for 26% of all diatoms by biovolume. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of Frustrulia sp. The different symbols in the ordination diagrams represent different sample times (Figure l-2a), elevation (m) (Figure l-2b) or plant species (Figure l-2c). 181 Diatom biovolume • I (a) t S • • • • • • Stress = 0.130 t 1 • 2 3 Axis 1 Diatom biovolume (b) 4 • * • • • Stress = 0.130 | Elevation (m) • 76 • 78 80 Axis 1 182 Diatom biovolume (c) Species • Woolgrass • Sedge Stress = 0.130 Axis 1 Figure 1-3. NMS ordination diagrams for Frustrulia rhomboides, accounting for 15% of all diatoms by biovolume. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of Frustrulia rhomboides. The different symbols in the ordination diagrams represent different sample times (Figure l-3a), elevation (m) (Figure l-3b) or plant species (Figure l-3c). 183 Diatom biovolume (a) Stress = 0.130 Time • 1 • 2 3 Axis 1 Diatom biovolume (b) • • Stress = 0.130 Elevation (m) • 76 • 78 • 80 Axis 1 184 Diatom biovolume (c) • • * Stress = 0.130 Species • Woolarass • Sedae Axis 1 Figure I-4. NMS ordination diagrams for Eunotia sp., accounting for 15% of all diatoms by biovolume. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of Eunotia sp. The different symbols in the ordination diagrams represent different sample times (Figure l-4a), elevation (m) (Figure l-4b) or plant species (Figure l-4c). 185 Diatom biovolume (a) • ' i * • • » • t • • • • • * Stress = 0.130 Axis 1 Diatom biovolume •|» • (b) •* » . • • • • * % " i • • • • • • • Stress = 0.130 Axis 1 Time • 1 • 2 3 Elevation (m) • 76 • 78 1 80 186 Diatom biovolume • • (c) Species • Woolarass • Sedae Stress = 0.130 Axis 1 Figure 1-5. NMS ordination diagrams for Tabellaria fenestrata, accounting for 14% of all diatoms by biovolume. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of T. fenestrata. The different symbols in the ordination diagrams represent different sample times (Figure l-5a), elevation (m) (Figure l-5b) or plant species (Figure I-5c). 187 Appendix J: Benthos Summary Data oo oo Table J-1. "Number of benthic organisms associated with woolgrass, sedge, fall rye and control samples at each elevation over time. Data has been averaged over the three sample locations. Elevation (m) : : Plant ... Spec|es . substrate 06-Jul-00 09-Aug-00 08-Sep-00 06-Jul-00 09-Aug-00 06-Jul-OO 09-Aug-OO 08-Sep-00 06-Jul-OO 09-Aug-OO 08-Sep-00 06-Jul-OO 09-Aug-OO 08-Sep-00 06-Jul-OO 09-Aug-OO 06-Jul-OO 09-Aug-OO 06-Jul-OO 09-Aug-OO 08-Sep-00 06-Jul-OO 09-Aug-OO 06-Jul-OO 09-Aug-OO 08-Sep-00 06-Jul-OO 09-Aug-OO 08-Sep-00 06-Jul-OO 09-Aug-OO 06-Jul-OO 09-Aug-OO 06-Jul-OO 09-Aug-OO 08-Sep-00 06-Jul-OO 09-Aug-OO 06-Jul-OO 09-Aug-OO 08-Sep-00 06-Jul-OO 80 Woolgrass leaves "Taxa Code s1 s2: s3 s4 s5 2.0 2.0 0.0 0.0 1.3 0.0 16.7 2.7 0.7 0.7 1.3 0.0 0.0 0.0 1.3 0.0 0.0 2.7 0.7 0.0 0.0 0.0 0.0 0.0 1.3 0.7 18.7 76.0 s9. . s10 s11 s12. s13. s14 s1S s16 s17 s18 s19 s20 s21 s22 s23 s24 s25s26 ,s27 s28 s29 80 Woolgrass leaves 2.0 80.0 0.0 0.0 0.0 0.0 37.3 10.0 10.0 0.7 3.3 0.0 7.3 0.0 20.7 0.0 0.0 4.7 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 6.0 127.3 0.7 80 Woolgrass leaves 0.7 5.3 0.0 0.0 0.0 1.3 1.3 2.0 0.0 0.7 0.0 8.0 1.3 0.0 2.0 0.0 0.0 2.7 2.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.7 39.3 0.0 78 Woolgrass leaves 0.7 10.0 0.0 0.0 1.3 2.7 8.7 0.0 5.3 0.0 0.0 0.7 1.3 0.7 2.0 0.7 0.0 5.3 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 4.0 127.3 1.3 78 Woolgrass leaves 16.7 92.0 0.0 0.0 11.3 0.7 57.3 8.7 33.3 6.0 0.7 0.0 5.3 0.0 72.7 0.0 0.0 20.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 44.7 109.3 2.7 08-Sep-00 78 Woolgrass leaves 2.7 60.0 0.0 0.0 30.0 0.0 77.3 57.3 78.0 0.0 1.3 0.7 38.0 0.0 70.7 0.0 0.0 35.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 33.3 96.0 8.0 76 Woolgrass leaves 0.0 0.7 0.0 0.0 0.7 0.0 10.0 1.3 0.7 0.0 0.7 0.0 4.7 0.0 2.7 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 1.3 0.7 4.0 189.3 4.0 76 Woolgrass leaves 4.0 6.0 0.0 0.0 0.0 0.0 6.7 0.0 2.7 10.7 1.3 0.0 0.0 0.0 28.7 0.0 0.0. 2.7 1.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 6.7 112.7 76 Woolgrass leaves 0.0 4.0 0.0 1.3 4.0 0.0 16.0 14.0 18.7 6.7 3.3 0.0 10.0 0.0 44.7 0.0 0.0 12.7 0.0 2.0 0.0 0.0 0.0 0.0 0.0 0.0 20.0 140.0 12.7 2.0 80 Sedge leaves 6.7 2.0 0.0 0.0 6.7 0.7 48.0 16.0 2.7 2.7 1.3 2.7 12.0 0.0 10.0 0.0 0.0 8.