Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Distribution and abundance of nearshore aquatic habitat, Fraser River, British Columbia Perkins, Ashley 2007

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

Item Metadata

Download

Media
24-ubc_2008_spring_perkins_ashley.pdf [ 2.41MB ]
Metadata
JSON: 24-1.0066171.json
JSON-LD: 24-1.0066171-ld.json
RDF/XML (Pretty): 24-1.0066171-rdf.xml
RDF/JSON: 24-1.0066171-rdf.json
Turtle: 24-1.0066171-turtle.txt
N-Triples: 24-1.0066171-rdf-ntriples.txt
Original Record: 24-1.0066171-source.json
Full Text
24-1.0066171-fulltext.txt
Citation
24-1.0066171.ris

Full Text

DISTRIBUTION AND ABUNDANCE OF NEARSHORE AQUATIC HABITAT, FRASER RIVER, BRITISH COLUMBIA  by ASHLEY ANNE HORNE PERKINS B.Sc., The University of British Columbia, 2005  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Geography)  THE UNIVERSITY OF BRITISH COLUMBIA December 2007 © Ashley Anne Horne Perkins, 2007  Abstract Physical habitat for instream biota derives from a combination of stream system structural and hydraulic phenomena. Consequently, the quantity and quality of physical habitat is dynamic both over time and in space along the river, laterally, longitudinally and vertically. Its characterization through stream assessment and classification leads to a better understanding of factors that determine and limit habitat extent and quality. This thesis investigates the effects of space and time on nearshore aquatic habitat in the gravel reach of Fraser River, British Columbia by employing a large river, stage-adaptive habitat classification system. The distribution and abundance of habitat are spatially quantified at the reach scale (32 km), and temporally quantified through a period of about 60 years at several adjacent gravel bars (7 km), and at approximately 500 m3 s-1 increments in discharge during the declining limb of the flood hydrograph at two well-developed gravel bars. Of the ten habitat types evaluated, the bar edge habitat type is most abundant by length and number of units. However, its relative importance is reduced when weighted by fish-habitat association characteristics. Preferred habitat types (channel nook, eddy pool and open nook) are frequent and available to aquatic organisms, and most common at welldeveloped bars and in zones of equilibrium long-term sedimentation. Preferred habitat was at a maximum 30 years ago when major new bars developed and the thalweg shifted, effectively increasing the amount of bar shoreline and nearshore habitat. This increase is due to substantial change in river planform morphology following a 30-year period of large annual floods. However, amounts of habitat did not increase exclusively during periods of higher than average flows, or decrease exclusively during periods of lower than average flows. Instead, habitat abundance response to flow may occur with a two- or three-year lag. Short term changes in stage are critical to amount of preferred habitat. Optimal discharge for maximum preferred habitat availability is in the range of approximately 2500 m3 s-1 to 4000 m3 s-1, which approximates long term mean flow. As flow increases, the proportion of preferred habitat compared with total bar shoreline decreases. Comparison ii  with the 2006 flow duration curve shows that 15 – 30 % of discharges are optimal for maximum fish density and biomass. These discharges occurred during April 27 to May 17 and July 14 to August 7, 2006.  iii  Table of Contents Abstract ........................................................................................................................................... ii  Table of Contents ............................................................................................................................iv  List of Tables ...................................................................................................................................vi  List of Figures .............................................................................................................................. viii  Glossary ...........................................................................................................................................xi  Acknowledgments ......................................................................................................................... xii  Dedication .................................................................................................................................... xiii  Chapter 1. Introduction..................................................................................................................... 1  1.1 Habitat Classification .................................................................................................. 1  1.1.1 Definition ........................................................................................................... 1  1.1.2 Issues Associated with Spatial and Temporal Scale .......................................... 1  1.1.3 Literature Review .............................................................................................. 2  1.2 Large River, Stage-Adaptive (LaRSA) Habitat Classification System ....................... 9  1.3 Thesis Objective ........................................................................................................ 12  1.4 Research Hypotheses ................................................................................................. 13  Chapter 2. Study Site ...................................................................................................................... 15  2.1 Gravel Reach, Fraser River, British Columbia .......................................................... 15  2.2 Site Selection ............................................................................................................. 18  2.2.1 Long Reach ...................................................................................................... 18  2.2.2 Historical Reach ............................................................................................... 18  2.2.3 Seasonal Reach ................................................................................................ 20  Chapter 3. Methods ........................................................................................................................ 21  3.1 Data Collection .......................................................................................................... 21  3.1.1 3.1.2 3.1.3 3.1.4 3.1.5 3.1.6  Aerial Photographs .......................................................................................... 21  Geographic Information Systems .................................................................... 23  Habitat Classification ....................................................................................... 24  Ground-truthing ............................................................................................... 27  Fish-habitat Association Data .......................................................................... 27  Observer Variability......................................................................................... 28   3.2 Data Analysis ............................................................................................................ 30  3.2.1 3.2.2 3.2.3 3.2.4  Data Characteristics ......................................................................................... 30  Numerical Analyses: Fish-Habitat Association Indices ................................... 31  Statistical Analyses .......................................................................................... 32  Graphical Analyses .......................................................................................... 33   Chapter 4. Results and Discussion ................................................................................................. 34  4.1 Introduction ............................................................................................................... 34  4.2 Habitat Observer Variability ..................................................................................... 34  iv  4.3 Spatial Habitat Occurrence: Long Reach .................................................................. 39  4.3.1 Long Reach Habitat Abundance ...................................................................... 39  4.3.2 Variability Amongst Individual Bar Complexes ............................................. 46  4.3.3 Variability Amongst 1-km Sedimentation Cells .............................................. 49  4.4 Temporal Change of Habitat: Historical Reach ........................................................ 54  4.4.1 Maps of Temporal Habitat Change .................................................................. 54  4.4.2 Temporal Change of Habitat Trends................................................................ 54  4.5 Stage-dependant Change of Habitat: Seasonal Reach ............................................... 67  4.5.1 Maps of Stage-dependent Change of Habitat .................................................. 67  4.5.2 Stage-dependent Change of Habitat Trends..................................................... 71  Chapter 5. Conclusions................................................................................................................... 77  References ...................................................................................................................................... 79  Appendix A: Fish Species Known to Occupy the Fraser River Gravel Reach ............................. 82  Appendix B: Habitat Type Variations by Year ............................................................................. 84  Appendix C: Maps of Stage-dependent Change of Habitat .......................................................... 86  Appendix D: Data Tables .............................................................................................................. 95   v  List of Tables  Table 1-1. Habitat classification units proposed by Bisson et al. (1982) based on morphological units. ..........................................................................................................................................3  Table 1-2. Summary of habitat unit types developed in studies on non-wadeable rivers (Murphy et al. 1989 and Beechie et al. 2005). .............................................................................................6  Table 1-3. Summary of habitat unit types developed for high water stage (Schwartz and Herricks 2005). .........................................................................................................................................8  Table 1-4. Habitat types corresponding to Level Three of the LaRSA habitat classification for the gravel reach of Fraser River (two-letter abbreviations are given in parentheses). Adapted from Rempel, 2004. Flooded bar top and flooded vegetation are not included. ……………11  Table 2-1. Summary characteristics of morphologically homogeneous sub-reaches located within the gravel-bed reach. Adapted from Church et al. (2000). ..........................................................16  Table 3-1. Summary specifications of vertical air photos used for historical habitat mapping. ........21  Table 3-2. Summary of conditions during 2006 oblique air photography. ........................................23  Table 3-3. Rules for mapping vertical air photos using the LaRSA habitat classification system. ...25  Table 3-4. The proportion of total shoreline length mapped by habitat type for each year of vertical photo data. ...............................................................................................................................26  Table 3-5. List of normalized habitat indices used to weight the length of mapped habitat with fishhabitat association data. The highest values (1.000) are bolded. n indicates the habitat unit population. Two-letter habitat abbreviations are given in Table 1-4. .....................................32  Table 3-6. Summary of correlation strength for Spearman’s rank correlation coefficient. ................33  Table 4-1. Ranked p-values obtained from chi-square tests between the number of habitat units classified by the assistant (considered the expert) and all other habitat classifiers, including the author (“Auth”) at Tranmer and Lower Herrling Bars. Participant 2 did not classify Tranmer habitat, while Participant 3 did not classify Lower Herrling habitat. Preferred habitat types are italicized. ......................................................................................................37  Table 4-2. List of sedimentation classes used in the Fraser River gravel reach sediment budget and amount of each class present in the Long Reach study area. ..................................................49  Table 4-3. Results of chi-square tests for sedimentary environments effects. ...................................50  Table 4-4. p-values from chi-square tests of the number of units per habitat type observed in each sedimentation zone class and the number of units in respective proportional length of the Long Reach study area. p-values showing statistically significant difference are italicized..51  Table 4-5. List of proportions of preferred habitat lengths to total mapped shoreline displayed in Figure 4-19, and their departures from 2006 values. ..............................................................60  Table 4-6. List of r2 values obtained from correlations of flow and departure of proportions of preferred habitat lengths to total mapped shoreline from 2006 values. ..................................64  Table 4-7. Spearman’s rank correlation coefficients and their associated strengths resulting from tests of habitat indices with departure of proportions of preferred habitat lengths to total mapped shoreline from 2006 values. .......................................................................................66  vi  Table A-1. List of fish species known to occupy the Fraser River gravel reach. ...............................83 Tables D-1 to D-10. Data tables used for analysis in this thesis. .......................................................96  vii  List of Figures Figure 1-1. Hierarchical habitat classification structure for the LaRSA habitat classification system and approximate spatial scale of each level (from Rempel 2004). ………………………..…10  Figure 1-2. Schematic diagram of Level Three alluvial habitat types in the gravel reach of Fraser River (adapted from Church et al. 2000). The approximate characteristic spatial scale of the gravel bar is 103 m and of the habitat units is 101 to 102 m. ....................................................12  Figure 2-1. Maximum, minimum, long-term daily-average and 2006 daily-averaged discharges at Hope [08MF005] for the period of 1912 to 2006. ...................................................................17  Figure 2-2. Location map of the Fraser River gravel reach (Rice et al. submitted). Data collection occurred between rkm 105 and 139. The red box shows the boundaries of the Long Reach, the blue box shows the limits of the Historical Reach and the green box show the limits of the Seasonal Reach. (N Bar is called Minto Island in this thesis.)................................................19  Figure 3-1. Example of a georeferenced mosaic of oblique photographs of Tranmer Bar. The view is to the southeast. Photos taken August 31, 2006, at 1605 m3 s-1. ............................................24  Figure 4-1. Distribution of habitat observer variability at Tranmer and Lower Herrling Bars. “Asst” refers to the vertical photo mapper and “Auth” refers to the oblique photo mapper. Numbers 1 to 7 refer to test mappers in rank order of habitat classification experience and are consistent across both bars. Participant 2 did not classify Tranmer habitat, while Participant 3 did not classify Lower Herrling habitat. Legend colours correspond to habitat mapping linework. See Table 1-4 for habitat type abbreviations and Figure 2-2 for bar locations. .....35  Figure 4-2. Long Reach habitat map. Photography taken March 3, 2006 at 695 m3 s-1. Flow is right to left. The reach extends from the downstream end of Wellington Bar at rkm 105 to the upstream end of Herrling Island at rkm 139. ...........................................................................40  Figure 4-3. Measurements of habitat abundance from 2006 air photos. a) Total length (km) and proportion of total length of habitat by habitat unit type, b) total count and proportion of total counts of habitat units by habitat unit type. Preferred habitat types are coloured dark grey. Two-letter habitat abbreviations are given in Table 1-4. ........................................................41  Figure 4-4. Example of two views of Tranmer Bar (July 7, 2006; 4100 m3 s-1). The west (downstream) view in the top panel clearly shows an eddy pool unit at the bar head. No eddy pool is evident in the northwest view of the bottom panel, which was photographed moments after the top panel. ...................................................................................................................43  Figure 4-5. Summary of weighted fish-habitat characteristics for the six core habitat types derived from 2006 air photo habitat mapping. Preferred habitat types are coloured dark grey. Twoletter habitat abbreviations are given in Table 1-4. ................................................................45  Figure 4-6. Distributions observed at each bar unit of a) all habitat type lengths and b) habitat type proportions of total shoreline length. Legend colours correspond to habitat mapping linework. See Table 1-4 for habitat type abbreviations and Figure 2-2 for bar locations. ....47  Figure 4-7 The sum totals of a) preferred habitat type lengths, and b) proportions of habitat lengths to total mapped shoreline observed at each bar unit. Note that Lower Herrling Bar is omitted from the results. Bars are listed left to right with upstream progression. See Figure 2-2 for bar locations. ...........................................................................................................................48  Figure 4-8. The sum totals of proportions of preferred habitat type lengths to total mapped shoreline observed in the four sedimentation classes observed in the Long Reach. ..............................52  viii  Figure 4-9. The sum totals of proportions of each habitat type length to total mapped shoreline observed at a) modest degradation, b) equilibrium, c) modest aggradation, and d) major aggradation zones in the Long Reach. Two-letter habitat abbreviations are given in Table 1-4. ................................................................................................................................53  Figure 4-10. Habitat map of the Historical Reach, 5 December, 1943 (930 m3 s-1). ..........................55  Figure 4-11. Habitat map of the Historical Reach, 31 March, 1949 (650 m3 s-1). ..............................55  Figure 4-12. Habitat map of the Historical Reach, 28 April, 1963 (2130 m3 s-1). .............................56  Figure 4-13. Habitat map of the Historical Reach, 22 March, 1979 (955 m3 s-1). Big Bar and Hopyard Island are newly developed. A new thalweg emerged on the right bank (north) side of the river near Gill Bar. ........................................................................................................56  Figure 4-14. Habitat map of the Historical Reach, 4 September, 1986 (1890 m3 s-1). .......................57  Figure 4-15. Habitat map of the Historical Reach, 20 March, 1999 (700 m3 s-1). ..............................57  Figure 4-16. Habitat map of the Historical Reach, 7 March, 2001 (535 m3 s-1). ...............................58  Figure 4-17. Habitat map of the Historical Reach, 17 December, 2003 (835 m3 s-1). .......................58  Figure 4-18. Habitat map of the Historical Reach, 3 March, 2006 (695 m3 s-1). ...............................59  Figure 4-19. The sum totals of proportions of preferred habitat type lengths to total mapped shoreline observed each year of Historical Reach vertical photography. Results from 1963, 1986, and 2001 are grey-screened and should be interpreted with care because flows at the time of vertical photography were notably different than most in this analysis (Table 3-1)...59  Figure 4-20. Cumulative departure from mean flow curve for Fraser River between 1912 and 2006. Dashed lines shown inflexions in the flow departure. Variation of average active channel width for the gravel reach (Ham, 2004). Normalized time history of the PDO (http://jisao.washington.edu/pdo/). Departures from 2006 values of the ratio of preferred habitat lengths to total shoreline are in the lower panel. …………………………………….61  Figure 4-21. Correlation plots of a) annual peak flows, and b) annual mean flows with departure of proportions of preferred habitat lengths to total mapped shoreline from 2006 values. ...........64  Figure 4-22. Amount of habitat index change relative to 2006 conditions. Data are historical proportions of total shoreline length occupied by the three preferred habitat types, weighted by fish-habitat characteristics. .................................................................................................65  Figure 4-23. