Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Channel geomorphic units as benthic macroinvertebrate habitat in small, high gradient streams on Vancouver… Halwas, Karen L. 1998

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

Item Metadata


831-ubc_1998-0157.pdf [ 9.46MB ]
JSON: 831-1.0099283.json
JSON-LD: 831-1.0099283-ld.json
RDF/XML (Pretty): 831-1.0099283-rdf.xml
RDF/JSON: 831-1.0099283-rdf.json
Turtle: 831-1.0099283-turtle.txt
N-Triples: 831-1.0099283-rdf-ntriples.txt
Original Record: 831-1.0099283-source.json
Full Text

Full Text

CHANNEL GEOMORPHIC UNITS AS BENTHIC MACROINVERTEBRATE HABITAT IN SMALL, HIGH GRADIENT STREAMS ON VANCOUVER ISLAND, BRITISH COLUMBIA By Karen L. Halwas B.Sc., University of Victoria, 1994 A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE DEGREE OF M A S T E R OF SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES Department of Geography We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A April, 1998 © Karen L . Halwas, 1998 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of 4Q PEe?fa (LLP t4v. The University of British Columbia Vancouver, Canada DE-6 (2/88) Abstract Headwater streams typically have no fish, owing to steep gradients and impassible barriers; therefore, scientific research and protection measures have been focused on fish bearing streams. The Scientific Panel for Sustainable Forest Practices in Clayoquot Sound (CSP) developed a channel classification system which is pertinent to all streams, Ashless and fish bearing alike, and upon which management prescriptions in Clayoquot Sound, Vancouver Island, British Columbia are based (CSP, 1995). The CSP classification delineates channels according to four physical criteria: bed material, gradient, entrenchment, and width. The current study was undertaken to determine the efficacy with which the CSP classification system delineates small, steep streams, on the basis of channel geomorphic units within them, and to examine the benthic macroinvertebrate habitat capability of these geomorphic units. Falls, bedrock cascades, boulder cascades, rapids, chutes, riffles, glides, and pools were described according to their bed slope and dominant channel-material type and organization. In addition, the area of each geomorphic unit was measured. Seventeen streams were grouped into four CSP channel classes which were compared with respect to the mean relative proportion of class area in geomorphic units. Stratified random benthic samples were extracted from geomorphic units in order to investigate and to compare their habitat capability. "Alluvial channels" in the study exhibited only weak, very infrequent fluvial transport; therefore, they were termed semi-alluvial. In general, high gradient geomorphic units (i.e. bedrock and boulder cascades) were dominant in steep, largely non-alluvial channels. Lower gradient units (i.e. riffles and rapids) were common in semi-alluvial streams with more mild slopes. Accordingly, channel classes with opposing bed material and gradient designations exhibited notable differences with respect to relative proportions of geomorphic units while width and entrenchment designations exerted little influence on channel organization. Ultimately, only two of the four CSP classification criteria effectively systematized channels on the basis of channel geomorphic units within them. Abundance of benthic macroinvertebrates was greatest in riffles («100 individuals per two minute kick sample), followed by rapids («80 individuals/sample), pools («70 individuals/sample), boulder cascades («60 individuals/sample), chutes (»50 individuals/sample), and lowest in bedrock cascades («25 individuals/sample). In addition, abundance of invertebrates in channels with ephemeral flow regimes was considerably lower compared to channels with seasonal or perennial flow regimes. Ordination of macroinvertebrate taxa showed that community structure of bedrock cascades and chutes were similar but different from other habitats. Similarly, the benthic macroinvertebrate community structure of channels with ephemeral flow regimes was very distinct. iv Table of Contents Abstract i i List of Tables viii List of Figures ix Acknowledgments xi Chapter 1: Study Rationale and Objectives 1.1 Study Rationale and Background Information 1 1.2 Research Objectives 5 1.3 Thesis Structure 7 Chapter 2: Physical Character of the Clayoquot Sound Region and the Study Reaches 2.1 Introduction 9 2.2 Clayoquot Sound Area Description 9 2.2.1 Climate 9 2.2.2 Terrain 12 2.2.3 Vegetation 14 2.2.4 Hydroriparian Ecosystem and Hydrology 15 2.3 Classification of the Hydroriparian System: Stream Environment 16 2.4 Study Reach Descriptions 22 Chapter 3: Channel Geomorphic Unit Representation in Stream Hydroriparian Classes 3.1 Introduction 28 3.2 Methods 31 3.2.1 Data Collection 31 3.2.2 Data Analysis 35 3.3 Results 37 3.3.1 Geomorphic Unit Descriptions 37 Fall 37 Bedrock Cascade 41 Boulder Cascade 41 Rapid 43 Chute 44 Riffle 44 Glide 45 Primary Pool 45 3.3.2 Channel Class Distinctions 46 Hydroriparian Stream Classification Criteria 46 Channel Classes 52 3.4 Discussion 59 3.4.1 Geomorphic Units 59 3.4.2 Channel Classes 64 Chapter 4: Variation in Benthic Macroinvertebrate Abundance and Community Structure