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 40.0 129.3 0.7 80 Sedge leaves 6.0 52.0 0.0 0.0 10.7 0.0 133.3 48.7 58.7 2.0 0.0 13.3 6.7 0.0 12.0 0.0 0.0 12.0 0.0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 35.3 67.3 0.0 80 Sedge leaves 4.7 B4.0 0.0 0.0 0.0 0.0 20.0 6.0 7.3 4.7 0.0 0.0 0.0 0.0 1.3 0.0 0.0 4.7 0.0 0.0 0.0 0.0 0.0 0.0 1.3 0.0 12.7 264.7 0.0 78 Sedge leaves 0.0 0.7 0.0 0.0 0.7 0.0- 4.7 0.0 2.7 0.0 0.7 0.0 0.7 0.0 0.7 0.0 0.0 1.3 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.3 110.7 0.0 78 Sedge leaves 2.0 14.0 0.0 0.0 9.3 0.0 8.7 0.7 5.3 4.0 2.0 0.0 1.3 0.0 11.3 0.0 0.0 2.7 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 11.3 129.3 1.3 78 Sedge leaves 0.0 14.7 0.0 0.0 12.7 1.3 32.7 18.7 47.3 0.0 0.0 0.0 13.3 0.0 30.7 1.3 0.0 2.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 22.7 222.0 4.7 76 Sedge leaves 2.0 2.7 0.0 0.7 0.0 0.0 0.0 0.0 0.0 2.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 76 Sedge leaves 0.7 5.3 0.0 1.3 5.3 0.0 10.0 4.7 3.3 7.3 2.7 3.3 12.7 0.0 24.7 0.0 0.0 2.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 20.7 364.7 0.0 0.0 0.0 19.3 194.7 0.0 08-Sep-00 76 80 Sedge 0.0 5.3 0.0 0.0 5.3 0.0 13.3 8.0 10.7 28.0 2.7 0.0 16.0 0.0 57.3 0.0 0.0 12.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 23.3 82.0 3.3 Fall Rye leaves 2.7 10.7 0.0 0.0 12.0 0.0 25.3 29.3 0.7 0.0 27.3 0.0 35.3 2.7 5.3 0.0 0.0 14.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 12.0 20.7 2.0 80 Fall Rye leaves 1.3 15.3 0.0 0.0 0.0 0.0 19.3 86.0 48.0 8.0 11.3 2.7 33.3 0.0 26.0 0.7 1.3 28.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 66.0 14.7 08-Sep-00 80 Fall Rye leaves 0.0 10.7 0.0 0.0 3.3 0.0 32.7 34.7 13.3 2.0 7.3 38.7 49.3 0.0 4.7 0.0 1.3 11.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 12.0 18.0 2.7 0.7 78 Fall Rye leaves 0.0 2.7 0.0 0.0 28.0 0.7 18.7 8.0 4.0 0.0 12.0 0.7 0.7 0.0 4.0 0.0 0.7 4.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 14.0 22.7 35.3 78 Fall Rye leaves 5.3 13.3 0.0 0.0 23.3 0.0 2.7 17.3 15.3 28.0 44.7 1.3 4.7 0.0 79.3 0.0 0.0 13.3 0.0 0.0 0.0 0.0 5.3 0.0 0.0 0.0 34.7 0.7 0.7 78 76 Fall Rye leaves 0.0 28.0 0.0 0.0 9.3 0.0 33.3 22.0 18.0 2.7 6.0 0.7 14.0 0.0 14.0 0.7 0.0 12.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 22.0 6.0 4.7 76 Fall Rye leaves 0.0 0.0 0.0 0.0 5.3 4.7 8.0 1.3 2.0 1.3 3.3 4.0 10.7 0.0 7.3 0.0 0.0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 10.7 162.7 3.3 Fall Rye leaves 0.0 1.3 0.0 0.0 0.0 0.0 2.7 0.0 0.0 79.3 17.3 1.3 4.0 0.0 10.0 1.3 0.7 3.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 3.3 19.3 12.0 08-Sep-00 76 Fall Rye leaves 0.0 14.0 0.0 0.0 5.3 0.0 22.0 24.7 6.0 24.7 15.3 0.7 18.0 0.0 26.7 0.0 1.3 11.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 16.0 6.7 80 Woolgrass roots 0.0 0.0 0.0 0.0 0.0 0.0 21.3 5.3 0.0 2.7 2.7 2.7 2.7 0.0 2.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.7 0.0 0.0 0.0 0.0 302.0 80 Woolgrass roots 0.0 21.3 0.0 0.0 2.7 0.0 34.7 6.0 5.3 16.0 0.0 16.0 31.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0-0 0 0 0 0 0-0 0 0 124.0 184.0 80 Woolgrass roots 0.0 0.0 0.0 2.7 0.0 0.0 10.7 0.0 0.0 0.0 0.0 24.7 0.7 0.0 3.3 0.0 0.0 2.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.7 155.3 68.7 Woolgrass roots 0.0 1.3 0.0 10.7 2.0 0.0 10.7 2.7 0.0 0.0 2.7 0.7 2.0 0.0 1.3 0.0 0.0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 134.7 40.0 78 Woolgrass roots 5.3 0.0 0.0 0.0 27.3 0.0 48.0 0.0 5.3 5.3 0.7 5.3 50.7 0.0 26.0 0.0 0.0 5.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 74.7 20.0 78 Woolgrass roots 2.7 11.3 0.0 0.0 37.3 0.0 51.3 26.7 57.3 0.0 2.7 0.0 57.3 0.0 54.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0-0 2.7 37.3 4.0 76 Woolgrass roots 0.0 5.3 0.0 0.0 0.7 0.0 18.7 0.0 0.0 0.0 0.0 4.7 0.0 0.0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0 0 0 0 5.3 238.0 8.0 76 Woolgrass roots 0.0 1.3 0.0 0.0 16.0 0.0 13.3 0.0 12.0 26.7 0.7 5.3 22.7 0.0 22.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0 0 0.0 0 0 0 0 60.0 9.3 08-Sep-00 76 Woolgrass roots 0.0 5.3 0.0 0.0 5.3 0.0 32.0 0.0 21.3 5.3 0.0 0.0 68.0 0.0 58.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0.0 0 0 10-7 46.0 24.7 80 Sedge 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.7 0.0 0.0 1.3 1.3 8.7 0.0 6.0 0.0 0.0 0.0 2.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 254.7 26.0 80 Sedge 0.0 10.7 0.0 0.0 0.0 0.7 38.7 29.3 16.0 1.3 7.3 31.3 43.3 0 .0- 2.7 0.0 0.0 0.0 0.0 0.0 0.0 1.3 0.0 0.0 0.0 0.0 6.7 106.0 50.0 08-Sep-00 80 Sedge 0.0 2.7 0.0 16.0 5.3 0.0 0.0 5.3 5.3 0.0 5.3 24.0 8.