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (middle and bottom). 6 August, 2006. 2570 m3 s-1. ..................................................................................................68  Figure 4-24. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (bottom). 16 August, 2006. 2110 m3 s-1. ..............................................................................................................................69  Figure 4-25. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (bottom). 31 August, 2006. 1605 m3 s-1. ..............................................................................................................................70  Figure 4-26. Habitat variation with discharge at Tranmer and Lower Herrling Bars measured by a) assumed length of all units, b) assumed length of preferred habitat units, c) count of all units, and d) counts of preferred habitat units. Diagonally-hatched regions show the range of optimal discharge for maximum habitat abundance. ...............................................................71   ix  Figure 4-27. 2006 flow duration curve at Hope. 1954-1976 average duration curve for historic high flow period, and 1977-2000 averaged duration curve for historic low flow period. The red segments indicate the percentage of time discharge is equalled or exceeded during the range of optimal flows for maximum habitat availability (15-30%). ................................................73  Figure 4-28. 2006 daily-averaged discharge, 1954-1976 daily-averaged discharge for historic high flow period, and 1977-2000 daily-averaged discharge for historic low flow period at Hope. Red segments indicate seasonal timing of discharges with optimal habitat (April 27 to May 17, and July 14 to August 7 for 2006). ....................................................................................73  Figure 4-29. Variation in the proportion of habitat type relative to all habitats over increasing discharge at Tranmer and Lower Herrling Bars measured by a) BA, BE, CB, CN, ON length, b) BH, BT, EP, RI length, c) BA, BE, CB, CN, ON count, and d) BH, BT, EP, RI count. Habitat types are plotted in separate groups to emphasize the ordinate scale. Habitat type abbreviations are defined in Table 1-4. Colours correspond to habitat maps. .......................75  Figure 4-30. The proportion of preferred habitat units relative to all units over increasing discharge, measured by a) length and b) count. ........................................................................................76  Figure B-1. Long term variation in habitat types, measured by proportion of each habitat type relative to total bar margins. ....................................................................................................85 Figures C-1 to C-10. Stage-dependent change of habitat maps of Lower Herrling Bar and Tranmer Bar. ..........................................................................................................................................87  x  Glossary    analytic stereoplotter an instrument that mathematically relates two-dimensional positions on  photographs to their real-world three-dimensional equivalents using standard equations for interior, relative and absolute orientation diapositive  a positive photographic image produced for viewing by transmitted light; a transparency  georeference  a geographic information systems technique used to align raster data sets with other spatial data; the determination of the coordinates, in absolute earth space, of any identified point on Earth’s surface  orthophotograph  an aerial photograph in which image displacement has been removed and distortion due to tilt, curvature, and ground relief has been corrected; a photographic map that can be used to measure true distances  photogrammetry  the science of obtaining accurate measurements and maps from photographs  register  a geographic information systems technique used to apply known, real-world coordinates to a planar map which allows features to be digitized directly in geographic space  RMS  root mean square sum of all residual error in distance between two points; a mathematical measurement comparing the location of the map coordinate in real world space with the transformed position in the raster space  shapefile  a file used in a Geographic Information System that contains geographic features and their data  xi  Acknowledgments I wish to acknowledge my supervisor, Michael Church, for the inspiration that lead me to pursue geomorphology. I also offer him many thanks for the mentorship, scientific insight, thoughtful feedback and eagerness to help that made this thesis possible. Financial support was provided by Géoide (the National Centre of Excellence in Geomatics), awarded to Mike. Dr. Laura Rempel’s cheerful advice and confidence in my abilities were an invaluable contribution to this work. She has my gratitude for her involvement in directing and reviewing this thesis and for her support of my other professional pursuits. A number of individuals have my sincere appreciation for their help in the air, on the ground, on the water and in the office: Brett Eaton, Dylan Hedden-Nicely, Hale Jones-Cox, Dave Luzi, Kathy Macdonald, Jeff Phillips, Laura Rempel, Elisa Scordo, Toby Perkins, Kyle Terry and Andre Zimmermann. Special thanks go to my assistant, Alissa Cullum, who worked in all the above environments with enthusiasm. Sandy Lapsky organized financial and administrative matters and Vincent Kujala assisted with computing concerns. Jose Aparicio gave GIS advice and the GIC staff (Rosemary, Kevin and Jenn) helped with air photos. Big thanks go to my officemate and good buddy, Dave Luzi, for his friendship and his uncanny ability to crack jokes at just the right time. As with all my endeavours, my family played a fundamental role in this chapter of my life. Thank you so much to Mom, Marshall and Mimi for their never-ending support. Big hugs to all the rest of my family spread out over the globe. Finally to Toby, my husband and my best friend, thank you for your encouragement, patience and your faith in me.  xii  Dedication This thesis is dedicated to my father, Ian W. Horne.  xiii  Chapter 1. Introduction 1.1 Habitat Classification 1.1.1 Definition Stream channels are composed of organized erosional and depositional units which, when combined with a certain discharge, lend themselves particular hydraulic properties and morphologies. Physical habitat for instream biota is a function of these hydraulic and geomorphologic properties, which vary longitudinally, laterally and vertically in the river. As a result, the quantity and quality of physical habitat is dynamic in both space and time. In turn, physical habitat is a major determinant of the distribution and abundance of fish and ecosystem productivity. Its characterization through stream assessment and habitat classification leads to a better understanding of limiting factors affecting the distribution and the productive potential of the aquatic community in general. Stream habitat classification is necessary for both research and the effective and efficient management of fish (the resource) and fisheries (activities which affect the resource). Classification systems play several roles. First, they are tools that order, synthesize, inventory and compare habitat characterization data (Church et al. 2000). This organizes variability and emphasizes patterns of significance, such as morphology and/or habitat units. Second, classification systems serve as a “common language” used to make decisions and communicate results to different parties interested in habitat study applications. Last, they act as project design or strategy templates, which aid site selection, field data sampling, and consistency of effort over long-term studies and management. The purpose of this chapter is to review relevant stream habitat classification examples in the literature that provide background to the specific objectives of this research on the spatial and temporal dynamics of nearshore aquatic habitat in Fraser River.  1.1.2 Issues Associated with Spatial and Temporal Scale Although streams vary in size, they retain the same basic morphological units (riffles, pools, and bars) through a large range of scales. In small or wadeable streams, 1  morphological units serve as functional habitat units. However, during high flow stages, habitat units become inundated with water and the distinction between them, both as morphologic units and habitat types, becomes blurred and habitat value varies. This variation is often gradational rather than discrete, particularly on increasing scales, which can make identification of morphologically-based habitat types difficult. As spatial scales increase, morphologically-defined units have less specific value as habitat indicators. Riffle and pool units, for example, still exist, but they are too large to provide habitat function. Water levels may be too high, or flows too strong, because the size of aquatic organisms does not scale proportionately with increased stream size. Instead, smaller habitat divisions are nested within morphological units. In large river systems, like Fraser River, functional habitat units are often defined at bar and bank margins.  1.1.3 Literature Review Very little literature currently exists on large river habitats and their ecological importance to aquatic organisms. The simple reason for such a paucity of information lies in the difficulty to sample both physical and ecological habitat characteristics in fast, deep water. Sampling of adult and migratory fishes is particularly difficult because they generally occur offshore. In a system as large as the lower Fraser River, it can be necessary to use commercial fishing gear to sample such fish populations representatively. Conversely, small streams can usually be waded and practical fish sampling techniques such as electro-fishing and block netting are effective. Stream habitat classification studies became prominent in peer-reviewed literature in the early 1980’s with increasing research on the effects of habitat characteristics and quality on aquatic populations. The habitat classification systems that resulted from these studies were generally applicable only to small, wadeable streams. This section introduces foundational habitat classification systems and discusses them in the context of applicability to large rivers, such as Fraser River. Although stream size classifications exist in the literature (cf. Church 1992), stream size is discriminated as either wadeable or non-wadeable for the purposes of this discussion. 2  Bisson et al. (1982) published a widely cited habitat classification based on three general morphological units: riffle, pool and glide. They attached descriptive information to these morphologies and established a nomenclature for habitat unit types (Table 1-1). These habitat types were field-investigated for salmonid distribution in low flow conditions and it was found that certain species preferred certain habitat types, as Bisson et al. (1982) had defined them. This classification system has been successfully applied to both management and research projects.  Table 1-1. Habitat classification units proposed by Bisson et al. (1982) based on morphological units. Unit Morphology  Riffle  Pool  Glide  Characteristics High bed topography Shallow flow depth Coarse-grained bed material Rapid flow velocities Low bed topography Deep flow depth Fine-grained bed material Low flow velocities  Habitat Types Low gradient Rapids Cascades Secondary channel Backwater Lateral Scour  Trench Plunge Dammed  Relatively shallow flow depth Non-turbulent flow  While the nomenclature of Bisson et al. (1982) is useful for descriptions of habitat on a single, small stream, it is inappropriate for many applications. Frissell et al. (1986) contended that a more useful classification system should consider the changes of a stream (or streams) over formative time scales and the factors that determine physical habitat characteristics, which develop on smaller spatial and temporal scales. Classification variables and the number of unit types should be selected so that they are meaningful to the stream under study, at the time of study, and so that they may vary with specific research or management objectives. The classification will then account not only for the present state, but will also acknowledge a range of possible conditions. Hierarchical methods for stream habitat classification systems were developed in response to the above observations. Frissell et al. (1986) proposed a hierarchical 3  classification framework based on a conceptual view of how streams are arranged in space and change through time. They recognized that smaller channel systems develop within constraints set by the larger-scale watersheds of which they are a part. For example, the morphology of a riffle or pool unit is controlled by the slope of the reach in which it resides and by the sediment and water inputs from the contributing drainage basin. Consequently, Frissell et al. (1986) developed a spatially-nested hierarchy of five levels, which allows the user to focus on a small set of variables at each level that best determines system behaviours and capacities within a relevant spatio-temporal time frame. The two finest classification levels of Frissell et al. (1986), the pool/riffle and microhabitat systems, are most appropriate for habitat studies. Definitions of pool and riffle units are based on Bisson et al. (1982), and the suitability of each as habitat to support aquatic biota is defined by their characteristic flow velocities, depths and sediment dynamics. The specified unit scales are suitable for small stream systems. Each pool or riffle habitat unit was characterized by a sequence of spatially associated microhabitat subsystems, which had relatively homogeneous velocities, water depths and substrate types (Frissell et al. 1986). This homogeneity makes a microhabitat subsystem appear simplified when compared to its larger-scale counterparts, and therefore demonstrates a potential for distinctive ecological communities. The microhabitat subunit is most susceptible to change resulting from variable flow conditions and sediment regime compared with higher hierarchical levels (Frissell et al. 1986); thus it is most sensitive in space and time domains. However, the organisms that are the focus at this division limit the possibility to apply the system to large, non-wadeable streams like Fraser River. While this division is appropriate for micro-organisms, such as benthic invertebrates, it is too fine for fish. Hawkins et al. (1993) later published a hierarchical stream habitat classification which described units in three levels of increasing detail. The work was complementary to a hierarchical classification of morphological channel units presented earlier by Sullivan (1986). All levels can be visually identified in the field. Level 1, the coarsest level of resolution, distinguishes between primary geomorphologic channel unit types, riffles and pools. These units represent distinctly different ecological habitats because the species 4  inhabiting these units have differing taxonomy and morphological, physiological, and behavioural characteristics (Hawkins et al. 1993). Level 2 separates geomorphologic units into sub-classes, which again are similar to the habitat type descriptions presented by Bisson et al. (1982). The pool sub-classes are distinguished by scour or damming formation, and riffle sub-classes are based on the presence or absence of turbulence (although they would be described better as having “smooth” or “rough” flows since all streams in nature are turbulent). This subdivision further refines physical and biological functions of the units in that, for example, dammed pools tend to have greater amounts of cover than scour pools because they are formed behind wood, debris, or large substrates, while the abundance of riffle-dwelling benthic organisms is affected by the intensity of turbulence (Hawkins et al. 1993). Level 3, the finest resolution, implicitly codifies additional information about riffles and pools, such as mean velocity, longitudinal and cross-sectional profiles and substrate characteristics. Explicitly, it describes morphological and/or genetic variants. This level also adopts some of the nomenclature used by Bisson et al. (1982). This system allows the user to choose the resolution necessary to meet specific objectives and to make comparisons between streams of similar spatial scale. It was adopted for the Fish Habitat Assessment Procedures of the Province of British Columbia as an example of hierarchical classification in an attempt to correctly identify fish productivitylimiting habitats such that they could be compared to those in similar watersheds (Johnston and Slaney 1996). Nonetheless, this particular system does not scale to identification of habitat units in large river systems. Geomorphologic channel units become proportionately larger with increases in stream size, but the boundaries between them may become less obvious as water depth, flow velocities and substrate size increase as well. Therefore a channel unit that is geomorphologically significant, such as a riffle or pool, may not possess the same habitat characteristics in a large stream as it does in a small stream. Channel units in large streams can consist of many smaller scale habitat patches that are physically and biologically equivalent to entire stream channel units in small streams (Hawkins et al. 1993). The stream habitat classification systems discussed thus far, have been developed for small, or wadeable, streams. In larger, non-wadeable rivers, habitat utilized by juvenile fishes is typically studied because successful juvenile rearing is the key to sustaining 5  populations. Two peer-reviewed studies have attempted to classify stream habitat and describe habitat utilization by juvenile salmonids in non-wadeable rivers (described as large rivers by the respective authors). Both classification efforts were river system-specific. Murphy et al. (1989) distinguished between active channel and off-channel habitat types for the glacial Taku River, Alaska, while Beechie et al. (2005) isolated habitat units in mid-channel and channel margin regions for Skagit River, Washington. From these spatial conditions, habitat units were classified based primarily on water velocities, but also on bed morphologies in active channel types (Table 1-2). Neither study was able to adequately sample the active channel so each presented results for channel margins only.  Table 1-2. Summary of habitat unit types developed in studies on non-wadeable rivers (Murphy et al. 1989 and Beechie et al. 2005). Authors  Location  Murphy et al. 1989  Taku River Alaska A: 16,000 km2  Channel Type  Habitat Unit Name  Habitat Unit Description  Active Channel  Main channel Braid Channel edge Slough Backwater Terrace tributary Tributary mouth Beaver pond Upland slough Pool Riffle Glide Bank edge Bar edge Backwater  Velocity >30 cm s-1 Velocity 10-3 cm s-1 Velocity <30 cm s-1 Velocity 0-15 cm s-1 Slack water, ~0 cm s-1 Stream flowing to main channel Lower reach of tributary; slack water Impounded tributary Slough with outlet to river Scoured bed depression High velocity, turbulent No bed depression, little turbulence Near vertical shore Low gradient shore Low velocity, partially enclosed  Offchannel  Beechie et al. 2005  Skagit River Washington A: 8,017 km2  Midchannel Channel Margin  A: drainage area  Murphy et al. (1989) adapted habitat classification systems defined by the Alaska Department of Fish and Game and the Unites States Department of the Interior, National Parks Service, for their study reaches on Taku River. They tested physical characteristics of habitat types and found statistically significant differences between habitat units in water velocity, water depth, turbidity, amount of large woody debris and water temperature. They also found that juvenile salmon distribution was most closely related to water velocity, and 6  secondly to turbidity. All juvenile salmonids were virtually absent from areas with currents > 30 cm s-1, which corresponded to main channel velocities (Table 1-2). Beechie et al. (2005) tested for differences in habitat characteristics (velocity, substrate and cover) among unit types for three different time periods, winter, end of spring and end of summer, at low flow conditions. They found that bar and bank edge units had similar velocity distributions, which were faster than backwaters and within the range preferred by most juvenile salmonids. Mid-channel units had velocities that exceeded the preferred range (> 45 cm s-1). Mean depths differed significantly among habitat units and juvenile salmonids selected mean depths most commonly found in edge units. For each respective non-wadeable stream, both Murphy et al. (1989) and Beechie et al. (2005) isolated the thalweg from shoreline areas and successfully identified unique habitat types within these two zones (Table 1-2). The identified shoreline habitat types were not strictly geomorphologic units, as described above for wadeable streams (although Beechie et al. (2005) used riffle, pool and glide terminology for mid-channel units), but instead were distinguished by flow characteristics found at channel margins. Habitat unit types, their areal extents and their positions change throughout the year with varying hydrograph stage and associated water levels. Therefore, stage-specific habitat mapping is necessary in order to adequately describe habitat utilization as fish relocate with changes in flow. For example, increased flows may require juvenile fishes to take refuge in sheltered, low-velocity habitat types. Conversely, during low discharge conditions, fish may seek out relatively deep areas as habitat availability becomes compromised. Indeed, Huntington et al. (1999) found that, during simulated drought conditions in a flume setting, 72% of juvenile Atlantic salmon residing in riffle habitats moved upstream to an available pool and 8% moved downstream to an available pool. Although their study was conducted with the purpose of investigating fish behaviour, it can be extrapolated to trends in habitat utilization. In terms of habitat classification, it is important to quantify how much habitat and which habitat unit types exist at flows suitable for unstressed fish behaviours. Flow patterns are more complex at high flows than at low flows, especially in small, wadeable streams, so a habitat classification system should address this complexity. Schwartz and Herricks (2005) identified three main principles that apply to development of a 7  framework for high flow habitat classification. These principles are as follows: (1) The hydraulic environment changes as stage increases such that depth-averaged downstream velocities converge in greater magnitude in both riffles and pools, (2) three-dimensional flow patterns depend on hydraulic resistances associated with all morphological surfaces, including large in-channel features (such as riffles or pools), and (3) flow masses are separated by distinct hydraulic boundaries between the high-velocity thalweg and lateral areas (Schwartz and Herricks 2005). Based on this reasoning, they developed a habitat classification system which incorporates unique habitat types for flood stages that inundate lateral transition areas between the main channel and the floodplain, and for flood stages that inundate floodplains (Table 1-3). Statistically significant differences in fish utilization of each habitat type were demonstrated.  