Among Habitats and Streams 4.1 Introduction 70 4.1.1 Invertebrate Biology 75 Heptageniidae 77 Leptophlebiidae 78 Baetidae 79 Simuliidae 80 Chironomidae 80 Nemouridae 82 Chloroperlidae 82 Lepidostomatidae 83 Enchytraeidae 84 Tricladida 84 vi 4.2 Methods 85 4.2.1 Data Collection 85 4.2.2 Data Analysis 87 4.3 Results 89 4.3.1 Quality Control 89 4.3.2 Macroinvertebrate Abundance 89 4.3.3 Macroinvertebrate Community Structure 112 Dominant Taxa 112 The Physical System 112 The Biological System 120 Linking the Systems 124 4.4 Discussion 127 4.4.1 Inter-habitat Variation 127 4.4.2 Inter-stream Variation 131 4.4.3 Temporal Variation and Interactions 135 Chapter 5: Conclusions, Limitations, and Final Comments 5.1 Conclusions 138 5.1.1 Fluvial Geomorphology 138 5.1.2 Aquatic Ecology 139 5.2 Study Limitations 140 5.2.1 Microhabitat 140 5.2.2 Biotic Forces 142 5.2.3 Invertebrate Measures 143 5.3 Final Comments 145 5.3.1 Integration of Ecological and Geomorphological Results 145 5.3.2 Ecological Significance of the CSP Hydroriparian Classification 146 References • 149 Appendices Appendix viii List of Tables 2- 1 Summary of study reach characteristics and channel class 26 3- 1 Study reach gradients as determined by four different methods 35 3-2 Channel geomorphic unit bed slope summary 41 3-3 Comparisons of mean area in geomorphic units among paired groups of reaches divided by a width, entrenchment, bed material, and gradient criterion 48 3-4 Comparisons of mean area in geomorphic units among reaches with gradients greater than and less than 15% gradient and among reaches greater than and less than 25% gradient. 51 3-5 Relative proportions of class area in geomorphic units 51 3-6 T-test and Mann-Whitney U-test results of pairwise comparisons among channel classes with respect to mean area in geomorphic units 54 3-7 Reclassification of study reaches to recognise a semi-alluvial category 56 3- 8 Gradient ranges of geomorphic units from various previous studies 61 4- 1 Insect data quality control summary 90 4-2 Total number of individuals in the ten most abundant taxa and their proportion of the grand total 90 4-3 Results of three-factor A N O V A models for effects of stream, habitat, date, and interactions on total invertebrate abundance and on individual taxon abundance for the ten most abundant taxa 95 4-4 Multiple pairwise comparisons of means amongst streams, habitats, and dates for total invertebrate abundance and individual taxon abundance for the ten most abundant taxa 96 4-5 Results of three-factor A N O V A models for effects of stream, habitat, date, and two-way interactions on total invertebrate abundance and on individual taxon abundance for the ten most common taxa I l l 4-6 Proportional abundance of most common taxa in each habitat type and each stream.... 113 4-7 Correlation matrix of taxa, environmental variables, and principal components 119 IX List of Figures 2-1 Mean monthly precipitation and temperature 11 2-2 Classification of the stream hydroriparian system 18 2-3 Determining (non-) entrenchment status by the position of terrestrial vegetation in relation to the stream bed 23 2- 4 Map of the study area and positions of study reaches 27 3- 1 A stylized example of a scaled diagram produced during reach surveys 33 3-2 Definition diagrams for boulder cascades, rapids, riffles, falls, pools, and glides 38 3-3 Photographs of typical falls, bedrock cascades, boulder cascades, rapids, chutes, riffles, and pools 39 3-4 Box plot of channel geomorphic unit vs. bed slope 42 3-5 Box plot of channel geomorphic unit vs. water velocity ; 47 3-6 Box plot of channel class vs. relative proportion of class area in falls, bedrock cascades, boulder cascades, rapids, chutes, riffles, and pools 53 3- 7 Scatterplots of channel geomorphic units vs. relative proportion of reach area 57 4- 1 Mean invertebrate abundance per sample for habitat types and streams 92 4-2 Inter-habitat variation in mean total invertebrate abundance for each stream 93 4-3 Plot of means illustrating details of stream x habitat interactions among the Heptageniidae 98 4-4 Plot of means illustrating details of stream x habitat x date interactions among the Nemouridae 100 4-5 Plot of means illustrating details of stream x date interactions among the Chironomidae. 98 4-6 Plot of means illustrating details of stream x habitat interactions among the Leptophlebidae 103 4-7 Plot of means illustrating details of stream x habitat interactions among the Enchytraeidae 103 4-8 Plot of means illustrating details of stream x date and habitat x date interactions among the Lepidostomatidae 107 4-9 Results of a principal components analysis processed on the array of 95 sites by 4 environmental variables 117 4-10 Results of a principal components analysis processed on the array of 182 sites by 18 taxa 121 4-11 Results of a principal components analysis processed on the array of 52 sites by 18 taxa (composite data) 123 4- 12 Scatter plot of biological PC 1 factor scores vs. physical PC 1 factor scores 125 5- 1 Mean invertebrate abundance (+ 1 SE) expressed as animals per sample for each study stream pooled across habitat types and sampling dates 147 xi Acknowledgments I wish to express sincere gratitude to those who have contributed significantly to