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0 0 8.0 30.0 44.7 78 Sedge roots 0.0 0.0 0.0 0.0 0.0 0.0 2.7 2.7 0.0 2.7 0.0 0.0 0.7 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.7 0.0 0.0 201.3 3.3 78 Sedge roots 78 Sedge 2.7 8.0 0.0 0.0 5.3 0.0 16.0 2.7 5.3 16.0 6.0 0.0 16.7 0.0 17.3 0.7 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 55.3 43.3 2.7 24.0 0.0 0.0 0.0 0.0 21.3 11.3 32.0 0.0 3.3 8.7 64.7 0.0 26.7 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 27.3 114.7 50.0 76 Sedge 0.0 0.0 0.0 0.0 0.0 1.3 21.3 2.7 1.3 22.7 0.7 0.0 2.0 0.0 1.3 1.3 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1 3 0.0 0.7 244.0 4.0 Sedge roots 0.0 10.7 0.0 0.0 5.3 0.0 10.7 0.0 5.3 90.7 8.7 0.0 10.0 0.0 69.3 5.3 0.0 0.0 0.0 0.0 0.0 5.3 5.3 0.0 0.0 0.0 0.0 331.3 106.0 08-Sep-00 76 Sedge roots 0.0 0.0 0.0 0.0 0.0 0.0 10.7 10.7 26.7 0.0 18.7 0.7 49.3 0.0 3.3 0.0 0.0 0.7 0.0 0.0 0.0 0.0 2.7 0.0 0.0 0.0 0.7 26.7 34.0 80 Fall Rye roots 0.0 0.0 0.0 0.0 0.0 0.0 10.7 0.0 0.0 0.0 12.0 0.7 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10-7 10.7 80 Fall Rye roots 0.0 0.0 0.0 0.0 0.0 0.0 10.7 85.3 0.0 0.0 22.0 21.3 18.7 0.0 1.3 0.0 0.0 21.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 128.0 80 Fall Rye roots 0.0 5.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 12.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0 0 0 0 0 0 42.7 0.0 85.3 78 Fall Rye 0.0 0.0 0.0 0.0 0.0 21.3 0.0 0.0 0.0 0.0 8.0 0.0' 0.0 0.0 21.3 0.0 0.0 0.0 0.0 0;0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 20.0 Table J-1. "Number of benthic organisms associated with woolgrass, sedge, fall rye and control samples at each elevation over time. Data has been averaged over the three sample locations. Elevation Plant";.: (m)-.;"t" Speciesv 09-Aug-00 08-Sep-00 06-Jul-00 09-Aug-00 06-Jul-00 09-Aug-00 08-Sep-00 06-Jul-00 09-Aug-00 06-Jul-OO 09-Aug-00 08-Sep-00 s10* s11 "812-513. S l 4 ; s 1 5 s 1 6 s17. ,s18/s19:s20 S21 's22;s23s24 s25 s26 s27 s28; -S29/ 78 Fall Rye "Taxa Code s1 s2 S3 0.0 0.0 0.0 0.0 10.7 0.0 0.0 0.0 0.0 0.0 42.7 0.7 9.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0-0 0 0 00 0 0 00 0 0 00 0 ° 236.0 78 Fall Rye 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 32.0 0.0 5.3 0.0 122.0 0.0 21.3 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 0.0 45.3 76 Fall Rye roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 o.o o.o o.o o.o o o o o o o o o o o o o 1 2 0 76 Fall Rye 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 85.3 34.7 21.3 36.7 0.0 11.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0-0 0-0 00 00 00 102-0 08-Sep-OO 76 Fall Rye roots 0.0 11.3 0.0 0.0 0.0 0.0 0.7 0.0 0.0 0.0 2.0 0.7 54.7 0.0 22.0 0.0 0.0 0.0 0.0 0.0 0.0 00 00 0 0 00 00 42.7 0.7 54.7 80 Control 0.0 0.0 0.0 0.0 0.7 0.0 0.0 2.0 0.0 0.0 0.0 0.0 0.3 0.0 0.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00 0.0 0.0 0 0 00 16.3 0.0 80 Control 80 Control 0.0 0.7 0.0 0.0 0.3 0.0 2.0 8.3 2.0 1.3 0.3 2.0 1.0 0.0 0.7 0.0 0.0 2.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0-0 3.3 2.7 0.0 0.3 0.0 0.0 0.0 0.0 0.3 6.7 3.7 1:0 0.0 2.0 1.7 0.0 0.0 0.0 0.0 0.0 0.0 0-0 0 0 0-0 00 0 0 0 0 0 0 4.3 0.7 15.7 78 Control 0.0 0.0 0.0 0.0 0.7 2.0 0.0 1.0 0.0 3.0 0.0 0.0 0.3 0.0 0.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0-0 00 00 1 0 0.3 78 Control 0.3 0.3 0.0 0.0 0.0 0.0 5.3 4.0 1.3 3.7 0.3 0.0 0.7 0.0 2.3 0.0 0.0 1.7 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 3.0 0.3 08-Sep-OO 78 Control 0.0 0.7 0.0 0.0 0.7 0.0 1.0 1.0 1.0 0.3 0.3 1.0 16.7 0.0 2.0 0.0 0.3 1.0 0.0 0.0 0.0 0.3 0.0 0.0 0.0 0.0 0.0 0.3 12.3 76 Control 0.0 0.0 0.0 0.0 0.0 0.7 0.7 0.0 0.7 3.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0-0 0-0 0-0 00 00 00 0 0 00 0 0 0 0 1-7 1-7 76 Control 0.0 0.7 0.0 0.0 0.0 0.0 4.3 0.0 0.3 3.3 0.0 1.0 0.0 0.0 0.7 0.3 0.0 0.3 0.0 0.0 0.0 0.3 0.0 0.7 0.0 0.0 0.3 1.7 2.7 76 Control 0.0 0.0 0.0 0.0 0.0 0.0 2.0 2.7 1.3 9.3 0.0 1.7 1.7 0.0 3.3 5.3 0.0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0.0 0.0 2.7 4.7 * Note that total number of invertebrates has been calculated by multiplying the observed number by two to account for subsampling (i.e. 1/2 sample went for vegetation/nutrient analysis) for all samples except control ** Refer to Table 15 in this thesis for key to taxa codes Appendix J: Benthos Summary Data o Table J-1. "Number of