Table 1-3. Summary of habitat unit types developed for high water stage (Schwartz and Herricks 2005). Flow Stage Lateral area inundated  Habitat Type Channel thalweg  Maximum depth and velocity flow path; Convergent-divergent flow path  Deflection Eddy  Recirculating flow; Intermediate bed elevation at margin of main channel at point bar face Recirculating flow; Zone of channel width expansion Stagnant or low-velocity areas produced by high hydraulic roughness; Channel margin location  Expansion Eddy Local dead zone Floodplain inundated  Habitat Characteristics  Vegetated point bar  Near zero velocity or recirculating flow; Elevated floodplain platform: Grassy vegetation  Concave-bank bench  Recirculating flow resulting from enlargement of expansion eddy; Elevated surface  Remnant channels  Recirculating flow resulting from enlargement of expansion eddy; Elevated surface; Gradual sloping re-entrance  8  In summary, four key challenges surrounding stream habitat classification have arisen: 1. One universal habitat classification system is not appropriate for all streams; 2. Hierarchical habitat classification systems have the advantage of allowing analysis over a range of detail and spatial scales because of a choice in resolution; 3. Wadeable stream classification systems are not transferable to non-wadeable streams because the characteristics of morphologically-based habitat units differ, and; 4. Habitat unit types should be applicable across all flow conditions, although the location of individual units may change with stage.  These challenges are satisfied in the large river, stage-adaptive (LaRSA) habitat classification system developed specifically for the gravel reach of Fraser River by Rempel (2004) as a part of her doctoral dissertation. (The LaRSA acronym was created for this thesis and does not appear in Rempel (2004).) The system was first documented in a report prepared for Fisheries and Oceans Canada (Church et al. 2000). However it was further refined and tested following the 2000 report and therefore is cited as Rempel (2004) with the exception of occasional specific references.  1.2 Large River, Stage-Adaptive (LaRSA) Habitat Classification System Laura Rempel’s doctoral dissertation, which was prepared in the Department of Geography at the University of British Columbia, applied a hierarchical concept to stream morphology and habitat classification to the gravel reach of Fraser River, British Columbia. Her large river, stage-adaptive (LaRSA) habitat classification system uses three levels of increasing detail (Figure 1-1) and addresses unique applications for each level. Level One is a reach-scale classification that can be used in strategic planning for fisheries management. Level Two is a gravel bar unit-scale classification that can support field studies and operational management of fisheries. Information for these two levels can be gained from map analysis and air photograph interpretation. The finest level of classification, Level 9  LEVEL ONE  LEVEL TWO  LEVEL THREE  Reach Scale  Bar Unit Scale  Habitat Unit Scale  >104 m  103 m  101 - 102 m  Figure 1-1. Hierarchical habitat classification structure for the LaRSA habitat classification system and approximate spatial scale of each level (from Rempel 2004).  Three, identifies discrete habitat units around the margins of bar complexes and channel shores. Many habitat units exist within a reach, and each is classified uniquely as a specific habitat type. The Level Three habitat classification is of primary interest for this thesis. Ten alluvial habitat types and three bank types of relatively homogeneous physical character were identified as Level Three units in the field (Table 1-4). The habitat units potentially exist at all stages of the hydrograph throughout the gravel reach but their locations shift with changing water levels. The spatial scale of individual units was intended to be ecologically relevant to aquatic organisms, particularly juvenile fish. Habitat types are physically distinct and possess moderately distinct and differentiated assemblages of both fish and invertebrate taxa (Rempel 2004). A schematic diagram of alluvial nearshore habitats is given in Figure 1-2. In terms of aquatic species utilization of the LaRSA habitat units, most fish species were distributed amongst all habitat types, although for most species, highest densities were associated with a limited number of habitat types. Most notably, three habitat types (eddy pool, open nook and channel nook) proved to be particularly unique because relatively distinct species composition was observed. This suggests that they offered distinct functional opportunities to rearing fish. Open nooks and channel nooks also had higher overall fish densities than other habitat types. Finally, benthic invertebrate community structure showed modest dissimilarity among Level Three habitat types. Two additional potential habitat types were field-identified by reconnaissance of the study areas during the spring 2006 freshet, which was shortly before data collection for this thesis commenced. These habitats are not part of Rempel’s (2004) LaRSA classification system and have not been field-sampled, but they were included in this thesis for stagedependent mapping of oblique air photos to make qualitative descriptions of their changes. 10  Table 1-4. Habitat types corresponding to Level Three of the LaRSA habitat classification for the gravel reach of Fraser River (two-letter abbreviations are given in parentheses). Adapted from Rempel, 2004. Flooded bar top and flooded vegetation are not included.  Habitat Type  Definition  Bay (BA)  Semi-enclosed area with no flow velocity and fine bed material (sand/silt). Occurs on the lee side of large sediment accretions that are deposited in the shape of a crescent-dune.  Bar Edge (BE)  Any length of bar edge not occurring at the head or tail of a bar that is oriented parallel to the flow and subject to constant and consistent flow forces. Banks slope is variable and a range of velocities and substrate types is possible. Riparian influence is variable.  *  *+  Upstream end of a gravel bar. Surface substrate is characteristically coarse and flow velocity is usually high.  *+  Bar Head (BH) Bar Tail (BT)  Downstream end of a gravel bar, usually with moderate flow velocity. The habitat is often depositional and surface substrate consists of smaller cobbles and gravels.  *+  *+  Channel Nook (CN)  Eroding bank of fine sediment that is steeply sloped or vertical. Dense riparian vegetation is often present. Large woody debris is common and flow is variable.  *  Cut Bank (CB)  Eddy Pool (EP)  Riffle (RI)  Area bound by fast rough water that creates a back eddy in the lee of the flow. Common on the inside edge of riffles and at the upstream end of some bar head habitats. Bank slope is invariably steep and the substrate is usually embedded cobble.  *+  Open Nook (ON)  Dead-end channel or narrow embayment of standing water and concave geometry. Substrate material usually consists of sand/silt and embedded gravel.  Shallow indentation along a bar edge of reduced velocity and variable substrate that is openly connected to the channel with no sedimentary barrier (unlike channel nook). An ephemeral habitat that often disappears with a relatively small change in water level.  *+  High-gradient area of shallow, fast water flowing over well-sorted substrate that often has granular structures and is stable. The flow is rough. Common at bar heads.  *  Natural rock bank, possibly with openings and cracks, that is invariably steep. The water is deep immediately offshore and currents are either fast or form a back eddy.  Rock Bank (RB)  Artificial Bank (AB)  *  Bank is invariably steep and consists of riprap or rubble rock that may have significant openings within its structure. The water is usually deep and fast immediately offshore, particularly at high flow.  Bar top (BP)  Bar top surface inundated only during high flow with reduced velocity and shallow water depth relative to open water and the thalweg. Substrate is variable.  Vegetation (VG)  Area of flooded island of bank vegetation where velocity is reduced and substrate is relatively fine. Submerged only at very high flow.  * Habitat unit types mapped in this study. +  Habitat unit types used in fish-habitat index analyses.  11  Figure 1-2. Schematic diagram of Level Three alluvial habitat types in the gravel reach of Fraser River (adapted from Church et al. 2000). The approximate characteristic spatial scale of the gravel bar is 103 m and of the habitat units is 101 to 102 m.  Flooded vegetation (FV) units are areas inundated at high discharges and may serve as flow refugia when other “sheltered” habitat types become unavailable. Flooded bar top (FB) units represent ponded areas that develop as stage drops as well as depressions that become isolated from the flow.  1.3 Thesis Objective The main objective of this thesis is to employ the LaRSA habitat classification system to characterize the variation in space and time of Fraser River gravel reach habitat units. While the system was tested previously for physical and ecological relevance following an intensive field-based development program (Rempel 2004), it has not been applied to determine reach-wide habitat distribution and trends. The results of this research will contribute to a large suite of projects being completed by the Fraser River Research Group at  12  the University of British Columbia’s Department of Geography, which are aimed at contributing to management of the reach. More generally, this thesis seeks to advance the characterization of physical habitat in large rivers. No known research has investigated the spatial distribution of aquatic habitat at the scale of Fraser River’s gravel reach, or the historical changes in habitat over a period of decades. Although some recent efforts have examined the issue of habitat classification over a range of discharge and associated water level conditions, these studies were limited to flumes and small stream settings (Huntington et al. 1999; Schwartz and Herricks 2005). The specific goals of this work are to quantify the distribution and abundance of physical habitat 1. spatially, at the reach scale (32 km); 2. temporally, through a historical period of about 60 years, and; 3. temporally, throughout the seasonal stage variation of the 2006 hydrograph descending limb.  To address these goals, I map current and historical configurations of nearshore habitat types as defined by Rempel (2004), then measure and compare relative amounts of habitat. Further details about methods are presented in Chapter 3.  1.4 Research Hypotheses Upon beginning this research program, I established two research hypotheses. Hypothesis 1: The amount of habitat will show little variation along the reach and throughout historical time, however the distribution of units will vary. This statement is based on the idea that, although bar features evolve and shift laterally across the stream channel, the amount of secondary channels and bar edge features should remain relatively consistent. Secondary channels and channel margins provide valuable habitat in large rivers like the Fraser because, in the main channel, aquatic 13  organisms expend significant energy to cope with high water velocities and a limited food supply. Notably, Fraser River gravel reach secondary channels were identified as important spawning habitat for endangered white sturgeon (Perrin et al. 2003). Hypothesis 2: With rising water level, the total amount of habitat will decrease and only certain habitat types will be observed. As water level rises, the channel margin shifts laterally across gravel bars and inundates available habitat. Although all LaRSA habitat units have a likelihood of occurring at all water levels and at all locations, certain habitat types, such as shallow open nooks, will become scarce or absent at high stages, while others will account for a higher proportion of total bar shoreline. In particular, cut banks and the less morphologically complex bar edge habitat units will become more prevalent as water levels rise to, and flood bar top vegetation.  14  Chapter 2. Study Site 2.1 Gravel Reach, Fraser River, British Columbia Fraser River drains approximately 230,000 km2 of south-central British Columbia. Its 1375 km length originates in the Rocky Mountain trench, flows through the Columbia and Rocky Mountains, the Interior Plateau, and the Coast Ranges, and terminates at the Strait of Georgia. Lower Fraser River is measured upstream from river kilometre (rkm) zero at the mouth to rkm 190 at Yale. As the river emerges from the mountains at Hope, it encounters a sharp decrease in topographic gradient and deposits its cobble- and gravel-sized coarse sediment load on a partially confined alluvial fan. The length of river that flows on this fan is known as the gravel reach because its bed is composed of these coarse, alluvial sediments. The gravel reach extends approximately from rkm 100 to 165. The river flows onto a sand bed at Sumas Mountain (rkm 100) near Mission. The gravel reach is characterized by a wandering morphology (Desloges and Church 1989). The term “wandering” was coined because of the river’s irregular, lateral channelshifting pattern. Both anastomosing and braiding channel morphologies occur: large, vegetated mid-channel islands frequently divide the channel and, during low flow conditions, relatively stable gravel bars are exposed at mid-channel, lateral and point locations. The resulting morphological configuration presents a complex network of channels that may or may not convey flow at various stages throughout the year. However, a principal channel is always clearly identifiable. The network provides aquatic habitat of exceptionally high quality and its associated fishery is of high ecologic and economic value (Rosenau and Angelo 2007). Bar growth occurs as sediment “sheets” migrate along the channel and become attached or incorporated into the bars (Church et al. 2001). The sheets are a result of the transfer of material to the active channel from upstream channel migration or bank erosion. The new bar shape then diverts the flow toward subsequent downstream banks and initiates further erosion and channel instability. At Agassiz, the annual bed load transport is estimated  15  to be about 227 x 103 t yr-1. Significant gravel transport occurs at flows greater than 5000 m3 s-1 (McLean et al. 1999). Five morphologically homogeneous sub-reaches, which correspond to Level One of the hierarchical habitat classification, have been identified within the gravel reach (Church et al. 2000). Each sub-reach is characterized by a distinct gradient, grain size, aggradation tendency, and morphology (Table 2-1). Each also provides distinctive distributions of fish habitat (Rempel 2004). The data for this study were collected in the Chilliwack, Rosedale and Cheam sub-reaches. The Chilliwack and Rosedale sub-reaches are aggrading, whereas the Cheam sub-reach is degrading.  Table 2-1. Summary characteristics of morphologically homogeneous sub-reaches located within the gravelbed reach. Adapted from Church et al. (2000).  Name  River km  Mean Gradient  Mean Grain Size (mm)  Aggradation Tendency  Sumas  89 - 100  0.000085  sand - 16  degrading  Chilliwack  100 - 118  0.00018  26  strong aggradation  channel bars with subordinate islands  Rosedale  118 - 130  0.00047  40  strong aggradation  island-bar complexes; channel commonly divided; laterally unstable  Cheam  130 - 149  0.00052  50  mild degradation  major islands with surrounding bars; single dominant channel  Hope  149 - 165  0.00055  nd  stable  single thread; cobble-gravel channel with stable lateral bars  Major Features single thread; gravel-sand transition; bars submerged  nd: no data  The Fraser River basin provides habitat for 53 fish species, including marine and introduced species. However, the basin is considered taxonomically sparse as a result of refuge disconnectivity during glaciation (Northcote and Burwash 1991). The gravel reach supports at least 28 fish species and is considered more species-rich than all upstream reaches (Rempel 2004). A list of species known to utilize the gravel reach appears in Appendix A. 16  Fraser River’s annual runoff pattern is dominated by a snowmelt freshet, with peak flow typically occurring in late May through early June. The main channel of Fraser River is unregulated and two major tributary confluences exist in the gravel-bed reach: Harrison River enters from the north (rkm 115) and Chilliwack River enters from the south (rkm 100). Measured at Hope (Water Survey of Canada hydrometric station 08MF005). The mean annual flow is 2700 m3 s-1 (1912 to 2006) and the mean annual flood is 8645 m3 s-1, while the measured flood of record reached 15,200 m3 s-1 in 1948. The 2006 data collection season had slightly lower than average flows. Peak flow occurred on May 27 and measured 8020 m3 s-1, while mean annual flow was 2144 m3 s-1 (Figure 2-1).  16 Max & Min 14  2006 Long-term daily-average  Discharge (103 m3 s-1)  12  10  8  6  4  2  0  Figure 2-1. Maximum, minimum, long-term daily-average and 2006 daily-averaged discharges at Hope [08MF005] for the period of 1912 to 2006.  17  2.2 Site Selection Three study reaches located within the Fraser River gravel reach were selected for the three components of this study, as described in Section 1.3. The reaches are herein referred to as the “Long Reach”, the “Historical Reach” and the “Seasonal Reach”. The Historical and Seasonal Reaches are nested within the Long Reach, but do not overlap (Figure 2-2).  2.2.1 Long Reach The Long Reach extends from the downstream end of Wellington Bar at rkm 105 to the upstream end of Herrling Island at rkm 139 (red box, Figure 2-2). It spans part of the Chilliwack sub-reach, all of the Rosedale sub-reach and part of the Cheam sub-reach. This length of river was selected for reach-scale spatial analysis (Goal #1) because it includes the largest and most highly developed gravel bar complexes in the gravel reach. Both upstream and downstream of the Long Reach, the river is generally a single-thread channel (Table 2-1). The Long Reach has also been the object of intensive research by the Fraser River Gravel Reach Studies Group at the UBC Department of Geography for both academic and public sector studies.  2.2.2 Historical Reach The Historical Reach is nested within the Long Reach and extends from the downstream end of Gill Island at rkm 121 to the upstream end of Big Bar at rkm 128 (blue box, Figure 2-2). It lies entirely within the Rosedale sub-reach (Table 2-1). This particular reach was selected for historical study because it is the best example of wandering morphology along the river: it is historically unstable and has experienced much lateral change in channel morphology over the period of photographic record (Church and Ham 2004). Many secondary channels have evolved with time and some of its bar complexes are relatively young. Most notably, Big Bar began developing in the early 1970’s, following downstream expansion of Ferry Island (Church and Ham 2004). A reach that experiences high morphodynamic evolution was desirable for this component of the study because the 18  19  Figure 2-2. Location map of the Fraser River gravel reach (Rice et al. submitted). Data collection occurred between rkm 105 and 139. The red box shows the boundaries of the Long Reach, the blue box shows the limits of the Historical Reach and the green box show the limits of the Seasonal Reach. (N Bar is called Minto Island in this thesis.)  