this thesis project. First and foremost my supervising professor, Dr. Michael Church, who provided affluent knowledge, unconfining guidance, and subtle encouragement at both critical and less important moments throughout the duration of the project. Second, a thesis supervisory committee member, Dr. John Richardson, who graciously provided expert advice, moral support, laboratory space and equipment, and supplies. Third, my field and lab assistants, Ralph Halwas, Ryan Letchford, Jessica Kaman, and Christy Steckler, whose competence, perseverance through undesirable conditions, and companionship merit high accolades. Fourth, my fellow students and friends whose support and understanding provided me with momentum to reach the finish. Fifth, Alan Chapman (Chapman Geoscience) who kindly approved several requests for advice and assistance. Sixth, Arlene Suski (Long Beach Model Forest), and Ken Matthews (B.C. Forest Service) whose help with site selection guided me to some beautiful study channels. Seventh, Dr. Art Borkent (Royal British Columbia Museum and American Museum of Natural History), Craig Logan (Environment Canada), Karen Needham (UBC), and Don Stacey (Royal Ontario Museum) who provided taxonomic expertise. Finally, my husband, Tim Boulton who despite the physical distance between us remained my closest friend and strongest crutch. This project was funded by a Forest Renewal British Columbia research grant awarded to M . Church and by The University of British Columbia. Chapter 1: Study Rationale and Objectives 1.1 Study Rationale and Background Information Small, steep streams that typically have little direct fishery value have not attracted the same attention, in terms of scientific research and protection measures, as fish bearing streams (Beschta and Platts, 1986). However, while these upstream reaches may have no fish, owing to steep gradients and impassable barriers, they contribute to downstream fishery success through their role in organic carbon recruitment and by influencing runoff timing and water quality. The view that river systems are connected from headwaters to mouth is paramount to the river continuum concept which proposes that a continuous gradient of physical conditions, created by the drainage network, elicits a series of biotic adjustments within the constituent populations. The balance of producer and consumer communities, therefore, parallels longitudinal transition in channel size, degree of shading, and organic matter loading, transport, and storage along the length of a river (Vannote et al, 1980). The basis for management of fish bearing streams and their riparian zones in the Province of British Columbia has been classifications based on easily measurable channel features and characteristics of the in-stream fish populations (cf. Forest Practices Code of British Columbia, 1995). These criteria do not represent the hydrological, sedimentological or ecological connectivity of the entire drainage basin. In recognition that the aforementioned approach to aquatic and riparian ecosystem management is neither satisfactory nor logical (Church, 1996), The Scientific Panel for Sustainable Forest Practices in Clayoquot Sound 2 (CSP) developed a physically based channel classification system (CSP, 1995). In Clayoquot Sound, management prescriptions for all streams and their adjacent terrestrial surfaces, to the limit of riparian influence, are based on this classification. Since first order channels compose approximately half of the total length of channel network (Horton, 1945; Montgomery and Buffington, 1997), special attention to headwater streams is long overdue. In addition, many issues challenging the land manager originate at the tips of drainage networks; therefore, it is here that a classification would be most suitable (Whiting and Bradley, 1993). The physically based CSP system, however, has been criticized for not accounting for attendant biological properties and ecological dynamics of hydroriparian ecosystems (Chan-McLeod, 1996). The classification and its criticism form the basis of this study. A strict definition of classification in general is the arrangement of objects or phenomena into groups on the basis of their similarities or relationships (Platts, 1980). Perhaps the most successful example of this is the binomial system of classification [Carl Linnaeus (1708-1778)] which is the methodical nomenclature of taxa (Weinstock, 1985; Larson, 1994). Stream classifications are tools most commonly applied for the purposes of resource management. They allow communication through a common language (MacKenzie and Banner, 1995; Rosgen, 1996) and they aid resource management decisions by allowing planners to inventory, compare, evaluate, and prioritize resource values (Lotspeich and Platts, 1982; Garcia, 1985). Classification of the natural environment is difficult, at best, because it is complicated by both longitudinal and lateral linkages, by morphological changes that occur over time, by tremendous spatial heterogeneity, and by indiscrete boundaries between apparent patches (Minshall, 1988; Naiman et al, 1992). Typologies of rivers and streams in the past have concentrated primarily on downstream parts of networks where rivers are large, slopes