benthic organisms associated with woolgrass, sedge, fall rye and control samples at each elevation over time (continued) Data has been averaged over the three sample locations. • Elevation ; ':; Date. .' . (m) .; 06-Jul-OO 09-Aug-00 08-Sep-OO 06-Jul-00 06-Jul-00 06-Jul-00 06-Jul-00 06-Jul-OO 06-Jul-00 06-Jul-OO 06-Jul-OO 06-Jul-OO 09-Aug-00 06-Jul-OO 09-Aug-00 08-Sep-OO 06-Jul-OO 09-Aug-00 08-Sep-OO 06-Jul-OO 09-Aug-00 08-Sep-OO 06-Jul-OO 09-Aug-00 08-Sep-OO 06-Jul-OO 09-Aug-00 06-Jul-OO 09-Aug-OO 06-Jul-OO Plant Species v. substrate "Taxa Code • • -s30: :s31 s32>"s33 s34 • - s35 80 Woolgrass leaves 0.0 0.0 0.7 0.7 0.0 s36 s37 S38'S39'.'JS'40:.'S41' s42- s43 s44 s45 -s46- s47 s48 s'49j^ s50,s51: s52 s53s54 s55 's56-;s57_ s58Js59 19.3 0.7 0.0 0.0 0.0 0.0 4.0 1.3 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00 0 0 0 0 00 00 80 Woolgrass leaves 2.7 0.0 3.3 0.0 0.0 6.0 5.3 0.0 0.0 0.0 0.0 0.0 2.0 1.3 0.0 6.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0 0 00 0 0 4.0 2.0 0.0 80 Woolgrass leaves 1.3 0.0 1.3 0.0 2.7 2.7 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00 00 00 00 00 0 0 00 0 0 1-3 6.0 2.0 0.0 0.0 78 Woolgrass leaves 2.0 0.0 0.7 0.0 0.0 11.3 0.0 0.0 0.0 0.0 2.0 0.0 0.7 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 00 0 0 09-Aug-OO 78 Woolgrass leaves 5.3 0.0 12.7 0.7 0.7 4.0 12.0 1.3 0.0 0.0 56.0 1.3 2.7 5.3 1.3 101.3 4.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0 0 00 08-Sep-OO 78 Woolgrass leaves 0.7 2.7 6.0 0.0 1.3 2.7 68.0 1.3 0.0 6.7 0.0 0.0 0.7 57.3 6.7 2.7 0.0 0.0 0.0 0.0 0.0 2.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 76 Woolgrass leaves 6.0 0.0 0.0 0.0 0.0 10.0 18.7 0.0 0.0 0.0 0.0 0.0 00 0.0 00 1.3 0.0 0.0 0.7 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 09-Aug-OO 76 Woolgrass leaves 9.3 0.0 1.3 0.0 0.0 3.3 21.3 0.0 0.0 0.0 1.3 0.0 1.3 1.3 0.0 60.0 0.0 0.0 0.0 2.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 08-Sep-OO 76 Woolgrass leaves 7.3 0.0 2.7 0.0 0.0 9.3 5.3 2.7 0.0 0.7 4.0 0.0 0.0 0.0 1.3 1.3 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 Sedge leaves 0.0 0.0 0.0 0.0 0.0 20.0 26.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 13.3 0.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 2.0 09-Aug-OO 80 Sedge leaves 0.7 0.0 0.0 0.0 0.0 42.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 00 2.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00 0 0 93 0.0 2.0 08-Sep-OO 80 Sedge leaves 16.7 0.0 0.7 0.0 0.0 62.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00 00 0.0 1-3 0.0 4.0 0.7 78 Sedge leaves 2'o 0.0 0.0 0.0 0.0 9.3 0.0 0.0 0.0 0.0 0.0 2.0 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 09-Aug-OO 78 Sedge leaves 16J 0^ 0 0.7 0.0 0.0 13.3 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.7 0.0 3.3 2.7 0.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 08-Sep-OO 78 Sedge leaves 0.7 0.0 1.3 0.0 0.0 20.0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 4.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 00 2.0 76 Sedge leaves 0.7 0.0 0.0 0.0 0.0 2.0 7.3 0.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 00 0 0 00 0 0 00 0.7 0.0 0.0 0.0 09-Aug-OO 76 Sedge leaves 4^ 0 0.0 1.3 0.0 0.0 16.0 5.3 0.0 0.0 0.0 0.7 0.0 0.7 0.7 0.0 0.0 0.0 0.0 2.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 08-Sep-OO 76 Sedge leaves 5.3 0.0 1.3 0.0 0.0 2.7 48.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0-7 0.0 0.0 4.7 0.0 0.0 0.0 0.0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 80 Fall Rye leaves 0.0 0.0 0.0 0.0 0.0 6.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 00 0.0 00 00 0 0 00 0 0 0-7 00 0 0 00 00 0 0 0-7 0 0 00 09-Aug-OO 80 Fall Rye leaves 0.7 0.0 0.0 0.0 0.0 17.3 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0 0 0.0 00 00 0 0 00 0 0 0 0 0 0 00 0 0 o o o o o o 08-Sep-OO 80 Fall Rye leaves 4.0 0.0 6.0 0.0 0.0 14.0 6.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00 00 00 0 0 0 0 0 0 0 0 00 0 0 00 0 ° 0 ° 78 Fall Rye leaves 24.7 0.0 8.0 0.0 0.0 6.7 0.0 0.0 0.0 0.0 0.0 0.0 00 0 0 00 00 0 0 00 00 00 00 0-7 0 0 00 0 0 00 0 0 00 00 00 09-Aug-OO 78 Fall Rye leaves 40.7 0.0 6.0 0.0 0.0 10.7 24.0 1.3 0.0 0.0 1.3 0.0 0.0 0.0 0.0 0.0 2.