objective was to describe changes in distribution and abundance of physical habitat over time. Little morphodynamic activity would not adequately capture potential for changes to habitat. Further, this reach was also most frequently captured in the available aerial photography resources.  2.2.3 Seasonal Reach The Seasonal Reach includes Lower Herrling Bar and Tranmer Bar between rkm 132 and 137 (green box, Figure 2-2). It lies entirely within the Cheam sub-reach. Lower Herrling Bar has shown modest degradation at its upstream end to equilibrium sedimentation at its downstream end between 1952 and 1999 (Church et al. 2001). Tranmer Bar has experienced equilibrium sedimentation at its upstream end and major aggradation at its downstream end between 1952 and 1999 (Church et al. 2001). Although Tranmer Bar is therefore growing and Lower Herrling Bar is eroding, the effect was considered to be negligible over the falling limb of the 2006 hydrograph during which period data for this study were collected. Tranmer Bar and Lower Herrling Bar were chosen for the seasonal component of this study because they are known to be some of the oldest bar complexes in the Fraser River gravel reach (Church and Ham 2004). It was assumed that increased age is related to increased complexity of bar surface topography (M. Church 2006, pers. comm.). Therefore a maximum amount of change in exposed bar area, form and surface topography with falling water levels was expected at Tranmer and Herrling compared with younger, less developed bars, such as Big Bar. Finally, a boat launch was conveniently located on the right channel bank near Tranmer Bar, which allowed for easy access to both Tranmer and Lower Herrling for ground-truthing of habitat maps interpreted from aerial photography. It was desirable to minimize river travel by boat because navigation of the gravel reach can be difficult. Swift water conditions can be dangerous and shallow areas, such as emerging riffles or bar surfaces, can be difficult to distinguish from the deeper channel.  20  Chapter 3. Methods 3.1 Data Collection 3.1.1 Aerial Photographs The photographic record of the Lower Fraser River is extensive. The Department of Geography at the University of British Columbia possesses a comprehensive archive of vertical aerial photographs (herein called air photos) at its Geographic Information Centre. The archive consists of photos purchased from the BC Ministry of Agriculture and Lands and its predecessors, the Canadian federal government, and those obtained from privately contracted flights for research projects, including flights contracted by the Fraser River Gravel Reach Research Group. A catalogue was compiled of all available hard copy vertical air photos depicting the study reaches. For the Historical Reach study, nine large-scale photo sets were selected from the catalogue based on similarity of low flow conditions (Table 3-1). 1963 and 1986 air photos were captured at greater discharges than the rest of the data set; therefore measurements are drawn from them with less confidence. March 2006 photography was chosen for the Long Reach study because it was the most current and was taken at low flow.  Table 3-1. Summary specifications of vertical air photos used for historical habitat mapping. Scale  Discharge (m3 s-1)  Flight Line  December 5 March 31  1:15,000 1:19,000  930 650  A7077/8 BC 720  April 28 March 22  1:13,000 1:12,000  2130 955  BC 5062 BC 79003  1999  September 4 March 20  1:12,000 1:40,000  1890 700  BCC 536 BCB99001  2001 2003  March 7 December 17  1:12,000 1:12,000  535 835  SRS 6348 SRS 6906  Year 1943 1949 1963  +  1979 1986  +  +  Date  2006 March 3 1:15,000 695 SRS 7250 Air photos taken at discharges too great for accurate habitat comparisons.  21  Each vertical photo measures 23 x 23 cm inclusive of a 1-cm border on all sides. Photos were scanned at 1200 dpi resolution (8-bit greyscale) on a HP Scanjet 4850 scanner with a 22 x 31.5 cm bed. Since the scanner bed was narrower than the photo width, minor distortion error at the edges of the digital images may have been introduced during scanning because the photos did not sit flush on the scanner bed surface. Approximately 5 to 8 cm was cropped from two opposite sides of each photo to reduce this effect and to remove the 1-cm border. Minor distortion error inherent in the hardware may also have been transferred to the scanned images. It is assumed that error associated with scanning had a negligible effect on habitat unit lengths measured from the scanned vertical air photos (B. Klinkenberg 2005, pers. comm.). Air photos captured at oblique angles relative to the vertical dimension (herein called oblique air photos) were used for the seasonal component of the study (Table 3-2). The study design required that air photos be obtained at approximately 500 m3 s-1 flow intervals. It was prohibitively expensive to contract vertical air photo flights, therefore a Cessna 150 aircraft was chartered from Chilliwack airport. The author captured oblique air photos from outside the open window of the aircraft at a flying height of approximately 2000 ft using a Nikon D100 digital SLR camera with a 28 mm lens on automatic setting. The straight flight line passed the north and south sides (right channel bank and left channel bank, respectively) of Tranmer Bar and Lower Herrling Bar. Each photo had a large amount of image overlap with successive photos to ensure successful georeferencing 1. Several passes in a lateral direction were also flown to capture multiple views of the bar, ensuring a thorough photo survey that would increase the accuracy of habitat unit classification.  1  See glossary for definitions of bolded terms.  22  Table 3-2. Summary of conditions during 2006 oblique air photography. Date  Discharge (m3 s-1)  Weather  Time of Day  June 9 June 23 June 29 July 7 July 19 July 24 August 6 August 16 August 31 September 21  6785 5245 4545 4070 3720 3115 2570 2110 1605 1095  mostly sunny, patchy low elevation cloud sunny, clear sunny, high elevation haze sunny, clear high cloud, low elevation haze sunny, low elevation haze sunny, clear sunny, dense low elevation haze sunny, clear mostly cloudy, low elevation cloud  14:00 13:00 12:00 13:00 12:00 12:00 13:00 13:00 10:00 14:00  October 13  735  sunny, clear  13:00  3.1.2 Geographic Information Systems All Geographic Information Systems (GIS) analyses were executed in ESRI™ ArcGIS (version 9.0). All scanned vertical air photo sets were georeferenced from ground control points located on 1:20,000 TRIM (Terrain Resource Information Mapping) orthophotographs distributed by the Province of British Columbia, Ministry of Sustainable Resource Management. On average, 15 floodplain-level ground control points were matched between photos sets. RMS variation was generally less than ± 4 m. This amount of error is small relative to the dimensions of most habitat units, but may have affected the measured length of smaller units such as bar heads and bar tails. However, these habitat types are not critical to this study because they were found to be less important in terms of fish utilization during the development and testing of the habitat classification system (Rempel 2004). Digital georeferenced vertical photos were available for 1999, 1949 and 1943 from previous research conducted by Darren Ham (Ham 2005). The 1949 and 1999 photo sets were georeferenced using an analytic stereoplotter. Ground control points were obtained from 1987 1:70,000 TRIM diapositive air photos. The 1943 photo set was registered to 1949 mapping in a GIS, resulting in less than ±10 m variation in RMS for all points. Ham 23  (2005) suggested this degree of error was small relative to the size of the channel (mean width = 900 m), but would affect the accuracy of quantitative comparisons of channel change over time. This larger error would also affect the accuracy of habitat length measurements. The oblique air photos could not be registered to real-world coordinates because each was captured at slightly different angles with respect to the lateral, longitudinal, and most importantly vertical dimensions. For this reason, it was not possible to measure habitat unit lengths for the seasonal component of the study. Instead, the number of times a habitat type was mapped (called counts herein) was recorded. Georeferencing of oblique photos was instead performed by tying successive photos to upstream and downstream ends of a photo that displayed a centrally located position on the bar of interest. This process produced severe distortion at the upstream and downstream ends of the final image and therefore presented difficulty for mapping habitats located at the extreme ends of the bars (Figure 3-1).  Figure 3-1. Example of a georeferenced mosaic of oblique photographs of Tranmer Bar. The view is to the southeast. Photos taken August 31, 2006, at 1605 m3 s-1.  3.1.3 Habitat Classification Habitat types, as defined in Level Three of the LaRSA habitat classification system (Table 1-4), were mapped on all air photos by digitizing within the GIS. Artificial bank units were mapped on 2006 Long Reach photography only. Flooded bar top and flooded 24  vegetation were observed and mapped on oblique (Seasonal Reach) photography only. On vertical air photos, habitats located on mid-channel bars, side bars, and river bank edges were mapped, and mid-channel bar perimeters and total water’s edge lengths were also digitized. On oblique air photos, only bar shorelines were mapped. Mapping proceeded according to a set of rules defined by the author and a GIS assistant, which are listed in Table 3-3. Once mapping was complete, relevant characteristics for each shapefile were calculated and exported for analysis in spreadsheet software.  Table 3-3. Rules for mapping vertical air photos using the LaRSA habitat classification system. Rule  Habitat Classification Criteria  1  Habitats less than 5 m in length were not classified  2  Exposed bar surfaces less than 10 m in the longitudinal direction were not classified  3  An unclassified <10 m buffer was allowed between habitat units where boundaries were transitional or inaccurately known  4  Adjacent habitats of the same type were classified individually where distinct boundaries between them were observed, eg. multiple open nooks  5  Scalloped bar edge was classified as bar edge, not as a series of small open nooks  6  Bar heads could exist in locations other than at the upstream end of a bar if they deflected water in two directions  The author mapped oblique air photos and the assistant mapped vertical air photos. Some discrepancies between the author’s and the assistant’s mapping resulted. In particular, there was disagreement between vertical cut banks and steep-angled bar edges, and wide channel nooks and bays. Furthermore, eddy pools were difficult to identify without adequate sunlight reflection off the water surface. In order for the habitat classification system to be useful, multiple users should be able to apply it effectively. Therefore, rather than editing one mapper’s work to precisely match the other’s, an observer variability exercise was performed. The methods for this exercise are discussed in Section 3.1.6. Error due to scanning and georeferencing was discussed in Section 3.1.2; however error in terms of both accuracy and precision is also inherent in habitat classification. Accuracy refers to the degree of conformity to the real or true value, which in this case 25  addresses whether the unit was classified as the correct habitat type and whether the boundaries of the habitat unit are digitized on the air photos to the correct corresponding realworld locations. Accuracy, in terms of correct habitat type classification, was increased in the seasonal study by ground-truthing habitat maps (Section 3.1.4). It was not possible to ground-truth maps of the Long or Historical Reaches. In terms of accurately locating habitat unit boundaries, a buffer length of bar edge was left between classified habitat units (Rule 3, Table 3-3). The length of each buffer increased with uncertainty of the habitat unit boundaries, but never exceeded 10 m. Proportions of total shoreline length mapped are summarized in Table 3-4. Table 3-4. The proportion of total shoreline length mapped by habitat type for each year of vertical photo data. Year  Proportion  1943 1949 1963 1979 1986 1999 2001 2003  0.989 0.986 0.976 0.981 0.982 0.982 0.984 0.984  2006  0.982  Average  0.983  Precision refers to reproducibility, which in this study addresses whether a habitat unit would be classified as the same type by several users. The application of this concept is discussed in the observer variability exercise results (Section 4.2). The Long Reach and Historical Reach components of the study use the proportional length of bar margin as a measure of the amount of habitat. Length was chosen as the most appropriate measure to compare habitat abundance both numerically and statistically amongst unit types; however there are problems associated with using it. For example, the fish-habitat association value may be over-estimated for semi-enclosed habitat types, such as channel nooks and bays in comparison with linear types like bar edge or cut bank, due to a large length to area ratio. However, assigning an area to a linear habitat type has little 26  significance to fish-habitat associations, except in the context of the width of the beach seine that was used to capture fish for confirmation of the classification system (Section 3.1.5). Ultimately, using an area measurement for some habitat types and a length measurement for others would produce incomparable results.  3.1.4 Ground-truthing Ground-truthing of seasonal study habitat maps was performed within two to three days after oblique air photos were captured. It was crucial to visit the field sites as soon as possible after acquiring the air photos so that water levels would be similar to those depicted in the photos. In the period immediately following the first oblique photo flight (9 June 06), which occurred approximately two weeks after peak flow, water levels were dropping at a rate of 10 cm per day. In such flow conditions, a new habitat unit such as a shallow open nook could become exposed or secondary channels could cease conveying flow over a few days. Ground-truthing was performed by circumnavigating the bar of interest following main and summer channels (as flow conditions permitted) in a 16’ aluminium boat fitted with a 30-horsepower engine. In general, the crew was looking to confirm the presence of units that were difficult to distinguish between, or to observe, by air photo analysis. For example, a low sedimentary cut bank could exist between two bar edge units, or an eddy pool could be missed when classifying because the photos did not capture the sun’s reflection off the water. At mid- to low flow conditions, ground-truthing the interior region of the bars by foot confirmed habitat units that were difficult to observe at high flows. Examples of such units are a dry riffle bed or a low cut bank, both of which are remnants of bar growth by chute and lobe couplets and gravel sheet development, respectively.  3.1.5 Fish-habitat Association Data Fish-habitat associations were derived in order to weight habitat types by ecological relevance for more informed length measurement comparisons. Weighting habitat types 27  adjusts for differences due to shape and shoreline length. However, the ecological data used for adjustments are not meant to be extensive or thorough because the focus of this research is physical habitat characterization. The data used to develop fish-habitat associations were collected for Laura Rempel’s doctoral research during all seasons of 1999 to 2001. Although a variety of sampling gear was used to catch fish, this study reduced the dataset to fish that were caught by beach seine to ensure consistency between catches and data quality. Typically, a small seine (12.5 x 2 m, 6 mm mesh) was deployed from the shore, and less frequently, a larger seine (30 x 3.5 m, 9 mm mesh) was deployed from a boat in habitat units having greater water depths. Sampling was performed by dragging the net downstream within the habitat unit to trap fish within it. Capture efficiency likely varied with a variety of factors, including habitat type, species of fish, fish size, time of year, and time of day (Rempel 2004); however, the number of fish caught in a habitat unit is considered a complete population for the analyses in this thesis, irrespective of which beach seine size was used. Further details on sources of potential bias in the fish data are discussed in Rempel (2004). These data were used to establish relations between various fish characteristics (species, number of fish caught, fish density, total biomass of catch) and the habitat unit types in which the fish were caught. Fish-habitat association analyses were limited to six habitat unit types because of data availability (bar edge, bar head, bar tail, channel nook, eddy pool, open nook), although ten habitat unit types were mapped from the air photos (including artificial bank in the 2006 Long Reach). These six habitat types are herein referred to as the core habitat types. Bay, cut bank, riffle and artificial bank units were eliminated or absent from the dataset. Reasons for missing data include a lack of flow against which to drag the net, limitations of sampling in fast deep water, an insufficient sample size, and no sampling effort in the habitat unit type, respectively.  3.1.6 Observer Variability In order to address the variability introduced by the author’s and assistant’s habitat classification, an observer variability exercise was conducted. The goal of the exercise was 28  to assess the likely precision of habitat classification completed by the author and the assistant by determining the consistency of habitat unit classification at each bar among several classifiers. Variability in identifying habitat unit boundaries was not pursued because doing so would require considerable experience with the classification system and, preferably, observation of habitat units in the field. However, while some participants used the LaRSA habitat classification system for the first time during this exercise, some had participated in its development and testing and therefore were familiar with its application. Each of seven participants was ranked in order of experience with physical habitat classification and/or the stream morphology and hydraulic phenomena that affect the organization of habitat occurrence (i.e. observer 1 is the most experienced). Comparing observer outcomes relative to experience gives an indication of whether the LaRSA system is an expert system, or can be applied by one with little background in this topic. Habitats were classified at low flow at Tranmer Bar and Lower Herrling Bar. These bars were chosen because they were the only two that both the author and the assistant had previously classified, and therefore the results could be compared to their previous work. The author and the assistant never classified the same photo set; however each classified air photos of Tranmer and Lower Herrling that were captured at similar flow conditions. The author was very familiar with the assistant’s work by the time this exercise was performed and therefore did not contribute to the observer variability data set. Classifying the same vertical photos would have produced biased results. Participants were provided with one black and white photocopy of a vertical air photo mosaic of Tranmer Bar and one of Lower Herrling Bar to be classified with coloured pens. Each was also given 1) a PDF file of the same images so that the classifier could zoom in on a computer monitor to see more detail, 2) printed and PDF examples of previously classified bars at different locations in the river, 3) a list and description of all habitat types used in the classification system, and 4) a short tutorial by the author. The participants were permitted to ask questions during the tutorial but not once they had begun classifying because the author did not want to influence their work.  29  3.2 Data Analysis 3.2.1 Data Characteristics As previously discussed in Section 3.1.2, vertical air photos were registered to realworld space, whereas oblique air photos were not. Therefore, reasonably accurate habitat unit length measurements along the bar perimeter could be acquired from vertical air photos. Only habitat counts could be taken from classified oblique. Counts are less desirable than lengths so an attempt was made to project the photos to horizontal space and register them to real-world coordinates using digital photogrammetry software. Unfortunately, the task proved too complex and time-consuming for the purpose of this thesis. Instead, average lengths of each habitat unit type measured from the Long Reach and Historical Reach components of the study were applied to unit counts in the seasonal study. These measurements are herein called assumed lengths. Counts and lengths of habitat unit types are presented for all components of the study in Chapter 4. In the reach and historical components, total habitat type unit lengths are presented as proportions of the total shoreline habitat because the amount of exposed bar varied in each photo set with water level conditions and with evolution of channel morphology over time. Proportions were not calculated for the seasonal component because the cumulative effect of tallying assumed lengths would result in a grossly inaccurate estimate of total bar perimeter. Of the ten habitat unit types classified, the data were truncated to channel nook, open nook and eddy pool types for some analyses. Throughout Chapter 4, these three habitat types are collectively referred to as preferred habitat types because Rempel (2004) found they provided distinct functional opportunities for rearing fish due to highly dissimilar species assemblages. Moreover, higher juvenile fish densities were found in open nooks and channel nooks than in other habitats. Therefore these habitat types are of notable ecological importance, particularly for early life stages (Rempel 2004).  