are mild, and alluvial processes predominate (Whiting and Bradley, 1993). Traditional ideas regarding classification assume that all channels may be separated into two major groups: alluvial ones that flow through their own deposits and non-alluvial ones that are unable to transport their bed material (Schumm, 1977). Dominant channel-material particle sizes may range from boulders in headwater streams to sand grains in lowland rivers and the frequency with which bed particles are moved depends upon the competence of the channel (Baker and Ritter, 1975), which is the largest grain size that can be transported under the prevailing hydraulic conditions. Shallow and ephemeral flow in many headwater channels generally is not capable of transporting colluvial sediment introduced to the channel, resulting in substantial storage of this material (Benda, 1990; Whiting and Bradley, 1993; Abrahams et al, 1995; Montgomery and Buffington, 1997). Because grain size changes gradually along the stream continuum, a simple bifurcation system does not adequately represent zones of transition between non-alluvial and alluvial segments of the network. The term semi-alluvial, therefore, is used to describe those transitional channels exhibiting weak fluvial transport. Channels are composed of successive, meso-scale bedforms which vary longitudinally with bed topography and water surface slope, depth and velocity (Frissell et al, 1986; Hawkins et al, 1993) and are termed geomorphic units. Channel geomorphic unit descriptions in lowland rivers often are limited to pools and riffles (Leopold et al, 1964; Richards, 1976b), although Bisson et al (1981) assert that, depending upon location, flow patterns, and flow controlling structures, various pools and riffles offer different habitat potential. Few studies have documented the bedforms in small, high gradient channels, and certainly none have addressed bedforms in extremely steep channels similar to those discussed hereinafter, which has resulted in imprecise nomenclature and incomplete descriptions of bed features (except see Grant et al, 1990). Small, high gradient streams exhibit highly variable bedforms particularly conspicuous at the spatial scale of geomorphic units (Grant et al, 1990; Hawkins et al, 1993). Since small-scale bedforms develop within constraints imposed by the large-scale systems of which they are a part (Frissell et al, 1986), discrimination of geomorphic units may shed light on the processes that account for their pattern and origin (Grant et al, 1990). Ultimately it is the processes that influence both the physical environment and the biological communities. The structure and dynamics of the physical environment are the primary factors influencing the production and diversity of stream biota (Hawkins et al, 1993); therefore, characterisation of geomorphic units serves as a foundation to investigate the aquatic community. Current velocity and substratum, which vary among channel geomorphic units, determine to a large extent the microenvironmental conditions influencing stream dwelling benthic organisms (Minshall and Minshall, 1977). Dwellers of lotic systems rely on water to satisfy consumptive and respiratory requirements (Hynes, 1970). The substratum provides benthic organisms a medium upon which they may move, attach, feed, or seek refuge (Minshall, 1984). Current velocity varies in space with water surface slope, depth and substrate roughness and in time with discharge. Although substrate characteristics change through time, it is the spatial variability, which results from terrestrial and upstream inputs and sorting action of the current, that presents a major influence upon invertebrate distribution. Clearly, spatial and temporal variation in current and substratum create a broad diversity of environmental conditions for benthic dwellers. Such organisms exhibit various morphological and behavioural strategies to contend with different ambient factors. It is the form and function of the organisms, viewed in light of their immediate surroundings, which enables interpretation of the community structure. Other important factors affecting the occurrence and community structure of invertebrates are temperature and liability to drought. Temperature and day length affect the life histories of invertebrates, particularly those in small streams of temperate climates; therefore, season accounts for much of the temporal variation in benthic communities (Hynes, 1970). Aquatic habitats subject to frequent drying exhibit amplitudes in both physical and chemical parameters which are much greater than those found in permanently flowing areas (Williams and Hynes, 1977). These conditions represent a major disturbance to organisms in lotic environments since they rely on flowing water for survival. As a result, colonization of temporarily wet areas is precluded by species of normally perennial streams that require flowing water (Brussock and Brown, 1991; Ward, 1992). The fauna, therefore, is composed of some species distinct to temporary waters and some aquatic invertebrate species that have adapted to seasonal drying (Williams, 1987; Brussock and Brown, 1991; Ward, 1992). 