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.3 0.0 0.0 0.0 08-Sep-OO 78 Fall Rye leaves 0.0 0.0 1.3 0.0 0.0 2.7 14.7 0.0 0.0 2.0 0.0 0.0 0.0 0.0 0.0 1.3 4.0 0.0 0.0 0.0 0.0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 76 Fall Rye leaves 1.3 1.3 2.7 0.0 0.0 13.3 4.0 1.3 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 2.0 0.0 0.0 4.0 0.0 0.0 0.0 09-Aug-OO 76 Fall Rye leaves 6.0 2.7 0.0 0.0 0.0 4.7 325.3 0.0 0.0 0.0 0.0 0.0 1.3 0.0 0.0 0.0 0 0 0 0 00 00 0 0 00 0 0 0 0 0 0 00 0 0 00 0 ° 0 ° 08-Sep-OO 76 Fall Rye leaves 4.0 0.0 1.3 0.0 0.0 4.0 1172.7 1.3 0.0 0.0 0.0 0.0 0.0 0.0 00 00 1 3 0.0 0.0 0.0 00 0 0 0.0 00 0.0 00 0 0 00 00 00 80 Woolgrass roots 21.3 0.0 6.7 0.0 0.7 101.3 10.7 0.0 0.0 0.0 2.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00 00 0.0 0 0 00 0.0 0.0 0.0 0.0 0 0 00 80 Woolgrass roots 42.7 0.0 33.3 0.0 0.0 37.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 00 00 0 0 00 0.0 00 0.0 00 0.0 00 0 0 00 08-Sep-OO 80 Woolgrass roots 141.3 0.0 8.0 0.0 0.7 26.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00 0 0 00 00 0 0 0 0 00 0 0 00 00 00 0 0 0 0 0 ° 0 0 0 0 0 0 78 Woolgrass roots 5.3 0.0 30.7 0.0 0.7 149.3 5.3 1.3 0.0 0.0 0.0 0.0 0.0 0.0 00 00 00 5.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00 00 00 78 Woolgrass roots 37.3 2.7 6.0 0.0 0.0 16.0 16.0 0.0 0.0 0.0 10.7 0.0 5.3 0.0 5.3 0.0 0.0 0.0 0.0 0.0 0.0 00 0 0 00 0.0 0.0 0 0 00 00 00 78 Woolgrass roots 74.7 1.3 12.0 0.0 0.0 16.0 26.7 13.3 0.0 5.3 0.0 0.0 16.0 24.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00 0 0 00 0 0 00 00 00 76 Woolgrass roots 6.7 0.0 12.7 0.0 1.3 96.0 48.0 0.0'0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 12.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0.0 00 0.0 00 00 76 Woolgrass roots 58.7 1.3 17.3 0.0 0.0 64.0 325.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 00 1 3 0.0 0.0 00 0 0 00 0 0 00 00 00 0 0 00 0 0 00 76 Woolgrass roots 181.3 1.3 14.0 0.0 0.0 13.3 40.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00 0 0 0 0 00 0 0 00 00 00 00 00 0 0 0 ° 0 0 0 ° 80 Sedge roots Sedge roots 0.0 0.0 19.3 0.0 0.0 229.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 00 0 0 0.0 0.0 0 0 0 0 00 00 0 0 0 0 00 0 0 0-0 0 ° 0 ° 0 0 0 0 117.3 0.0 24.7 0.0 0.0 105.3 0.0 0.0 0.0 0.0 0.0 10.7 1.3 0.0 0.0 0.0 0.0 0.0 0 0 00 00 00 0 0 0 0 0 0 00 0 0 21.3 0.0 0.0 80 Sedge roots 61.3 10.7 26.7 0.0 0.0 253.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 00 00 0 0 0 0 00 00 00 00 0 0 0 0 0 0 0 ° 0 ° 2 7 00 00 78 Sedge roots 16.0 2.7 27.3 0.0 0.0 216.0 8.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 00 00 00 00 0 0 00 0 0 00 00 00 00 00 78 Sedge roots 61.3 1.3 14.0 0.0 0.0 90.7 40.0 2.7 0.0 0.0 0.0 5.3 0.0 0.0 0.0 0.0 2.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0-0 78 Sedge roots 76 Sedge 64.0 3.3 3.3 0.0 0.0 2.7 18.7 0.0 0.0 0.0 0.0 0.0 2.7 0.0 0.0 0.0 0.0 5.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 00 ~~2J7 OO 6.0 0.0 0.0 84l) 141.3 0.0 0.0 0.0 5.3 0.0 0.0 0.0 0.0 0.0 0.0 2.7 0.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 0.0 76 Sedge 165.3 0.0 96.0 0.0 0.0 85.3 853.3 16.0 0.0 0.0 5.3 0.0 5.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0 0 0 0 00 0 0 00 00 00 08-Sep-OO 76 Sedge roots 80.0 0.0 5.3 0.0 2.7 16.0 525.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 00 00 10-7 0.0 0.0 0.0 0.0 00 00 00 00 00 80 Fall Rye roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 21.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 21.3 0.0 0.0 0.0 80 Fall Rye roots 0.0 0.0 21.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 10.7 0.0 0.0 0.0 0.0 0.0 0.0 o.o o.o o.o o o o o o o o o o.o o o o o o.o o o o o 08-Sep-OO 80 Fall Rye 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 .0.0 0.0 0.0 o.o o.o o.o o.o o o o o o o o o o o o o 0 0 0-0 78 Fall Rye roots 26.7 0.7 2.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0' 0.0 0.0 ' 0.0 21.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Table J-1. "Number of benthic organisms associated with woolgrass, sedge, fall rye and control samples at each elevation over time (continued) Data has been averaged over the three sample locations. ; Elevation Plant "Taxa Code- ' •: : . ; •-''.^••:-i'-tr>-'U% : > ' ; • • • - ' : ' ' - - ••rV'vv.r-'?-j -".s-Jr-X'-•-'''" • - - - -•-»-." vS* Date • (m) : . Species !.substrate .'s30 s31; s32 s33 s34 s35; s36" s37 s38 s39 s40 s41- s 4 2 s 4 3 s44 s45; s46 s47 s48 s49 s50 ,s51 ' Note that total number of invertebrates has been calculated by multiplying the observed number by two to account for subsampling (i.e. 1/2 sample went for vegetation/nutrient analysis) for all samples except control ** Refer to Table 15 in this thesis for key to taxa codes Appendix J : Benthos Summary Data Table J-1. 