30  3.2.2 Numerical Analyses: Fish-Habitat Association Indices Indices were derived to compare fish-habitat associations with amounts of measured habitat types. The goal was to produce a metric that would differentially weight habitat types to demonstrate ecological significance. Three metrics describing fish-habitat characteristics, (number of fish caught by beach seine, fish biomass, and fish density (# m-2)) were applied to three species groupings: all fish species caught, chinook salmon only (Oncorhynchus tshawytscha), and mountain sucker only (Catostomus platyrhynchus). Chinook were chosen because they are economically important and known to use the gravel-bed reach in large numbers for rearing. Mountain sucker were chosen because they are a blue-listed species and are vulnerable to disturbance. The indices were calculated by averaging each of the nine fish-habitat characteristics (Table 3-5) measured in the six core habitat types discussed in Section 3.1.5 (1). These averages were then ranked and normalized by dividing the characteristic average with the maximum value by itself, effectively setting it to a value of 1 (2). The rest of the averages were therefore expressed as a proportion of the highest so that all resulting values ranged between 0 and 1. Finally these values were multiplied by the proportions of total shoreline length occupied by each core habitat type (3), resulting in 54 indices, herein referred to as weighted fish-habitat characteristics. n  c ij =  Nij =  ∑c k =1  ijk  (1)  nij  cij (max−l ) cijmax  I ij = N ij × p i  (2)  (3)  c = fish-habitat characteristic value = average fish-habitat characteristic value i = fish-habitat characteristic type j = habitat type n = number of observations of each characteristic in each habitat type k = observations of each characteristic in each habitat N = normalized fish-habitat characteristic value max = maximum habitat value l = ranked value of c, where 0 = maximum value, and 5 = minimum value I = weighted fish-habitat characteristic index pi = ratio of total habitat type length to total mapped bar perimeter  The values of the nine habitat indices applied to the six core habitat types are listed in Table 3-5. Eddy pools and open nooks tend to have higher values than other habitat types. Notably, eddy pools have the highest index values in all density characteristics, while open nooks have high values in the count category. 31  Table 3-5. List of normalized habitat indices used to weight the length of mapped habitat with fish-habitat association data. The highest values (1.000) are bolded. n indicates the habitat unit population. Twoletter habitat abbreviations are given in Table 1-4. Habitat n Count All Fish Count Chinook Count Mountain Sucker Biomass All Fish Biomass Chinook Biomass Mountain Sucker Density All Fish Density Chinook Density Mountain Sucker  BE 299 0.282 0.598 0.136 0.736 0.699 0.389 0.526 0.325 0.138  BH 120 0.262 1.000 0.080 0.791 0.936 0.694 0.622 0.442 0.367  BT 101 0.335 0.744 0.253 1.000 0.652 0.305 0.758 0.324 0.120  CN 78 0.591 0.494 0.063 0.499 0.548 0.278 0.864 0.637 0.176  EP 67 0.298 0.465 0.048 0.623 1.000 1.000 1.000 1.000 1.000  ON 80 1.000 0.330 1.000 0.483 0.926 0.698 0.558 0.789 0.627  3.2.3 Statistical Analyses The length data collected in this study represent a survey of all habitat located in each respective reach, at each respective date. It is not appropriate to perform hypothesis tests on a survey because the data are total values, not means. Furthermore, the data are nominal (because they were measured from habitat classes that cannot be ordered), dependent (because the presence of a habitat type was sometimes influenced by its presence at a higher discharge, most notably for the stage-dependent habitat study component) and were not randomly selected from a population. The data do not follow a normal distribution (i.e. nonparametric) and not all variables that contribute to habitat type were measured in this study (such as flow velocity, substrate texture). Because of the above-listed data characteristics, habitat unit count data are tested with nonparametric chi-squared tests for independence. The purpose of the chi-square is to test whether the observed values statistically differ from values that would be expected by chance or probability. The null hypothesis states that there is no statistically significant difference between the expected and observed results. An alpha value of 0.05 was used for all tests. The p-value for a one-tailed chi-square test is the probability that a value chosen at random from a particular chi-square distribution would be greater than or equal to the value of an observed value (the chi-square value) from the same distribution. In some of the following 32  analyses, p-values are used as a measure of similarity between the observed and expected values. p-values near one indicate good similarity and, conversely, low p-values suggest poor similarity. Spearman’s rank correlation coefficient was used to determine strength and direction (positive or negative) of the relation between fish-habitat association index results and historical habitat trends. Spearman’s rank correlation is a nonparametric test and does not assume that the relationship between variables is linear. The null hypothesis is that there is no relation between the two datasets being compared. The coefficient, ρ, will always be between -1 and 1. The closer ρ is to -1 or 1, the stronger the likely correlation. A coefficient of zero accepts the null hypothesis. A summary of correlation strength is givens in Table 3-6.  Table 3-6. Summary of correlation strength for Spearman’s rank correlation coefficient. ρ   Correlation strength  -1 -1 to -0.5 -0.5 to 0 0 0 to 0.5 0.5 to 1  perfect negative correlation strong negative correlation weak negative correlation no correlation weak positive correlation strong positive correlation  1  perfect positive correlation  3.2.4 Graphical Analyses The results of this study are largely presented in graphical formats. All graphs were prepared in either Microsoft Excel 2003 or 2007. All maps were produced in ESRI™ ArcGIS (version 9.0).  33  Chapter 4. Results and Discussion 4.1 Introduction This chapter presents and discusses the results of using the LaRSA habitat classification system to characterize the effects of space and time on Fraser River gravel reach aquatic habitat occurrence. As outlined in Section 2.2, the reaches selected for the three components of this study are referred to as the Long Reach, the Historical Reach and the Seasonal Reach. This chapter is organized accordingly. In general, the results include total habitat lengths, proportions of total shoreline length occupied by each habitat type, counts of habitat units, amounts of habitat weighted by habitat index, and habitat assumed lengths. It is necessary to present both lengths and counts because of the potential for overrepresentation of unit types that have long shorelines relative to the area of water they enclose, such as channel nooks and bays. Some analyses are reduced to the three preferred habitat types (eddy pool, open nook and channel nook) that were demonstrated by Rempel (2004) to be particularly valuable in terms of highly varied species composition and high fish densities.  4.2 Habitat Observer Variability Habitat observer variability was estimated by classification of Tranmer and Lower Herrling Bars. In order to characterize classification precision, each participant interpreted habitat unit boundaries as s/he classified. Classification accuracy could not be measured in this exercise because units were not discretized for the participants to assign a habitat type. All data were produced from March 3, 2006 vertical photography at 695 m3 s-1, with the exception of the data series labelled “Auth”, which was mapped by the author from October 13, 2006 oblique photography with flows of 735 m3 s-1. Some reworking (entrainment, transport and deposition) of nearshore habitat gravels may have occurred over the freshet between the dates of the two photosets. However, since the 2006 peak flow was average-low (8020 m3 s-1 compared with a historical average of 8645 m3 s-1), its potential to cause considerable topographic change on the bars is assumed to be small. Figure 4-1 shows the distribution of habitat observer variability by counts of habitat 34  90  Tranmer Bar  Unit Count  80  BA  70  BE  60  BH  50  BT  40  CB CN  30  EP 20  ON  10  RI  0 Asst Auth  1  2  3  4  5  6  7  80  Lower Herrling Bar 70  BA  60  BE BH  Unit Count  50  BT 40  CB  30  CN  20  EP ON  10  RI  0 Asst Auth  1  2  3  4  5  6  7  Figure 4-1. Distribution of habitat observer variability at Tranmer and Lower Herrling Bars. “Asst” refers to the vertical photo mapper and “Auth” refers to the oblique photo mapper. Numbers 1 to 7 refer to test mappers in rank order of habitat classification experience and are consistent across both bars. Participant 2 did not classify Tranmer habitat, while Participant 3 did not classify Lower Herrling habitat. Legend colours correspond to habitat mapping linework. See Table 1-4 for habitat type abbreviations and Figure 2-2 for bar locations.  35  units per each habitat type mapped at Tranmer and Lower Herrling Bars (data are provided in Appendix D). To statistically investigate comparability between classifications, chi-squared tests were performed. It was assumed that the assistant was the most expert classifier in the group and therefore the number of habitat units she identified was considered the expected value. The observed values were all other participant’s counts, including the author’s. Resulting p-values are considered a measure of similarity with the assistant’s work, whereby high p-values indicate similarity and low p-values indicate difference. p-values were ranked in decreasing order to demonstrate the similarity achieved by participants in terms of their experience with gravel bed river habitat (Table 4-1). These chi-square results should be interpreted with care because the numbers of habitat typed counts used to produce the chisquare scores were small. Therefore a small difference in habitat counts can result in a large statistical difference. Figure 4-1 and Table 4-1 suggest that the highest likelihood for reproducibility amongst all participants lies between the author’s and the assistant’s work at Tranmer Bar. The distributions of habitat types and number of habitat units identified are similar, and the author consistently ranks high in the chi-square comparison of classification performance amongst the other participants. Preferred habitat abundances are also well-matched between assistant and author at Tranmer Bar. These occurrences are favourable and important for consistency through the rest of this thesis. Another notable result is that the assistant and author classified a statistically similar number of bar edge units and number of total habitat units. The number of bar edge units demonstrates the degree of connectivity between other habitat type units, which are generally more specialized than bar edge in terms of their morphologic and hydraulic characteristics. Similarity amongst the total number of habitat units implies small variability in identification of habitat unit boundaries. This is desirable because identifying a habitat unit occurrence from air photos is of primary importance, while its habitat type classification may be arguable. For example, a sheltered habitat unit may be classified as a channel nook by one observer and as a bay by another observer. The correct classification can be field-checked, but observing the unit is critical to a habitat survey. These findings are important because  36  Table 4-1. Ranked p-values obtained from chi-square tests between the number of habitat units classified by the assistant (considered the expert) and all other habitat classifiers, including the author (“Auth”) at Tranmer and Lower Herrling Bars. Participant 2 did not classify Tranmer habitat, while Participant 3 did not classify Lower Herrling habitat. Preferred habitat types are italicized.  Tranmer Bar BA Auth 5 7 4 1 3 6  BE  BH  1.000 0.480 0.294 0.078 0.034 0.014 0.005  Auth 6 3 7 4 5 1  3 7 Auth 6 1 5  0.752 0.752 0.527 0.527 0.058 0.058  1 5 Auth 4 6 7  1.000 1.000 0.317 0.317 0.317 0.317  3 6 Auth 7 4 1  4  0.027  3  0.046  5  CN  0.752 0.048 0.021 0.009 0.005 0.005 < 0.0001 EP  Auth 1 6 7 3 4 5  1.000 1.000 0.480 0.480 0.034 0.005 < 0.0001 ON  BT  CB  5 Auth 3 6 7 1 4  0.046 0.003 0.003 0.003 0.003 < 0.0001 < 0.0001 RI  Auth 1 3 4 6 5 7  0.221 0.100 0.100 0.100 0.100 0.040 0.040 Total  0.655 0.371 0.254 0.168 0.011 0.003  1 4 6 Auth 3 7  1.000 0.317 0.317 < 0.0001 < 0.0001 < 0.0001  Auth 3 6 7 5 1  0.655 0.157 0.017 0.009 0.002 < 0.0001  0.001  5  < 0.0001  4  < 0.0001  Lower Herrling Bar BA Auth 4 5 7 2 6 1  BE 1.000 1.000 0.655 0.655 0.371 0.371 0.18  4 Auth 2 5 1 6 7  CN  BH 0.480 0.343 0.237 0.237 0.058 0.058 0.005  1 Auth 2 5 6 4 7  EP  BT  1.000 0.317 0.317 0.317 0.317 0.046 0.003 ON  5 6 2 Auth 1 4 7  CB 1.000 1.000 0.584 0.254 0.254 0.083 0.021  6 2 Auth 1 4 5 7  RI  0.129 0.058 0.058 0.024 0.024 0.008 0.008 Total  Auth 1 2 6 7 4  0.221 0.129 0.069 0.034 0.016 0.003  2 Auth 1 4 5 6  1.000 0.317 0.317 0.317 0.317 0.317  6 7 2 Auth 5 4  0.752 0.752 0.439 0.129 0.006 0.003  1 2 4 5 6 Auth  1.000 1.000 1.000 1.000 1.000 0.317  7 6 2 Auth 1 4  0.061 < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001  5  0.003  7  < 0.0001  1  0.001  7  < 0.0001  5  < 0.0001  37  they suggest that the assistant and the author can produce habitat maps that are similar in terms habitat accessibility and are therefore ecologically relevant. Lower Herrling Bar habitat classification showed high observer variability amongst all participants, including between the author and the assistant (Figure 4-1; Table 4-1). This was expected because Lower Herrling has a complex shoreline and even at low flow conditions has some shallow channels near its margins. In fact, Lower Herrling is in the author’s opinion the most difficult bar complex to classify in all of the study areas. Therefore the high variability in this case is considered extreme and has no clear implication for the reliability of the rest of the mapping completed by the author and the assistant. The variables that differed most across all observers’ interpretations at both bars were the presence of eddy pools and riffles and the number of bar edge units (Figure 4-1; Table 4-1). The ranges in p-values were large. This result is not surprising given that it is difficult to identify eddy pools and riffles from vertical photography because minimal sunlight reflection from the water surface is evident. (Identification of eddy pools is discussed more thoroughly in Section 4.3.1.) It was expected that bar edge would be variable because it typically lies immediately adjacent to other habitat types, or separates one type from another. Therefore the higher the number of bar edge units identified, the greater the likelihood that the total number of habitat units will be high. Classifications organized by ranked participants were highly variable, implying poor precision amongst the group. Furthermore, there appears to be little relation between experience with habitat classification and the expected classification, or the assistant’s work. The highest ranked participants did not consistently score high p-values. It is assumed that this occurrence is due to a small data set. In some cases, the assistant identified very few unit counts per habitat type (Appendix D). Had the participants classified many more bars, the effect of their experience may have been more evident. Roper and Scarnecchia (1995) recommend using the most simplified habitat classification in order to reduce observer bias and error likelihood, particularly if the classification is to be used for habitat assessment. This recommendation arose from indications that field workers often misclassify habitat units, resulting in biased, unrepeatable, and/or imprecise estimates of habitat availability (Roper and Scarnecchia 38  1995). Based on the simple results of this exercise, it appears that the LaRSA classification system is too complex for precise application without considerable training and/or experience with its use. It is, in effect, an expert system. However, based on the similarity of the assistant’s and author’s work demonstrated in this section, it is assumed that the habitat classifications used to derive the results presented in the rest of this thesis are sufficiently consistent and therefore reliable.  4.3 Spatial Habitat Occurrence: Long Reach Habitat measurements taken from the 2006 Long Reach aerial photography were analyzed from three perspectives: 1) for trends in total amount of Long Reach habitat during the 2006 season, 2) for trends in habitat amongst bar units and, 3) for trends in habitat delimited by 1 km grid cells used in a previous Fraser River Gravel Reach Research Group sedimentation study. Results appear respectively in the three following sections. Figure 4-2 shows the map produced from Long Reach habitat classification of vertical photos.  4.3.1 Long Reach Habitat Abundance Absolute measurements and proportions of total habitat by habitat unit type appear in Figure 4-3. Bar edge was the most abundant habitat type, both by length and by count. It accounted for 53% of the total mapped lengths and 41% of the total number of habitat units in the reach. This occurrence was expected because bar edge is the most general of all habitat types in terms of its form and location characteristics (Table 1-4). A high proportion of bar edge counts indicates a high degree of disconnection between other habitat types and low habitat heterogeneity. Bay habitats were also very abundant. They ranked third in the amount of habitat measured by length (14% of total length) and fourth in the amount of habitat by count (8% of total units). However, in terms of the physical and ecological habitat roles of bays in the Fraser River gravel bed reach, it may not be necessary to discretize bay and channel nook abundance. Rempel (2004) found no statistically significant physical dissimilarity between 39  Figure 4-2. Long Reach habitat map. Photography taken 3 March , 2006 at 695 m3 s-1. Flow is right to left. The reach extends from the downstream end of Wellington Bar at rkm 105 to the upstream end of Herrling Island at rkm 139.  40  0.7  100  a 80  0.6  60  0.4  40  0.3  Proportion  Length (km)  0.5  0.2 20  0.1  0  0.0 BA  BE  BH  BT  CB  CN  EP  ON  RI  AB  240  0.3  160 Count  0.4  120  0.2  80  Proportion  b  200  0.1  40 0  0.0 BA  BE  BH  BT  CB  CN  EP  ON  RI  AB  Figure 4-3. Measurements of habitat abundance from 2006 air photos. a) Total length (km) and proportion of total length of habitat by habitat unit type, b) total count and proportion of total counts of habitat units by habitat unit type. Preferred habitat types are coloured dark grey. Two-letter habitat abbreviations are given in Table 1-4.  