'Number of benthic organisms associated with woolgrass, sedge, fall rye and control samples at each elevation over time (continued) Data has been averaged over the three sample locations. >0 to. Elevation : >: Plant "Taxa Code Total Total benthos i f ; > . s . . Taxa Date (m) :' Species . substrate s60 s61 s62 s63 s64 s65 s66 s67 s68. s69 s70; Benthos (leaves;+;rbots)v" - ' i ' Count'' 06-Jul-OO 80 Woolgrass leaves 0.0 0.0 0.0 6.0 0.0 0.0 0.0 0.7 0.0 0.7 0.0 162.0 742.0 11.3 09-Aug-OO 80 Woolgrass leaves 1.3 0.0 0.0 10.0 0.0 0.0 0.0 0.0 3.3 0.0 0.0 355.3 920.7 17.3 08-Sep-OO 80 Woolgrass leaves 0.0 1.3 2.7 0.7 0.0 1.3 0.0 0.0 3.3 0.0 0.0 91.3 551.3 12.7 06-Jul-OO 78 Woolgrass leaves 0.0 0.0 0.0 5.3 0.0 0.0 0.0 6.7 0.0 3.3 0.0 196.0 610.7 13.3 09-Aug-OO 78 Woolgrass leaves 0.7 0.0 0.0 8.7 0.0 0.0 0.0 1.3 38.7 0.0 1.3 700.7 1100.7 20.0 08-Sep-OO 78 Woolgrass leaves 0.0 0.0 0.0 15.3 0.0 0.0 0.0 0.0 37.3 2.7 0.0 763.3 1350.0 18.7 06-Jul-OO 76 Woolgrass leaves 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.3 0.0 2.7 0.0 259.3 732.7 10.3 09-Aug-OO 76 Woolgrass leaves 0.0 0.0 0.0 0.7 0.0 0.0 0.0 3.3 22.0 0.0 0.7 298.7 962.0 14.3 08-Sep-OO 76 Woolgrass leaves 0.0 0.0 0.0 12.7 0.0 0.0 0.0 0.0 52.7 2.7 0.0 348.0 876.0 19.0 06-Jul-OO 80 Sedge leaves 0.7 0.0 0.0 0.7 0.0 0.0 0.0 10.7 0.0 0.0 0.0 353.3 905.3 12.7 09-Aug-OO 80 Sedge leaves 2.0 0.0 0.0 8.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 526.7 1152.7 16.0 08-Sep-OO 80 Sedge leaves 0.0 0.0 2.0 0.7 0.0 0.0' 0.0 0.0 0.7 1.3 0.0 481.3 991.3 11.7 06-Jul-OO 78 Sedge leaves 0.0 0.0 0.0 0.0 0.0 0.0 0.0 3.3 1.3 0.0 0.0 139.3 632.0 8.7 09-Aug-OO 78 Sedge leaves 0.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 7.3 0.0 0.0 243.3 657.3 14.3 08-Sep-OO 78 Sedge leaves 0.0 0.0 0.0 2.7 0.0 0.0 0.0 0.0 13.3 0.0 0.0 457.3 944.7 14.0 06-Jul-OO 76 Sedge leaves 0.0 0.0 0.0 0.0 0.0 0.0 0.0 5.3 2.7 0.0 0.0 406.0 956.7 7.3 09-Aug-OO 76 Sedge leaves 0.0 0.0 0.0 0.7 0.0 0.0 0.0 1.3 13.3 0.0 0.0 328.7 2230.0 14.3 08-Sep-OO 76 Sedge leaves 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 28.7 0.7 0.0 331.3 1156.0 15.3 06-Jul-OO 80 Fall Rye leaves 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 208.7 307.3 11.7 09-Aug-OO 80 Fall Rye leaves 0.0 0.0 0.0 14.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 398.7 739.3 13.3 08-Sep-OO 80 Fall Rye leaves 0.7 0.0 0.0 12.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 282.7 428.7 13.0 06-Jul-OO 78 Fall Rye leaves 0.0 0.0 0.0 2.7 0.0 0.0 0.0 2.7 0.0 0.0 0.0 199.3 320.7 12.3 09-Aug-OO 78 Fall Rye leaves 0.0 0.0 0.0 2.0 0.0 0.0 0.0 0.0 8.0 0.0 0.7 380.0 796.0 16.3 08-Sep-OO 78 Fall Rye leaves 0.0 0.0 0.0 11.3 0.0 0.0 0.0 0.0 9.3 0.0 0.0 232.0 642.7 16.0 06-Jul-OO 76 Fall Rye leaves 0.0 0.0 0.0 0.7 0.0 0.0 1.3 1.3 0.0 0.0 0.0 258.0 350.7 13.0 09-Aug-OO 76 Fall Rye leaves 0.0 0.0 0.0 0.0 0.0 0.0 0.7 0.0 2.7 0.0 0.0 496.0 1268.7 13.0 08-Sep-OO 76 Fall Rye leaves 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 2.7 0.0 0.0 1388.7 2817.3 15.0 06-Jul-OO 80 Woolgrass roots 0.0 0.0 0.0 0.0 0.7 0.0 0.0 0.0 5.3 24.0 0.0 580.0 9.0 09-Aug-OO 80 Woolgrass roots 0.0 0.0 0.0 10.7 0.0 0.0 0.0 0.0 0.0 10.7 0.0 565.3 9.7 08-Sep-OO 80 Woolgrass roots 0.0 0.0 0.0 12.0 0.0 0.0 0.0 0.0 0.0 8.0 0.0 460.0 9.7 06-Jul-OO 78 Woolgrass roots 0.0 0.0 0.0 1.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 414.7 9.7 09-Aug-OO 78 Woolgrass roots 0.0 0.0 0.0 26.7 0.0 0.0 0.0 0.0 0.0 37.3 0.0 400.0" 12.0 08-Sep-OO 78 Woolgrass roots 0.0 0.0 0.0 52.0 0.0 0.0 0.0 0.0 13.3 64.0 0.0 586.7 15.7 06-Jul-OO 76 Woolgrass roots 0.0 0.0 0.0 5.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 473.3 8.0 09-Aug-OO 76 Woolgrass roots 0.0 0.0 0.0 5.3 0.7 0.0 0.0 0.0 0.0 6.7 0.0 663.3 12.3 08-Sep-OO 76 Woolgrass roots 0.0 0.0 0.0 0.7 0.0 0.0 0.0 0.0 0.0 32.0 0.0 528.0 10.0 06-Jul-OO 80 Sedge roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 5.3 0.0 552.0 6.7 09-Aug-OO 80 Sedge roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 21.3 0.0 626.0 13.3 08-Sep-OO 80 Sedge roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 13.3 0.0 510.0 10.3 06-Jul-OO 78 Sedge roots 0.0 0.0 0.0 6.0 0.0 0.0 0.0 0.0 0.0 29.3 0.0 492.7 7.7 09-Aug-OO 78 Sedge roots 0.0 0.0 0.0 0.0 0.0 0.0 2.7 0.0 0.0 2.7 0.0 414.0 13.7 08-Sep-OO . 