channel nooks and bays. Further, the species most frequently associated with channel nooks were also collected in bays, although it was speculated that bays could support a wider range of fish sizes, as a consequence of the deeper water (Rempel, 2004). Combining bay and channel nook measurements amounts to nearly a third of total bar shoreline habitat (31%).and 14% of all units, however the functional value of this large proportion of shoreline length should be interpreted with care. Although measured bay and channel nook abundance suggests that there is a large amount of these habitat types available to aquatic organisms, the 41  units are not frequent, meaning they are less available than other unit types. Moreover, access to their sheltered environments is limited by either narrow entrances or entrances that are small relative to the large amount of shoreline length enclosed within the unit. Of the three preferred habitat types, two were of next highest abundance after bar edge. Channel nook was the second most abundant habitat type by length, while open nook was by count (Figure 4-3). Again, these observations are not surprising. Channel nooks are elongated, dead-end channel features with a large amount of shoreline. By definition, they are biased towards over-representing amount of habitat length relative to other habitat types that are open and exposed to constant flow forces throughout the year. Open nooks, however, are shallow indentations along a bar shoreline that can disappear with a small change in water level (Table 1-4). They are minor shoreline topographic irregularities and, because they have no sedimentary barriers, they require relatively little sediment displacement activity to evolve or change in shape. They are therefore likely to be highly abundant. The third preferred habitat type, eddy pool, made a small contribution to total lengths and counts of habitat. This occurrence was expected given that eddy pool lengths are relatively short in comparison with most habitat types and that their occurrences are defined by relatively specific topographic and hydraulic conditions (Table 1-4). Nonetheless, total eddy pool counts are likely underestimated because they (along with riffles) were particularly difficult to identify from vertical air photos, although they are known to be geographically associated with bar heads and to form in the lee of riffles. In addition, it is assumed that measured eddy pool lengths are substantially inaccurate because their effective boundaries were difficult to determine from only one photo view, and especially from the vertical orientation. Instead, it is more effective to classify eddy pools from multiple air photos that were taken from varying angles relative to the vertical, as in the stage-dependent habitat component of this study. This method increases the chances of identifying differential light reflections off the water surface, which indicates flow separation between an eddy pool backwater and the dominant downstream component of the flow. Figure 4-4 shows an example of two differing photo views of an eddy pool unit at the head of Tranmer Bar. The unit can be clearly observed in one view and is not evident in the alternate view. 42  Ed ddy pool  3 -1 -  Figure 4-4. Examplle of two viewss of Tranmer Bar B (July 7, 20006; 4100 m s ). The west (downstream) ( v view in the top panel clearly shhows an eddy pool p unit at the bar head. No eddy pool is evvident in the v of the botttom panel, whiich was photoggraphed moments after the toop panel. northwest view  43  The sum total of preferred habitat types (channel nook, eddy pool and open nook) amounted to 20% of total shoreline length mapped and 32% of all habitat unit counts mapped. This suggests that although the preferred units make up a relatively small fraction of total habitat length, they are frequent and relatively accessible to aquatic organisms. The difference in the count of open nook units relative to their length is the major contributor to the divergence between unit frequency and accessibility. These abundance values are used as baseline data to compare historical measurements of preferred habitat types in Section 4.5. Proportions of total shoreline length occupied by each core habitat type, weighted by fish-habitat characteristics, appear in Figure 4-5. Three notable results appeared from using this method to calculate relative abundance of habitat classified with the LaRSA system. First, the abundance of bar edge habitat among all fish-habitat weighted characteristics was considerably reduced when calculated by this method in comparison with non-weighted proportions of total length of habitat (Figure 4-3a). This produces an effective reduction in importance of the bar edge habitat type, compared with other habitat types. Second, eddy pools tended to score high values more frequently than any other habitat type by this analysis, despite their minimal contribution to total habitat length, as was demonstrated in Figure 4-3. In contrast to the effect of weighted bar edge, this suggests that the relative importance of eddy pools is increased by measuring habitat abundance with this method. Indeed, Rempel (2004) suggested that eddy pools may be exceptionally important to fish species because of their geographical locations and hydraulic conditions. Although they are hydraulically sheltered relative to the bar heads and riffles near which they are typically found, they are considerably more coupled with the flow than other sheltered units such as channel nooks. This allows fish to exploit a range of physical conditions and food sources with minimal relocation (Rempel 2004). The last notable result from weighting habitat abundance across reach-long measurements is that, in general, the three preferred habitat types demonstrated greater values than the other three core habitat types, with the exception of the biomass of all fish and count of chinook. Weighted channel nook was, however, generally less abundant relative to the other preferred habitat types than non-weighted channel nook, suggesting lesser ecological importance by this analysis (Figure 4-3a). This result is surprising because Rempel (2004) demonstrated that the channel nook habitat type has amongst the highest 44  1.0  Count of all fish  1.0  Count of chinook  Count of mountain sucker 1.0  0.8  0.8  0.8  0.6  0.6  0.6  0.4  0.4  0.4  0.2  0.2  0.2  0.0  0.0  1.0  Density of all fish  0.0 BE BH BT CN EP ON  BE BH BT CN EP ON  1.0  Density of chinook  BE BH BT CN EP ON  1.0  0.8  0.8  0.8  0.6  0.6  0.6  0.4  0.4  0.4  0.2  0.2  0.2  0.0  0.0  0.0  BE BH BT CN EP ON  1.0  Biomass of all fish  BE BH BT CN EP ON  BE BH BT CN EP ON  1.0  Biomass of chinook  Biomass of mountain sucker  1.0  0.8  0.8  0.8  0.6  0.6  0.6  0.4  0.4  0.4  0.2  0.2  0.2  0.0  0.0  0.0  BE BH BT CN EP ON  BE BH BT CN EP ON  Density of mountain sucker  BE BH BT CN EP ON  Figure 4-5. Summary of weighted fish-habitat characteristics for the six core habitat types derived from 2006 air photo habitat mapping. Preferred habitat types are coloured dark grey. Two-letter habitat abbreviations are given in Table 1-4.  juvenile fish densities of all habitat units and provides fish with distinct functional opportunities for rearing based on highly dissimilar species assemblages.  45  4.3.2 Variability Amongst Individual Bar Complexes The distributions of all habitat type lengths and the proportions of total shoreline length that were observed at each bar unit are presented in Figure 4-6. Raw length values sum to the total amount of shoreline perimeter mapped at each bar, which provides estimates of bar sizes. The application of these distributions to bar complex characteristics is discussed below in terms of the three preferred habitat unit types. The sum totals of preferred habitat type lengths and proportions of total shoreline length that were observed at each bar complex unit are presented in Figure 4-7. Lower Herrling Bar was removed from the analysis because it consisted of an artificially large proportion of channel nook habitat (Figure 4-6a). A causeway dammed the flow at the left channel bank and produced one very large channel nook resulting in a misleading overabundance of this habitat type. The three bars that possessed the greatest length of preferred habitat are, in descending order, Tranmer Bar, Greyell Island and Queens Bar (Figure 4-7a). With respect to the greatest proportions of preferred habitat to total shoreline length, Greyell Island, Queens Bar and Carey Bar, again in descending order, rate the highest (Figure 4-7b). A large amount of preferred habitat also is located at Wellington Bar, by both measures. Tranmer Bar, Greyell Island, Queens Bar and Carey Bar are amongst the oldest gravel bar complexes in the reach. Tranmer and Greyell have existed since before the earliest detailed Fraser River map of 1912. Air photo records show that first development of modern Queens Bar and Carey Bar had appeared by 1928 (Church and Ham 2004). Small sediment accumulations existed at present-day Wellington Bar between 1912 and 1928, and by 1938 had developed into a major bar feature (Church and Ham 2004). This suggests that the most valuable habitats are found at the most highly developed bars. Older bars and islands typically have a more complex network of secondary channels and are topographically higher than younger bars. At low flow conditions, channel nooks appear where these secondary channels have been obstructed at their upstream ends by sediments. Likewise, Big Bar and Hopyard Island are two of the youngest gravel bar complexes in the reach and possess the least amounts of preferred habitat, next to Minto Island which was attached to the left bank of the river during aerial photography. (During low flow 46  Preferred habitat length (km)  40  a  30  BA BE BH  20  BT CB  10  CN EP ON  0  1.0  Proportion of preferred habitat  0.8  RI  b  BA BE  0.6  BH BT  0.4  CB CN  0.2  EP ON  0.0  RI  Figure 4-6. Distributions observed at each bar unit of a) all habitat type lengths and b) habitat type proportions of total shoreline length. Legend colours correspond to habitat mapping linework. See Table 1-4 for habitat type abbreviations and Figure 2-2 for bar locations.  47  5  CN  EP  CN  EP  ON  a  Preferred habitat length (km)  4  3  2  1  0  Proportion of preferred habitat  0.4  ON  b  0.3  0.2  0.1  0.0  Figure 4-7 The sum totals of a) preferred habitat type lengths, and b) proportions of habitat lengths to total mapped shoreline observed at each bar unit. Note that Lower Herrling Bar is omitted from the results. Bars are listed left to right with upstream progression. See Figure 2-2 for bar locations.  48  conditions Big Bar and Hopyard Island are joined as one unit (Figure 4-2), and were therefore grouped together for this analysis, which used low flow aerial photography. They are otherwise separated throughout the season.) Two channel nooks, one eddy pool and one open nook were observed at Big Bar and Hopyard Island (Figure 4-2), although their combined perimeter (7.95 km) was approximately the median of all bar shorelines measured (8.41 km).  4.3.3 Variability Amongst 1-km Sedimentation Cells The Fraser River Gravel Reach Research Group estimated a sediment budget for the Fraser River gravel reach in order to make recommendations for gravel extraction (Church et al. 2001). The Group analyzed bathymetric surveys of channel morphology made in 1952, 1984 and 1999 within 1-km cells spaced longitudinally along the channel. The cells were classified according to degree of sedimentation over this time period (Table 4-2). To determine whether certain habitat types were dominantly located in areas of particular sedimentation tendency, habitats within each of the 1-km cells were isolated from the rest of the reach and the resulting data were then pooled by sedimentation class.  Table 4-2. List of sedimentation classes used in the Fraser River gravel reach sediment budget and amount of each class present in the Long Reach study area.  major degradation moderate degradation equilibrium moderate aggradation  Bed Level Change (m) > -1 0.3 - 1 ± 0.3 0.3 - 1  major aggradation  1-2  Class  Cumulative Length (km) 0 7 12 9 4  Proportion of Long Reach (%) 21.9 37.5 28.1 12.5  Chi-square tests were performed to investigate whether the data were indeed affected by their sedimentary environment. The observed value was the number of habitat units in each sedimentation class and the expected value was the number that should be found in the same relative length of the Long Reach. The expected score was obtained by multiplying the 49  proportional pooled length of each sedimentation class relative to the length of the Long Reach study area (Table 4-2) by the number of habitat units observed in the whole reach. Because the 1-km cell boundaries bisected many habitat units during the data isolation procedure, leading to cases where an individual unit was classified with two sedimentation types, the total number of observed units differed from the expected. To correct this difference, the expected value was multiplied by the ratio of the number of observed units to expected units. The null hypothesis states that there is no statistically significant difference between the number of habitat units found in each sedimentation environment type and the number found in the same proportionate length of the Long Reach (α = 0.05). Results of the chi-square tests appear in Table 4-3. All tests for differences between number of habitat units in each sedimentation environment and a proportionate length of the gravel reach accepted the null hypothesis, suggesting no effect of sedimentation tendency on number of habitat units.  Table 4-3. Results of chi-square tests for sedimentary environments effects. Sedimentation Zone Type Moderate Degradation  Equilibrium  Moderate Aggradation  Major Aggradation  Observed Units Expected Units  137 139  241 238  162 178  94 79  χ2  0.025  0.044  1.465  2.745  p-value  0.874  0.834  0.226  0.098  In order to investigate the effect of sedimentation tendency on the habitat types found in each zone, chi-square tests were performed between the number of units per habitat type observed in each sedimentation environment and the expected number. In this case, the expected value refers to the number of units obtained by multiplying the proportional length of each sedimentation class relative to the length of the Long Reach study area (Table 4-2) by the number of units per habitat type found in the Long Reach study area. (Data appear in Appendix D.) The null hypothesis states there is no statistically significant difference between the number of units per habitat type in each sedimentation environment. 50  Almost all tests showed no statistically significant difference between the number of habitat units per habitat type found in each sedimentation zone and the number expected to be found in the whole reach. Only two p-values were less than α = 0.05 (Table 4-4). In order to proceed with further analysis in this section, it is assumed that the sedimentary environment does have an effect on occurrence of particular habitat types. It is possible that the 1-km sedimentation cell size is too fine a spatial scale for this analysis and that bisection of habitat units during the data isolation procedure has influenced the statistical outcomes. Bisection occurred most frequently with habitat types subjected to constant flow forces (e.g. bar edge) because they are typically longer than other habitat types. In some cases, bar edge measured more than a kilometre and consequently spanned three sedimentation cells.  Table 4-4. p-values from chi-square tests of the number of units per habitat type observed in each sedimentation zone class and the number of units in respective proportional length of the Long Reach study area. p-values showing statistically significant difference are italicized. Sedimentation Zone Type Habitat Type  Moderate Degradation  Equilibrium  Moderate Aggradation  Major Aggradation  BA BE BH BT CB CN EP ON  0.411 0.597 0.219 0.745 0.371 0.754 0.332 0.768  0.775 0.454 0.596 0.281 0.401 0.256 0.913 0.462  0.523 0.828 0.920 0.989 0.238 0.407 0.666 0.041  0.308 0.056 0.064 0.847 0.166 0.255 0.135 0.950  RI  0.250  0.089  0.044  0.989  Reach scale graphical analysis shows that preferred habitat is dominantly located in zones where riverbed sedimentation is in equilibrium over time (Figure 4-8). This result is not surprising and follows the findings in the previous section regarding preferred habitat  51  0.10 CN  Proportion of preferred habitat  EP ON  0.08  0.06  0.04  0.02  0.00 Moderate degradation  Equilibrium  Moderate aggradation  Major aggradation  Figure 4-8. The sum totals of proportions of preferred habitat type lengths to total mapped shoreline observed in the four sedimentation classes observed in the Long Reach.  types and bar age or degree of development. It is important that habitat gravels are continually renewed by natural rates of erosion, transport and deposition for optimum rearing and spawning conditions. Old bars exchange sediment, but exhibit little or no net growth. In contrast, net riverbed aggradation increases bar development, while net degradation results in a local loss of habitat. Both of these sedimentation types alter the shoreline shape configuration and affect fish use of specific habitat classes due to habitat availability. Channel nook habitat is most abundant in equilibrium zones (Figure 4-8). This is not surprising because, as discussed previously, channel nooks are typically dead-end remnants of secondary channels that conveyed flow at higher discharges. Secondary channels occur at older and more highly developed bars, where there has been long-term equilibrium sediment exchange. It was not expected that the next highest abundance of channel nook habitat would be found in moderate degradation sedimentation zones.  52  In terms of the hazard effects of localized aggradation, it is notable that there is the least amount of preferred habitat in aggradation zones. Aggradation presents flood concerns with rising water levels due to increasing bed elevation. Such areas could be targeted for flood mitigation by gravel removals taken from either the wetted channel or dry bar surfaces. Removals alter channel morphology and potentially impact river ecosystem health. Impacts to physical habitat include modifications of flow velocity, substrate texture, channel depth, riparian vegetation cover, simplification of bar topography and reduction of shallow-water habitat available at high flows. These factors affect the distribution and abundance of aquatic organisms. Figure 4-9 shows the distribution of total lengths and counts of habitat types across the four sedimentation zones. Both lengths and counts of habitat types generally follow the same trends when analyzed over the Long Reach as a whole (Figure 4-3). However, two notable patterns exist when sedimentation zones are compared: Proportional length of bay units is greater in areas of modest degradation and proportional length of cut bank units is greater in aggradation regions. Modest Degradation  Equilibrium  0.7 0.6 Proportion  0.7  Count  a  Length  Count  0.6  0.5  0.5  0.4  0.4  0.3  0.3  0.2  0.2  0.1  0.1 0.0  0.0 BA  BE  BH  BT  CB  CN  EP  ON  RI  BA  BE  BH  BT  CB  CN  EP  ON  RI  Major Aggradation  Modest Aggradation 0.7  0.7 Count  0.6  Count  0.6  c  Length  d  Length  0.5  0.5  Proportion  Proportion  b  Length  0.4 0.3  0.4 0.3  0.2  0.2  0.1  0.1 0.0  0.0 BA  BE  BH  BT  CB  CN  EP  ON  RI  BA  BE  BH  BT  CB  CN  EP  ON  RI  Figure 4-9. The sum totals of proportions of each habitat type length to total mapped shoreline observed at a) modest degradation, b) equilibrium, c) modest aggradation, and d) major aggradation zones in the Long Reach. Two-letter habitat abbreviations are given in Table 1-4.  53  The greater presence of cut banks in known aggradation zones compared with the other sedimentation classes is particularly surprising because, by definition, cut banks are vertically orientated and eroding (Table 1-4). This result may be due to two causes. First, bar edge units may have been misclassified as cut banks. However, this scenario is unlikely because habitat classifiers found it difficult to identify vertical features from vertically oriented air photos. Therefore, cut banks are more likely to be under-represented and classified as bar edge. The second and most probable cause of high abundance of cut banks in aggradation zones is that sediment accumulation causes the channel to shift and to erode facing banks.  4.4 Temporal Change of Habitat: Historical Reach 4.4.1 Maps of Temporal Habitat Change The maps produced to assess temporal habitat change are presented in Figures 4-10 through 4-18. Detailed information about the methods and materials used to create these maps is in Chapter 3. Table 3-1 lists the summary specifications of the vertical air photos, including dates and discharges during the time of photography.  4.4.2 Temporal Change of Habitat Trends The general long term trend in Fraser River preferred habitat abundance demonstrates an increase in the amount of habitat from 1943 to 1979, a decrease from 1979 to 2001, and renewed increase from 2001 to 2006 (Figure 4-19). However, the results derived from 1963, 1986 and 2001 photography should be interpreted with care because the air photos were captured at flows considerably different from the rest of the dates: 1963 and 1986 photography was captured at higher flows, while 2001 was captured at a very low flow (Table 3-1). In particular, the 2001 photography was captured when the amount of channel nook habitat was considerably minimized (Figure 4-16). The same long term trend exists if these dates are omitted from the analysis, although the inflexion at 2001 would shift to 2003. The effect of variable water levels on habitat is discussed in Section 4.6. 54  Figure 4-10. Habitat map of the Historical Reach, 5 December, 1943 (930 m3 s-1).  Figure 4-11. Habitat map of the Historical Reach, 31 March, 1949 (650 m3 s-1).  55  Figure 4-12. Habitat map of the Historical Reach, 28 April, 1963 (2130 m3 s-1).  new thalweg Gill Island  Hopyard Island  Big Bar  Figure 4-13. Habitat map of the Historical Reach, 22 March, 1979 (955 m3 s-1). Big Bar and Hopyard Island are newly developed. A new thalweg emerged on the right bank (north) side of the river near Gill Bar.  56  Figure 4-14. Habitat map of the Historical Reach, 4 September, 1986 (1890 m3 s-1).  Figure 4-15. Habitat map of the Historical Reach, 20 March, 1999 (700 m3 s-1).  57  Figure 4-16. Habitat map of the Historical Reach, 7 March, 2001 (535 m3 s-1).  Figure 4-17. Habitat map of the Historical Reach, 17 December, 2003 (835 m3 s-1).  58  Figure 4-18. Habitat map of the Historical Reach, 3 March, 2006 (695 m3 s-1).  