78 Sedge roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 10.7 0.0 0.0 487.3 12.3 06-Jul-OO 76 Sedge roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 14.7 0.0 550.7 10.0 09-Aug-OO 76 Sedge roots 0.0 0.0 0.0 10.7 0.0 0.0 0.0 0.0 10.7 5.3 0.0 1901.3 12.7 08-Sep-OO 76 Sedge roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 18.7 0.0 824.7 11.0 06-Jul-OO 80 Fall Rye roots 10.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 98.7 3.0 09-Aug-OO 80 Fall Rye roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 340.7 5.0 08-Sep-OO 80 Fall Rye roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 146.0 2.0 06-Jul-OO 78 Fall Rye roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 121.3 3.3 Table J-1. 'Number of benthic organisms associated with woolgrass, sedge, fall rye and control samples at each elevation over time (continued) Data has been averaged over the three sample locations. 'Elevation Plant . ?*Taxa Codes -V-'~ ••-!> :•' . Total • Total benthos •"• " ?;vi Taxa\ Date (m) • Species': -V; substrate. 's6q.Vs6l'.s62ji>63;;s64 s65.s66s67.js68' s69 s70; Benthos - (leaves + roots) -'.•Count . l t | \ 09rAug'-00 78 Pall Rye roots 0.0 0.0 0.0 2.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 416.0 5.0 08-Sep-00 78 Fall Rye roots 0.0 0.0 0.0 12.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 410.7 5.7 06-Jul-OO 76 Fall Rye roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 92.7 2.0 09-Aug-OO 76 Fall Rye roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 772.7 6.0 08-Sep-OO 76 Fall Rye roots 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1428.7 6.7 06-Jul-OO 80 Control 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 25.0 25.0 3.7 09-Aug-OO 80 Control 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.3 0.0 30.3 30.3 8.7 08-Sep-OO 80 Control 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.3 0.0 40.0 40.0 6.7 06-Jul-OO 78 Control 0.0 0.0 0.0 0.0 0.0 0.0 0.3 0.0 0.0 0.3 0.0 15.3 15.3 6.3 09-Aug-OO 78 Control 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.7 0.0 0.0 38.3 38.3 9.7 08-Sep-OO 78 Control 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 47.0 47.0 10.0 06-Jul-OO 76 Control 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 12.3 12.3 4.7 09-Aug-OO 76 Control 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 25.0 25.0 9.0 08-Sep-OO 76 Control 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 70.3 70.3 7.7 * Note that total number of invertebrates has been calculated by multiplying the observed number by two to account for subsampling (i.e. 1/2 sample went for vegetation/nutrient analysis) for all samples except control ** Refer to Table 15 in this thesis for key to taxa codes Appendix K: Complete ANOVA tables for analyses related to benthos Table K-1. Complete ANOVA table for benthic organism abundance (log10(x) transformed data) Source SS MS Num DF Num *F Ratio Prob > F Site&Random 0.43562 0.21781 2 1.5616 0.3043 Elevation 0.18933 0.09467 2 0.8467 0.4936 Species 26.2968 8.7656 3 •116.3134 <0001 Site*Elevation&Random 0.44723 0.11181 4 2.3446 0.1136 Site*Species&Random 0.45217 0.07536 6 1.5804 0.2352 Elevation*Species 0.52952 0.08825 6 1.8507 0.1715 Site*Elevation*Species&Random 0.57224 0.04769 12 0.8020 0.6465 Time 1.46706 0.73353 2 *12.3369 < 0001 ElevationTime 0.34095 0.08524 4 1.4336 0.2373 Species*Time 0.52924 0.08821 6 1.4835 0.2039 Elevation*Species*Time 0.4447 0.03706 12 0.6233 0.8118 * Indicates a significant F-test result Table K-2. Complete ANOVA table for benthic organism abundance per gram of plant material (log10(x) transformed data) Source SS MS Num DF Num *F Ratio Prob > F Site&Random 0.54515 0.27257 2 0.4391 0.6766 Elevation 1.17356 0.58678 2 4.0845 0.1080 Species 5.01685 2.50842 2 3.8860 0.1155 Site*Elevation&Random 0.57465 0.14366 4 0.8534 0.5301 Site*Species&Random 2.582 0.6455 4 3.8344 0.0501 Elevation*Species 0.70726 0.17682 4 1.0503 0.4394 Site*Elevation*Species&Random 1.34676 0.16835 8 1.1598 0.3323 substrate 3.61957 3.61957 1 *24.9377 <0001 Elevation*substrate 0.04905 0.02453 2 0.1690 0.8448 Species*substrate 1.57744 0.78872 2 *5.4340 0.0059 Elevation*Species*substrate 0.73833 0.18458 4 1.2717 0.2870 Time 2.01818 1.00909 2 *6.9523 0.0016 Time*Elevation 2.67388 0.66847 4 •4.6055 0.0020 Time*Species 0.83697 0.20924 4 1.4416 0.2268 Time*substrate 0.05781 0.02891 2 0.1992 0.8198 Time*Elevation*Species 1.13209 0.14151 8 0.9750 0.4607 Time*Elevation*substrate 0.95679 0.2392 4 1.6480 0.1691 Time*Species*substrate 1.03664 0.25916 4 1.7855 0.1386 Time*Elevation*Species*substrate . 