0.3 Proportion of preferred habitat length  CN  0.3  EP ON  0.2 0.2 0.1 0.1 0.0 1943 1949 1963 1979 1986 1999 2001 2003 2006  Figure 4-19. The sum totals of proportions of preferred habitat type lengths to total mapped shoreline observed each year of Historical Reach vertical photography. Results from 1963, 1986, and 2001 are greyscreened and should be interpreted with care because flows at the time of vertical photography were notably different than most in this analysis (Table 3-1).  59  The amount of preferred habitat length relative to total bar edge was at a maximum about 30 years ago (Figure 4-19; Table 4-5). This occurrence corresponds with substantial change in the river planform between 1949 and 1979 that increased bar complex perimeter and exposed surface area, thereby also increasing available habitat (Figures 4-11 to 4-13). The thalweg shifted position resulting in detachment of present-day Gill Island from the river’s right bank and isolation of modern Hamilton Bar. This occurred as a response to growing local sedimentary congestion (Church and Ham 2004). Several secondary channels also emerged throughout the reach at this time, including the transition of the previous thalweg to a secondary channel Figures 4-11 to 4-13. Moreover, two major bar complexes developed in this reach during the 1970s, Big Bar and Hopyard Island although, as discussed in the last section, they do not contribute a considerable amount of preferred habitat to the reach because of their early stage of development. Habitat type variation by year as a proportion of the total length of shoreline is given in Appendix B.  Table 4-5. List of proportions of preferred habitat lengths to total mapped shoreline displayed in Figure 4-19, and their departures from 2006 values.  Year  Proportion of preferred habitat length  Departure from 2006 values  1943 1949 1963 1979 1986 1999 2001 2003  0.142 0.195 0.204 0.248 0.241 0.190 0.114 0.144  -0.030 0.022 0.031 0.076 0.069 0.017 -0.059 -0.029  2006  0.172  -  Cumulative departure from mean annual discharge in Figure 4-20 shows that Fraser River flows were generally lower than average (decreasing slope) between 1925 and 1953, 1977 and 1996, and 2001 and 2006. Conversely, flows were higher than average (increasing slope) between 1912 and 1924, 1954 and 1976, and 1997 and 2000. There does not appear to 60  5  1200  800  0  600  400  Flow departure PDO Index Channel width  200  ‐5 1900  Channel width (m)  PDO Index & Flow Departure  1000  0 1920  1940  1960  1980  2000  Habitat Departure  0.10  0.05  0.00  ‐0.05  ‐0.10 1900  1920  1940  1960  1980  2000  Figure 4-20. Cumulative departure from mean flow curve for Fraser River between 1912 and 2006. Dashed lines shown inflexions in the flow departure. Variation of average active channel width for the gravel reach (Ham, 2004). Normalized time history of the PDO (http://jisao.washington.edu/pdo/). Departures from 2006 values of the ratio of preferred habitat lengths to total shoreline are in the lower panel.  61  be a direct correlation between flow record regime shifts and the historical abundance of preferred habitat mapped in this study. This means that amounts of habitat did not increase exclusively during periods of higher than average flows, or decrease exclusively during periods of lower than average flows. However, the progressive development of bars, islands and secondary channels during the 1949 to 1979 period, which resulted in the maximum amount of preferred habitat, appears to be related to prolonged higher than average flow conditions (Figure 4-20). Two notably large floods occurred in 1948 and 1972 and caused increased exchange and transport of habitat gravels leading to a greater number of secondary channels (Figure 4-20). As flows dropped, upstream entrances to secondary channels became cut off from the flow as bar surfaces emerged. The result was an increase in the amount of available channel nook habitat, the habitat type that makes up the highest proportion of the three preferred habitat types. The possible correlation of amounts of preferred habitat abundance with flow also extends to historical river width within the gravel reach (Ham and Church 2002), whereby the amount of preferred habitat increases and decreases with river width within the same time frames (Figure 4-20). Ham and Church (2002) describe how extended periods of low flows ultimately simplify and decrease channel width. Bar complexes connect to the floodplain and coalesce into mega islands as secondary channels become filled. In this scenario, less bar shoreline becomes accessible to aquatic species. Conversely, during a period of prolonged higher flows, more bar shoreline becomes accessible as secondary channels are cut and bar complexes become detached from the river banks. In a specific example, preferred habitat abundance declined as flows shifted to an above average period between 1997 and 2000, and then rebounded in 2001, when flows fell below average again. This effect is due to the three peak flows that exceeded 10,000 m3 s-1 in 1997, 1999 and 2002 (Figure 4-20). These flows were considerably larger than recent years. The major flow regime shifts in 1925, 1948, and 1977 accurately correspond to the Pacific Decadal Oscillation (PDO), which is a series of climate reversals centred over the mid-latitude North Pacific (Mantua et al. 1997). The 1948 to 1977 period coincides with the 62  cool phase of the PFO, while the flow records prior to 1948 and following 1977 relate to warm phases (Figure 4-20). Increased habitat abundance prior to 1979 appears to have been facilitated by the PDO cool phase and associated above average flows. PDO cycles appear to last approximately 30 years; therefore, we should expect a cool phase and an above-average flow regime shift in the near future. This should increase channel complexity and associated increase in habitat abundance. It is uncertain if this shift has already occurred. Flows since the late 1990s have been quite variable (Figure 4-20). From a shorter time scale perspective, Figure 4-21 shows habitat departures from 2006 values plotted against annual peak and mean flows for all years of historic air photography, and the five years preceding each date (Photo Years minus 1 through minus 5). Three years prior to the year of air photography, r2 values are largest (Table 4-6). Although the correlation is weak, this may suggest a two- or three-year lag in habitat response to flow. Most photo sets were captured in the spring, prior to freshet (Table 3-1), meaning that the flows that occurred during the year of photography affected next year’s habitat conditions. This result is consistent with Rempel (2004), who found that two freshets of below-average peak discharge resulted in substantial reworking and topographical adjustment of surface sediment across lower Harrison Bar post-gravel mining such that habitat was sufficiently reestablished following gravel scalping. Nonetheless the, r2 values produced here are small, suggesting that habitat departures do not correlate well with either annual scale peak or mean flows. Future work could investigate this topic more thoroughly. Historical proportions of total shoreline length occupied by preferred habitat types, weighted by fish-habitat characteristics relative to 2006 conditions are shown in Figure 4-22. The count of all fish and fish density indices generally followed the same trends as unadjusted proportions of preferred habitat length to total shoreline, meaning they are correlated with the historical flow regime and changes in the river planform morphology. The 1963 data appear to be anomalous, which is likely due to higher stage than what was typical during other years of air photography (Table 3-1). All indices scored high in 1979, which was expected given that preferred habitat abundance was at a maximum during this time. However, count, density and weight of mountain sucker scored high in 2001, which was the year with the least amount of preferred habitat mapped in this study. Although the 63  0.10  Photo Year Photo Year ‐ 1 Photo year ‐ 2 Photo Year ‐ 3 Photo Year ‐ 4 Photo Year ‐ 5  0.08  Departure from 2006 Value  0.06 0.04 0.02 0.00 ‐0.02 5.0  10.0  15.0  20.0  ‐0.04 ‐0.06 ‐0.08  Discharge (103 m3s‐1)  0.10 Departure from 2006 Value  Photo Year Photo Year ‐ 1 Photo year ‐ 2 Photo Year ‐ 3 Photo Year ‐ 4  b  0.08 0.06 0.04 0.02 0.00 ‐0.02 2.0  2.5  3.0  3.5  4.0  ‐0.04 ‐0.06 ‐0.08  Discharge (103 m3s‐1)  Figure 4-21. Correlation plots of a) annual peak flows, and b) annual mean flows with departure of proportions of preferred habitat lengths to total mapped shoreline from 2006 values.  Table 4-6. List of r2 values obtained from correlations of flow and departure of proportions of preferred habitat lengths to total mapped shoreline from 2006 values. r2 Peak Flow  Mean Flow  Photo Year Photo Year - 1 Photo Year - 2 Photo Year - 3 Photo Year - 4  0.355 0.002 0.005 0.336 0.230  0.088 0.009 0.009 0.391 0.178  Photo Year - 5  0.306  0.212  64  Weighted proportion  0.10 0.08 0.06 0.04 0.02 0.00 -0.02 -0.04 -0.06 -0.08  Weighted proportion  0.10 0.08 0.06 0.04 0.02 0.00 -0.02 -0.04 -0.06 -0.08  Weighted proportion  0.10 0.08 0.06 0.04 0.02 0.00 -0.02 -0.04 -0.06 -0.08  Count of all fish  Count of chinook  Density of all fish  Density of chinook  Density of mountain sucker  Biomass of mountain sucker  Biomass of chinook  Biomass of all fish  Count of mountain sucker  Figure 4-22. Amount of habitat index change relative to 2006 conditions. Data are historical proportions of total shoreline length occupied by the three preferred habitat types, weighted by fish-habitat characteristics.  representative proportion of mountain sucker was consistent among all years of sampling during the LaRSA classification system development program, the species was found to be virtually absent from hydraulically sheltered habitats like channel nooks and bays (Rempel 2004). Large sized mountain sucker were found almost exclusively in bar tail habitat units, while smaller-sized fishes were dominantly located in bar edge (Rempel 2004). Other  65  habitatindices failed to show a trend in habitat abundance that relates to the long-term time scales analyzed in this component of the study. In order to investigate how the index-weighted data compare to the unweighted data, Spearman’s rank correlation test was applied to departures of fish-habitat association indices from 2006 values (i.e. Figure 4-22 data) and to departure of proportions of preferred habitat lengths from total 2006 values. The correlation coefficient, ρ, demonstrates correlation strength and direction. The results listed in Table 4-7 show that most correlations were weak. Furthermore, the directional trends are not consistent across the group: there is a positive correlation amongst the density indices, a negative correlation amongst the biomass indices, and no uniform trend amongst the count indices. In general, the density indices correlated best to the unweighted departures, while the density of all fish index showed a notably strong correlation in particular. Counts of all chinook showed a notably weak correlation. Overall, these results suggest that the fish-habitat association indices should not be used as stand-alone measurements of amounts of habitat. Instead, measurements should Table 4-7. Spearman’s rank correlation coefficients and their associated strengths resulting from tests of habitat indices with departure of proportions of preferred habitat lengths to total mapped shoreline from 2006 values. ρ   Correlation strength  Count of all fish Count of chinook Count of mountain sucker Density of all fish Density of chinook Density of mountain sucker Biomass of all fish Biomass of chinook  0.633 0.033 -0.067 0.950 0.433 0.117 -0.200 -0.217  strong positive weak positive weak negative strong positive weak positive weak positive weak negative weak negative  Biomass of mountain sucker  -0.100  weak negative  Index  be obtained by lengths and/or counts, while the indices should serve as ecologically-relevant supplemental information.  66  4.5 Stage-dependant Change of Habitat: Seasonal Reach At high stage, the area of exposed bar surfaces is at a seasonal minimum because bar margins are inundated with water. Micro-topography, which dictates the shapeform characteristics of habitat, varies across a bar. Therefore, as stage rises and/or falls throughout the flood season, the types and amount of habitat found at associated exposed bar margins change accordingly. This section addresses the changes in aquatic nearshore habitat within the Seasonal Reach (Tranmer and Lower Herrling Bars, rkm 132 to rkm 137) during the falling limb of the 2006 hydrograph, measured at the Hope, B.C. gauge. Habitat measurements were obtained by using the LaRSA habitat classification system to map oblique air photos, taken at approximately 500 m3 s-1 flow increments. A detailed description of methods and photo characteristics is in Chapter 3. All results in this section are expressed as a function of discharge, which serves as a proxy for local water level, or stage.  4.5.1 Maps of Stage-dependent Change of Habitat Habitat mapping of the Seasonal Reach for the month of August is presented in Figure 4-23 to Figure 4-25. All other maps are available in Appendix C. Habitat type abbreviations are located in Table 1-4. The maps are organized by decreasing flow or, in other words, by date. In order to maximize the amount of visible habitat it was necessary to map each bar separately because each was best viewed at different orientations (note the north arrows). Results combine the data obtained from the two bars. Flooded bar top and flooded vegetation polygon units are included in the maps for interest, but are not used in any quantitative results. It was not possible to estimate change in the amounts of flooded bar top and flooded vegetation units because polygon areas could not be calculated due to oblique photo orientations.  67  Figure 4-23. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (middle and bottom), 6 August, 2006 (2570 m3 s-1).  68  Figure 4-24. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (bottom), 16 August, 2006 (2110 m3 s-1).  69  Figure 4-25. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (bottom), 31 August, 2006 (1605 m3 s-1).  70  4.5.22 Stage-dep pendent Ch hange of Haabitat Trends Plots depiicting countss and assumeed lengths of preferred habitat h typess at approoximately 50 00 m3 s-1 disccharge increements in thee Seasonal Reach R are preesented in Fiigure 4-26. Raw data values v appeaar in Append dix D. As exxpected, habbitat length and a total couunt increases until a threshold t disscharge is reeached, then decreases ass discharge continues c to increase. At low flows, a maaximum amoount of bar suurface area is i exposed annd most barss are singuular units. As A flow increeases, dry seccondary channnels fill andd dead-end channel c nookks and bays b form. Next, N sedimeentary barrieers located att the head off channel noooks become inunddated and seccondary or summer s channnels begin to t convey floow. Finally,, a network of o summ mer channelss disconnectss the bar surrface and onlly the most elevated e areaas are exposeed. The total t perimetter of these disconnected d d bar units iss less than thhe perimeter at moderate stagees, which results in less available a habbitat.  a  c  b  d  Figure 4-26. Habitaat variation witth discharge at Tranmer and Lower L Herrlingg Bars measureed by a) assum med length of all units, b) assuumed length off preferred habiitat units, c) count of all unitss, and d) countss of preferred haabitat units. Diagonally-hatched regions shhow the range of optimal disccharge for maximum habitat h abundannce.  71  Habitat assumed lengths (Section 3.2.1) sum to the total amount of shoreline perimeter mapped, providing an estimate of bar size (Figure 4-26a). True length could not be measured from oblique photography so assumed length values are reported. A chi-square test shows a statistically significant relation between habitat type counts and discharge (p-value < 0.0001). Habitat counts showed the same trends as assumed length changes, with a minor exception at 3700 m3s-1 in preferred habitat types. The difference at 3700 m3 s-1 is explained by a large proportional contribution of channel nook units, which have high mean habitat lengths compared to open nooks and eddy pools. Optimal discharge for maximum habitat abundance is in the range of 2500 m3 s-1 to 4100 m3 s-1 (Figure 4-26). Mean annual flow for the gravel reach is 2700 m3 s-1, at the lower end of this range. Peak amounts of all habitat types occurred at 3700 m3 s-1 when measured by length and 2550 m3 s-1 when measured by count. Preferred habitat abundance was highest at 2550 m3 s-1 by both length and by count. The 2006 flow duration curve indicates that the discharge range yielding the greatest amount of habitat occurs through about 15-30% of the year (Figure 4-27). It is important to note, however, that although this range constitutes a considerable amount of the year, it does not occur in continuous duration. Instead, the flow duration curve was created by ordering each discrete 2006 flow record by magnitude, not by date. Applying the discharge range to the 2006 hydrograph demonstrates that habitat was most abundant in April 27 to May 17 and July 14 to August 7 (Figure 4-28). High flows do geomorphic work on the substrate by entraining, transporting and depositing sediment. This potentially alters habitat unit shapeform configurations. Between 1952 and 1999, Lower Herrling Bar demonstrated a state of modest degradation at its upstream end to equilibrium sedimentation at its downstream end, while Tranmer Bar showed equilibrium sedimentation at its upstream end and major aggradation at its downstream end (Church et al. 2001). It is assumed that these trends are continuing, therefore amounts and types of habitat found during 2006 flows on the falling limb of the hydrograph would be somewhat different than those found during similar flows on the rising limb. Small and shallow topographic irregularities such as open nooks  72  10.0  Discharge (103 m3s-1)  2006 1977-2000 1954-1976  1.0  0.1 0  20  40  60  80  100  Percentage of time discharge was equalled or exceeded (%) Figure 4-27. 2006 flow duration curve at Hope. 1954-1976 average duration curve for historic high flow period, and 1977-2000 averaged duration curve for historic low flow period. The red segments indicate the percentage of time discharge is equalled or exceeded during the range of optimal flows for maximum habitat availability (15-30%).  10  Discharge (103 m3s-1)  8  2006 1977-2000 1954-1976  6  4  2  0  Figure 4-28. 2006 daily-averaged discharge, 1954-1976 daily-averaged discharge for historic high flow period, and 1977-2000 daily-averaged discharge for historic low flow period at Hope. Red segments indicate seasonal timing of discharges with optimal habitat (April 27 to May 17, and July 14 to August 7 for 2006).  73  would be particularly affected. Rempel (2004) found that below average flood discharges in 2000 and 2001 produced relatively large and balanced sediment exchange on a destabilized bar surface at Harrison Bar following gravel mining, resulting in topographical changes. Nonetheless, it was not possible to make a before and after freshet comparison at Lower Herrling and Tranmer Bars because the data gathering period was limited to late spring through early fall of 2006. In order to collect rising limb data for this type of analysis, the field season should have commenced in mid-March. In general, the amount of bar edge and cut bank habitat types increases with discharge, while the amount of all other habitat types decreases (Figure 4-29). This occurs because of unvegetated bar surface inundation with rising water levels. Generally speaking, the elevation of well-established and/or advanced succession vegetation is an indicator of the water level associated with average annual peak flow. The cohesion of unvegetated coarse gravel and cobble sediments found at Tranmer and Lower Herrling Bars is low. However, at and near the vegetation line, cohesion increases due to root structure and associated trapped fine sediments. Bank angles are high to vertical, and there is little shoreline topographic variability and no upstream sedimentary barriers to flow. Therefore, at high stage, the abundance of cut bank and steep bar edge habitat at the bar margin is large relative to other habitat types. Increasing proportions of bar edge and cut bank units with discharge indicate increasing disconnection between other habitat types. This also suggests homogenization and simplification of shoreline habitat. Refugium availability for aquatic organisms becomes critical when discharge is high and bar edges and cut banks are subjected to constant flow forces. It is assumed that summer and secondary channels, as well as flooded vegetation over main channel bars serve as appropriate refuge. Although this study cannot quantitatively investigate availability of flooded vegetation habitats because area cannot be measured from oblique photography, the area of flooded vegetation clearly increased with discharge (Figure 4-23 to Figure 4-25). The importance of flooded vegetation as habitat in the Fraser River gravel reach is not known because sampling for fish and invertebrates in flooded vegetation areas was not performed during the development of the LaRSA habitat classification system. Future work could investigate this topic.  