1.03711 0.12964 8 0.8932 0.5257 * Indicates a significant F-test result 194 Table K-3. Complete ANOVA table for benthic taxonomic richness analysis (non transformed data). Source SS MS Num DF Num *F Ratio Prob > F Site&Random 76.0173 38.0087 2 3.3275 0.1542 Elevation 48.2173 24.1087 2 2.6344 0.1861 Species 753.425 251.142 3 *28.7885 0.0006 Site*Elevation&Random 36.6058 9.15144 4 1.4189 0.2862 Site*Species&Random 52.3398 8.7233 6 1.3523 0.3078 Elevation*Species 58.3539 9.72564 6 1.5084 0.2557 Site*Elevation*Species&Random 77.3078 6.44232 12 0.6886 0.7539 Time 357.599 178.799 2 •19.1121 <0001 Elevation*Time 101.942 25.4855 4 •2.7242 0.0401 Species*Time 49.8422 8.30704 6 0.8879 0.5112 Elevation*Species*Time 73.7569 6.14641 12 0.6570 0.7825 * Indicates a significant F-test result 195 Appendix L: NMS ordination diagrams for benthos abundance analysis Benthos abundance (a) • • * * • • « a • * • * a % m • Stress = 0.144 Time • 1 • 2 3 Axis 2 Benthos abundance (b) • • • * • • #• * • a « ^ -Stress = 0.144 Elevation (m) • 76 • 78 80 Axis 1 Figure L-1. Ordination diagram for benthos abundance data. Note that there is a weak grouping by time (Fig. L-1a) and no clear grouping by elevation (m) (Fig. L-1b). Distances between sample units approximate dissimilarity in species composition. 196 Ostracoda Axis 1 Ostracoda Elevation (m) • 76 • 78 80 Axis 1 197 Ostracoda Species • Woolgrass • Sedge • Fall rye • Control Axis 1 Figure L-2. NMS ordination diagrams for subclass Ostracoda, accounting for 22% of all benthic organisms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of Ostracoda individuals. The different symbols in the ordination diagrams represent different sample times (Figure L-2a), elevation (m) (Figure L-2b) or plant species (Figure L-2c). 198 Enchytraeidae Axis 1 Enchytraeidae (b) • U5 Elevation (m) • 76 • 78 80 Stress = 0.144 Axis 1 Enchytraeidae (c) Species Woolgrass T # Sedge Fall rye Control CO X < Stress = 0.144 Axis 1 Figure L- 3. NMS ordination diagrams for family Enchytraeidae (Class Oligochaeta), accounting for 21% of all benthic organisms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of individuals from the family Enchytraeidae. The different symbols in the ordination diagrams represent different sample times (Figure L-3a), elevation (m) (Figure L-3b) or plant species (Figure L-3c). 200 Phaenopsectra C O x < Axis 1 Phaenopsectra (b) 1 • • 9 -Elevation (m) • 76 • 78 • 80 Stress = 0.144 Axis 1 201 Phaenopsectra oo </> x < (c) Species • Woolgrass • Sedge • Fall rye • Control Stress = 0.144 Axis 1 Figure L-4. NMS ordination diagrams for Phaenopsectra, the most abundant chironomid taxa noted during this study. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of Phaenopsectra. The different symbols in the ordination diagrams represent different sample times (Figure L-4a), elevation (m) (Figure L-4b) or plant species (Figure L-4c). 202 Nematoda Axis 1 Nematoda (b) Elevation (m) • 76 • 78 80 Stress = 0.144 Axis 1 203 Nematoda (c) Species • Woolgrass • Sedge • Fall rye • Control CO X < Stress = 0.144 Axis 1 Figure L -5. NMS ordination diagrams for Phylum Nematoda, accounting for 8% of all benthic organisms by abundance. The sizes of symbols within the ordination diagrams are proportional to the magnitude of abundance of nematodes. The different symbols in the ordination diagrams represent different sample times (Figure L-5a), elevation (m) (Figure L-5b) or plant species (Figure L-5c). 204 Appendix M: Complete ANOVA table for diatom to benthos ratio Table M-1. Complete A N O V A table for ratio of diatom biovolume to benthos density Source S S MS Num DF Num *F Ratio Prob > F Site&Random 0.75991 0.37996 2 1.2338 0.5004 Elevation 0.80429 0.40215 2 1.1520 0.4026 Species 0.76887 0.76887 1 3.4478 0.2045 Site*Elevation&Random 1.39639 0.3491 4 1.3216 0.3968 Site*Species&Random 0.44601 0.223 2 0.8443 0.4945 Elevation*Species 0.09321 0.0466 2 0.1764 0.8444 Site*Elevation*Species&Random 1.05658 0.26414 4 1.8730 0.1480 Time 1.06711 0.53356 2 *3.7833 0.0373 Elevation*Time 1.5466 0.38665 4 2.7416 0.0521 Spec iesT ime 0.12423 0.06211 2 0.4404 0.6489 Elevation*Species*Time 1.2952 0.3238 4 2.2960 0.0885 * Indicates a significant F-test result 205 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0092643/manifest

Comment

Related Items