74  1.0 0.8 Proportion - Length  0.014  BA BE CB CN ON  BH BT EP RI  a 0.012 0.010  0.6  0.008  0.4  0.006  b  0.004 0.2  0.002  0.0  0.000 0  2  4  6  8  0  2  4  6  8  0.10  0.7  d  c  0.6  0.08  Proportion - Count  0.5 0.4  0.06  0.3  0.04  0.2 0.02  0.1 0.0  0.00 0  2 4 6 Discharge (103 m3 s-1)  8  0  2 4 6 Discharge (103 m3 s-1)  8  Figure 4-29. Variation in the proportion of habitat type relative to all habitats over increasing discharge at Tranmer and Lower Herrling Bars measured by a) BA, BE, CB, CN, ON length, b) BH, BT, EP, RI length, c) BA, BE, CB, CN, ON count, and d) BH, BT, EP, RI count. Habitat types are plotted in separate groups to emphasize the ordinate scale. Habitat type abbreviations are defined in Table 1-4. Colours correspond to habitat maps.  As water levels drop, flooded bar top areas are disconnected from the flow and form ponds. With evaporation and continued dropping stage, aquatic organisms using flooded bar top ponds as habitat become trapped and die. This reduction in flooded bar top pond area was most evident throughout August by empirical observation although, again, area could not be measured from oblique photography (Figure 4-23 to Figure 4-25).  75  Changes in the proportion of preferred habitat units relative to all units at approximately 500 m3s-1 discharge increments are presented in Figure 4-30. The abundance of preferred habitat decreases with increasing discharge, however statistical correlation is not strong. This effect occurs because of increasing abundance of bar edge and cut bank habitat units with rising water levels. 0.5  a 0.2  Proportion - Count  Proportion- Length  0.3  r² = 0.39  0.1 0  b  0.4 0.3  r² = 0.60  0.2 0.1 0  0  2  4 6 Discharge (103 m3 s-1)  8  0  2 4 6 Discharge (103 m3 s-1)  8  Figure 4-30. The proportion of preferred habitat units relative to all units over increasing discharge, measured by a) length and b) count.  76  Chapter 5. Conclusions This thesis has investigated the effects of space and time on Fraser River gravel reach habitat occurrence by employing the Large River Stage-Adaptive habitat classification system to vertical and oblique air photographs. It has determined the distribution and abundance of physical habitat: 1.  at the reach scale between Wellington and Tranmer Bars (rkm 105 to 139);  2.  through a historical period of about 60 years between Gill Island and Big Bar (rkm 121 to 128), and;  3.  through the descending limb of the 2006 seasonal hydrograph at Lower Herrling and Tranmer Bars (rkm 132 to 137).  Two research hypotheses were put forward in the introduction: 1. that the amount of habitat would show little variation at the reach scale and throughout historical time, but the distribution would vary, and; 2. that with increasing water levels, the total amount of habitat would decrease and only certain habitat types would be observed. The first statement is not confirmed. Habitat types shown to be preferred by fish (Rempel 2004) were used to analyze this topic because total habitat length summed to estimates of shoreline perimeter. Preferred habitats were most abundant in older and welldeveloped bars and peaked in zones known to be in equilibrium sedimentation. Moreover, preferred habitat abundance fluctuated through the historical record and was at a maximum about 30 years ago due to favourable changes in the river planform at that time. Bar perimeter and exposed surface area increased, and the thalweg shifted position, thereby increasing available habitat. On a shorter time scale, a two- or three-year lag in habitat abundance response to sediment exchange by annual flow conditions may occur, although statistical correlation was weak. This finding is consistent with earlier work by Rempel (2004), who showed that surface sediments were sufficiently reworked to re-establish habitat following gravel scalping. The second statement was confirmed. From low flow conditions, habitat increased with discharge, peaked between 2500 m3 s-1 to 4100 m3 s-1, and then decreased with 77  continued rising water levels. Furthermore, the amount of preferred habitat relative to shoreline perimeter decreased with rising stage. At the highest water levels, bar edge and cut bank habitat types were most abundant and other habitat types were submerged. This finding leads to interest in availability and utilization of flow refugia during high discharges. In an Illinois stream with bankfull discharge of approximately 12 m3 s-1, Schwartz and Herricks (2005) found that fish used floodplain habitat units identified as vegetated point bars and concave-bank benches during bankfull stage. Fish densities were four times higher in these units than in thalweg units (Schwartz and Herricks 2005). Sampling of aquatic species in flooded vegetation on topographically high areas and in side channels on Fraser River would provide a better understanding of flood refugia utilization in large river systems. Three notable concerns related to data characteristics affected the level of accuracy of these results. Firstly, although oblique photography is relatively inexpensive to obtain and provides the benefit of viewing the study areas from several angles and orientations, it does not permit accurate length or area measurements. Projecting the oblique photos to horizontal space and registering them to real-world coordinates would correct this problem. However, before acquiring photographs, it is recommended to establish a thorough understanding of data requirements and photogrammetry software limitations. Secondly, the nature of habitat shapes and shorelines affects comparability of habitat type results. Sheltered habitat types such as channel nooks and bays have proportionately long shorelines compared with open habitat types like bar edge or bar head, so they will be more abundant if defined by length. Weighting habitat types by fish-habitat associations adjusts this effect. However, the adjustments performed in this thesis are rudimentary because the focus of the research was constrained by physical habitat characteristics. A more ecologically significant method of habitat type weighting is recommended. Lastly, it is difficult to accurately identify the true lengths of unit types that are defined primarily by flow configuration, like eddy pools and riffles. Assessing the number of these units is therefore necessary for a complete habitat survey.  78  References Beechie, T.J., Liermann, M., Beamer, E.M., and R. Henderson. 2005. A classification of habitat types in a large river and their uses by juvenile salmonids. Transactions of the American Fisheries Society, 134, 717-729. Bisson, P.A., Nielsen, J.L., Palmason, R.A., and L.E. Grove. 1982. A system of naming habitat types in small streams, with examples of habitat utilisation by salmonids during low stream flow. Acquisition of Utilisation of Aquatic Habitat Information (Ed. N.B. Armandtrout), pp. 62-73. Western Division of the American Fisheries Society, Portland, OR. Church, M. 1992. Channel morphology and typology. The river’s handbook: Hydrological and ecological principles. (Eds. Calow, P. and G.E. Petts) Oxford: Blackwell, 126143. Church, M. and D. Ham. 2004. Atlas of the alluvial gravel-bed reach of Fraser River in the lower mainland showing channel changes in the period 1912-1999. Department of Geography, The University of British Columbia. http://www.geog.ubc.ca/fraserriver/reports/morphatlas_all.pdf Church, M., D.G. Ham, and H. Weatherly. 2001. Gravel management in the lower Fraser River. Department of Geography, The University of British Columbia, Vancouver, British Columbia. http://www.geog.ubc.ca/fraserriver/reports/report2001.pdf Church, M., L. Rempel, and S. Rice. 2000. Morphological and habitat classification of the lower Fraser River gravel-bed reach. Department of Geography, University of British Columbia. http://www.geog.ubc.ca/fraserriver/reports/DFO_report.pdf Desloges, J.R. and M. Church. 1989. Wandering gravel-bed rivers: Canadian landform examples. Canadian Geographer, 33, 360-364. Frissell, C.A., Liss, W.J., Warren, C.E., and M.D. Hurley. 1986. A hierarchical framework for stream habitat classification: viewing streams in a watershed context. Environmental Management, 10, 199-214. 79  Ham, D.G. 2005. Morphodynamics and sediment transport: Fraser River, British Columbia. Ph.D. thesis. Department of Geography, The University of British Columbia. Ham, D.G. and M. Church. 2002. Channel island and active channel stability in the lower Fraser River gravel reach. Department of Geography, The University of British Columbia, Vancouver, British Columbia. http://www.geog.ubc.ca/fraserriver/reports/channelreport2002.pdf. Hawkins, C.P., Kershner, J.L., Bisson, P.A., Bryant, M.D., Decker, L.M., Gregory, S.V., McCullogh, D.A., Overton, C.K., Reeves, R.J., Steedman, R.J., and M.K. Young. 1993. A hierarchical approach to classifying stream habitat features, Fisheries, 18(6), 3-12. Huntington, F.A., Aird, D., Joiner, P., Thorpe, K.E., Braithwaite, V.A., and J.D. Armstrong. 1999. How juvenile Atlantic salmon, Salmo salar L., respond to falling water levels: experiments in an artificial stream. Fisheries Management and Ecology, 6, 357-364. Johnston, N.T. and P.A. Slaney. 1996. Fish habitat assessment procedures: Watershed restoration technical circular No. 8. Ministry of Environment, Lands and Parks and Ministry of Forests. Mantua, N.J., Hare, S.R., Zhang, Y., Wallace, J.M., and Francis, R.C. 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society, 78: 1069-1079. McLean, D.G. and M. Church. 1999. Sediment transport along lower Fraser River. 2. Estimates based on the long-term gravel budget. Water Resources Research, 35: 2549-2559. McLean, D.G., M. Church, and B. Tassone. 1999. Sediment transport along lower Fraser River. 1. Measurements and hydraulic computations. Water Resources Research, 35: 2533-2548. Murphy, M.L., Heifetz, J., Thedinga, J.F., Johnson, S.W., and K.V. Koski. 1989. Habitat utilization by juvenile Pacific salmon (Onchorynchus) in the glacial Taku River, Southeast Alaska. Canadian Journal of fisheries and Aquatic Sciences, 46, 16771685. 80  Northcote, T.G. and M.D. Burwash. 1991. Fish and fish habitats of the Fraser River basin. In: Water in Sustainable Development: Exploring our Common Future in the Fraser River Basin. Edited by A.H.J. Dorcey and J.R. Griggs. Westwater Research Centre, University of British Columbia, Vancouver. pp. 117-141. Northcote, T.G. and P.A. Larkin. 1989. The Fraser River: A major salmonine production system. In Proceedings of the International Large River Symposium. Edited by D.P. Dodge. Canadian Special Publication of Fisheries and Aquatic Science, 106, pp. 172204. Perrin, C.J., J.L. Taylor, and T.B. Stables. 2003. Effects of dredging and transfer pit operations on the aquatic community in the Lower Fraser River near Barnston Island. Report prepared by Limnotek Research and Development for Fraser River Port Authority, Richmond, British Columbia. Rempel, L.L. 2004. Physical and ecological organization in a large gravel-bed river and response to disturbance. Ph.D. thesis. Department of Geography, The University of British Columbia. Rice, S.P., M. Church, Wooldridge, C.L., and E.J. Hickin. 2007. Morphology and evolution of bars in a wandering gravel-bed river; lower Fraser River, British Columbia, Canada. Sedimentology, submitted March 2007. Roper, B.B. and D.L. Scarnecchia. 1995. Observer variability in classifying habitat types in stream surveys. North American Journal of Fisheries Management, 15: 49-53. Rosenau, M.L., and M. Angelo, 2007. Saving the heart of the Fraser: addressing human impacts to the aquatic ecosystem of the Fraser River, Hope to Mission, British Columbia. Pacific Fisheries Resource Conservation Council, Vancouver, BC. http://www.fish.bc.ca/files/SavingHeart-of-the-Fraser_2007_0_Complete.pdf Schwartz, J.S. and E.E. Herricks. 2005. Fish use of stage-specific fluvial habitats as refuge patches during a flood in a low-gradient Illinois stream. Canadian Journal of Fisheries and Aquatic Sciences, 62, 1540-1522. Sullivan K. 1986. Hydraulics and fish habitat in relation to channel morphology. Ph.D. thesis. The Johns Hopkins University. 81  Appendix A: Fish Species Known to Occupy the Fraser River Gravel Reach  82  Table A-1. List of fish species known to occupy the Fraser River gravel reach. Family Acipenseridae Salmonidae  Cyprinidae  Catostomidae  Gasterosteidae Cottidae Petromyzontidae  Osmeridae  Species Acipenser transmontanus Prosopium williamsoni Salvelinus confluentus S. malma Oncorhynchus clarki O. gairdneri O. gorbuscha O. keta O. kisutch O. nerka O. tshawytscha Hybognathus hankinsoni Mylocheilus caurinus Ptychocheilus oregonensis Rhinichthys cataractae R. falcatus Richardsonius balteatus Catostomus macrocheilus C. platyrhynchus C. columbianus Gasterosteus aculeatus G. aculeatus trachurus Cottus aleuticus C. asper Lampetra ayresi L. richardsoni L. tridentata Thaleichthys pacificus  Common Name White sturgeon Mountain whitefish Bull trout Dolly Varden Cutthroat trout Rainbow trout Pink salmon Chum salmon Coho salmon Sockeye salmon Chinook salmon Brassy minnow Peamouth Northern pikeminnow Longnose dace Leopard dace Redside shiner Largescale sucker Mountain sucker Bridgelip sucker Threespine stickleback Marine stickleback Coastrange sculpin Prickly sculpin River lamprey Western brook lamprey Pacific lamprey Eulachon  83  Appendix B: Habitat Type Variations by Year  84  CN  EP  ON  RI  CN  EP  ON  RI  EP  ON  RI  BT  CN  CB CB  BH  BE  BA  RI  EP  ON  CB  CN  BT  BT  2006  0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 BH  BT  BH  BE  BA  RI  EP  ON  CB  CN  BT  BH  2003  BE  CB  BE  BH  BA  RI  EP  ON  CB  CN  BT  BH  BE BE  BA  ON  RI RI  ON  EP  1999  0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0  0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 CN  BT  CB  BH  BE  2001  BA  1986  0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 CN EP  BT  CB  BE BH  BA  1979  1963  0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 BA  Proportion  0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0  0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0  BA  Proportion  0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0  1949  1943  BA BE BH BT CB CN EP ON RI  Proportion  0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0  Figure B-1. Long term variation in habitat types, measured by proportion of each habitat type relative to total bar margins.  85  Appendix C: Maps of Stage-dependent Change of Habitat  86  Figure C-1. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (bottom), 9 June, 2006 (6785 m3 s-1).  87  Figure C-2. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (bottom), 23 June, 2006 (5245 m3 s-1).  88  Figure C-3. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (bottom), 29 June, 2006 (4545 m3 s-1).  89  Figure C-4. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (bottom), 7 July, 2006 (4070 m3 s-1).  90  Figure C-5. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (bottom), 19 July, 2006 (3720 m3 s-1).  91  Figure C-6. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (bottom), 24 July, 2006 (3115 m3 s-1).  92  Figur C-7. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (bottom), 21 September, 2006, (1095 m3 s-1).  93  Figure C-8. Habitat maps of Lower Herrling Bar (top) and Tranmer Bar (bottom), 13 October, 2006 (735 m3 s-1).  94      Appendix D:  Data Tables   95  Table D-1. Counts of the habitat units identified in the habitat observer variability exercise. Numbers 1 to 7 represent participants and are ranked by experience with gravel bed river processes and habitat classification (i.e. 1 has the most experience). Habitat BA BE BH BT CB CN EP ON RI Total  Habitat BA BE BH BT CB CN EP ON RI Total  Asst 8 37 2 1 6 10 1 19 1 85  Asst 5 18 1 3 7 11 1 16 1 63  Auth 8 35 2 4 3 8 2 14 5 81  Auth 5 14 0 1 2 7 0 10 2 41  1 2 9 2 5 2 4 1 6 1 32  Tranmer Bar 2 3 1 23 5 4 2 9 3 17 8 72  4 3 20 6 7 2 3 0 8 0 49  5 6 20 9 3 1 4 1 4 9 57  6 0 25 1 4 2 8 0 23 0 63  7 5 21 3 4 1 9 0 13 5 61  1 2 10 1 1 1 6 2 3 1 27  Lower Herrling Bar 2 3 4 7 5 23 15 0 3 4 6 2 1 5 1 1 0 13 4 1 1 56 36  5 6 13 2 3 0 1 0 5 1 31  6 3 26 0 3 3 4 2 17 1 59  7 6 30 4 7 O 3 5 15 6 76  ON 4.40 79  RI 0.74 27  Table D-2. Total lengths and counts of habitat units mapped in the Long Reach.  Length (km) Count  BA 19.58 43  BE 77.02 218  BH 0.61 23  BT 0.85 18  Habitat Types CB CN 16.47 24.01 33 75  EP 0.35 15  AB 0.17 4  96  Table D-3. Total lengths of habitat units mapped in the Long Reach listed by bar. Total Length of Unit Type (km) Wellington Queens Minto Calamity Harrison Foster Carey Greyell Hamilton Gill Big & Hopyard Powerline Tranmer Herrling Total  BA 0.83 1.35 1.52 2.13 1.99 0.48 0.41 2.36 1.15 0.71 0.29 2.03 2.75 1.59 19.58  BE 16.93 4.01 1.88 0.87 8.96 2.55 2.91 4.96 4.92 5.30 5.94 2.84 7.98 5.90 75.93  BH 0.17 0.03 0.02 0.00 0.06 0.03 0.00 0.04 0.00 0.04 0.01 0.03 0.09 0.06 0.58  BT 0.17 0.06 0.05 0.07 0.00 0.01 0.00 0.08 0.03 0.00 0.15 0.08 0.01 0.14 0.85  CB 1.66 0.82 0.00 0.00 3.22 0.00 0.70 0.70 0.00 2.08 1.06 0.38 2.10 3.76 16.47  CN 1.60 2.48 0.00 0.34 1.17 0.32 1.16 3.15 1.12 1.30 0.38 0.35 3.14 21.48 37.98  EP 0.04 0.01 0.00 0.00 0.01 0.00 0.01 0.01 0.00 0.03 0.06 0.03 0.02 0.02 0.25  ON 0.44 0.11 0.07 0.00 0.64 0.06 0.31 0.23 0.39 0.18 0.02 0.21 0.81 0.95 4.40  RI 0.03 0.00 0.00 0.00 0.02 0.00 0.08 0.05 0.03 0.45 0.03 0.00 0.01 0.01 0.71  Table D-4. Total lengths of habitat units mapped in the Long Reach listed by sedimentation zone type.  Moderate Degradation Equilibrium Moderate Aggradation Major Aggradation Total  BA 7.07 4.62 5.96 2.83 20.48  BE 10.49 28.99 25.51 12.03 77.02  BH 0.11 0.31 0.19 0.00 0.61  Habitat Lengths (km) BT CB CN EP 0.14 1.66 7.62 0.05 0.28 5.76 11.39 0.11 0.24 5.51 2.99 0.20 0.18 3.54 1.85 0.00 0.85 16.47 23.84 0.35  ON 0.91 1.87 0.79 0.84 4.40  RI 0.08 0.51 0.02 0.13 0.74  Total 28.13 53.83 41.40 21.40 144.77  Table D-5. Counts of habitat units mapped in the Long Reach listed by sedimentation zone type.  Moderate Degradation Equilibrium Moderate Aggradation Major Aggradation Total  BA 14 18 12 9 53  BE 61 105 75 43 284  BH 3 12 8 0 23  BT 4 5 6 3 18  Habitat Counts CB CN 6 21 18 27 15 21 8 15 47 84  EP 2 7 6 0 15  ON 22 31 16 12 81  RI 4 18 3 4 29  Total 137 241 162 94 634  97  Table D-6. Expected counts of habitat units per sedimentation zone types used for chi-square tests. Counts were adjusted by proportion of Long Reach occupied by each sedimentation zone class.  Assumed Long Reach Value Moderate. Degradation. Equilibrium Moderate Aggradation Major Aggradation  BA  BE  Counts of Habitat Type BH BT CB CN EP  43 11 19 14 6  218 57 98 73 33  23 6 10 8 3  18 5 8 6 3  33 9 15 11 5  75 20 34 25 11  15 4 7 5 2  ON  RI  Total  79 21 35 27 12  27 7 12 9 4  531 139 238 178 79  Table D-7. Proportions of habitat length to total mapped shoreline listed by historical air photo set.  1943 1949 1963 1979 1986 1999 2001 2003 2006  BA 0.14 0.19 0.16 0.07 0.09 0.20 0.11 0.09 0.12  BE 0.61 0.52 0.44 0.56 0.48 0.48 0.61 0.57 0.57  BH 0.01 0.00 0.01 0.01 0.01 0.01 0.01 0.01 0.00  Proportion of Total of Mapped Length BT CB CN EP 0.01 0.05 0.13 0.000 0.01 0.04 0.15 0.000 0.01 0.15 0.17 0.005 0.01 0.09 0.20 0.003 0.03 0.15 0.19 0.002 0.02 0.08 0.14 0.002 0.01 0.09 0.05 0.002 0.01 0.13 0.09 0.002 0.01 0.11 0.15 0.005  ON 0.01 0.05 0.03 0.04 0.05 0.05 0.06 0.05 0.02  RI 0.000 0.002 0.031 0.010 0.008 0.003 0.026 0.003 0.015  Total 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0  Table D-8. Counts of habitat units measured from Tranmer and Lower Herrling Island Bars listed by date and discharge. Date  Discharge (m3 s-1)  9-Jun 23-Jun 29-Jun 7-Jul 19-Jul 24-Jul 6-Aug 16-Aug 31-Aug 21-Sep 13-Oct  6775 5250 4550 4075 3700 3125 2550 2100 1625 1100 735  Number of Habitat Units BA 2 9 9 13 12 8 10 11 14 12 13  BE 78 104 134 129 132 119 136 86 68 46 49  BH 0 3 3 4 5 1 4 3 2 2 2  BT 0 0 0 0 2 6 6 5 3 6 5  CB 42 42 40 32 55 39 42 23 10 7 5  CN 8 25 31 35 49 33 41 22 11 21 15  EP 0 2 7 6 7 8 4 4 4 3 2  ON 7 32 51 58 28 36 56 45 25 23 24  RI 0 13 20 28 16 16 11 5 13 8 7  Total 137 230 295 305 306 266 310 204 150 128 122  98  Table D-9. Mean habitat type lengths calculated from the historical study used to derive habitat unit lengths for stage-dependent study. Mean Habitat Type Length (km) from Historical Study BA 0.804  BE 0.34  BH 0.04  BT 0.073  CB 0.328  CN 0.249  EP 0.023  ON 0.074  RI 0.023  Table D-10. Approximated habitat unit lengths derived from counts and historical study mean lengths listed by date and discharge. Date  Discharge (m3 s-1)  9-Jun 23-Jun 29-Jun 7-Jul 19-Jul 24-Jul 6-Aug 16-Aug 31-Aug 21-Sep 13-Oct  6775 5250 4550 4075 3700 3125 2550 2100 1625 1100 735  Approximated Habitat Unit Lengths (km) BA 1.61 7.23 7.23 10.45 9.65 6.43 8.04 8.84 11.25 9.65 10.45  BE 26.49 35.31 45.5 43.8 44.82 40.41 46.18 29.2 23.09 15.62 16.64  BH 0 0.12 0.12 0.16 0.2 0.04 0.16 0.12 0.08 0.08 0.08  BT 0 0 0 0 0.15 0.44 0.44 0.36 0.22 0.44 0.36  CB 13.76 13.76 13.1 10.48 18.02 12.78 13.76 7.53 3.28 2.29 1.64  CN 1.99 6.21 7.71 8.7 12.18 8.2 10.19 5.47 2.73 5.22 3.73  EP 0 0.05 0.16 0.14 0.16 0.18 0.09 0.09 0.09 0.07 0.05  ON 0.51 2.35 3.75 4.27 2.06 2.65 4.12 3.31 1.84 1.69 1.77  RI 0 0.3 0.46 0.64 0.37 0.37 0.25 0.11 0.3 0.18 0.16  Total 44.36 65.34 78.03 78.64 87.59 71.49 83.23 55.05 42.88 35.24 34.87  99  

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.24.1-0066171/manifest

Comment

Related Items