1.2 Research Objectives The current study was undertaken to evaluate the efficacy of the CSP classification system, the purpose of which is to identify structurally distinct channels in Clayoquot Sound, Vancouver Island, British Columbia. The major criticism of the classification system is that, because it is physically based, the entire biotic component of the hydroriparian ecosystem is not considered (Chan-McLeod, 1996). Therefore, in order to determine channel class distinctiveness with respect to biological function, benthic macroinvertebrate abundance and community structure were examined in relation to geomorphic units. Presumably, one may extrapolate information gathered at the scale of geomorphic units to higher levels (i.e. reaches or segments), within which the lower levels are spatially nested. Because this study has two components, fluvial geomorphology and aquatic ecology, the research hypotheses were twofold: (1) 1 hypothesized that inter-class variation in relative proportion of area in geomorphic units would be strongly linked to channel bed material, slope, entrenchment, and width, the criteria upon which the CSP classification system is based. (2) I hypothesized that the spatial heterogeneity in abundance and community composition of benthic macroinvertebrates would be strongly linked to variation among habitats, the characteristics of which determine, in part, ambient conditions. To test the hypotheses, study channels were classified according to the CSP system. Geomorphic units within the channels were identified, described, and their areas measured. The relative effectiveness of the classification criteria (bed material, gradient, entrenchment, and width) in distinguishing channels was determined by comparing groups of channels with respect to the mean relative proportion of stream area in geomorphic units. Finally, in order to determine i f channel classes are distinct, they were compared with respect to the mean 7 relative proportion of class area in geomorphic units. The benthic communities of geomorphic units (habitat types) were investigated, compared, and assessed in light of the environmental conditions determined by each habitat type. In addition, through casual observation in the early stages of this research, marked differences in the fauna among channels with ephemeral flow and those with seasonal or perennial regimes were apparent. Although not a primary inspiration initiating the research, the differences in benthic communities of ephemeral streams compared to streams with more permanent flow became a focus as well. The research summarized hereinafter has been confined, due to limited time and resources, to systems relatively unimpacted by human activities. Knowledge of pristine, high gradient, headwater environments is a prerequisite to the examination of morphology and habitat in disturbed environments. Without this knowledge, the benchmark necessary for pre- and post-harvest comparisons is lacking. 1.3 Thesis Structure The core of this thesis is composed of two interrelated yet relatively distinct units, chapters three and four, each with separate introduction, methods, results, and discussion sections. Because the hydroriparian classification was developed specifically for Clayoquot Sound, characterisation of the region is appropriate so that the reader may judge the transferability of the classification to other areas; therefore, chapter two is a brief account of the climate, terrain, vegetation, and hydrology of the study area. An explanation of the hydroriparian 8 system classification, the basis of the study, and of the study channels themselves also are included in chapter two. Chapter three contains descriptions of the eight channel geomorphic units identified, the relative proportions of which formed the basis of the comparisons among channel classes. Channel classification criteria also received emphasis in order to explain why the relative proportions of certain channel units were greater in one class than another. Attempts to determine i f channel geomorphic units serve as distinctive habitat are summarized in chapter four. Invertebrate abundance and community structure were compared among habitat types. Information regarding the biology of the most abundant organisms was reviewed and, along with details of the physical characteristics of the habitat, was used to interpret the distribution of benthic organisms. Also addressed in chapter four is the variability in abundance and community structure among ephemeral, seasonal, and perennial flow regimes. The fifth and final chapter is a general discussion summarizing the major findings and a deliberation of the study's weaknesses. 9 Chapter 2: Physical Character of the Clayoquot Region and the Study Reaches 2.1 Introduction Climate, the product of geographical location and large scale topography (Schaefer, 1978), determines the temperature and precipitation of a region. Terrain, the result of rock type, tectonic history and climate, determines present landforms and the nature and distribution of surficial materials (Ryder, 1978). Subsequently, characteristics of drainage systems largely are determined by precipitation, topography, and surficial material. In addition, the morphology of small streams with restricted riparian zones in particular is influenced by upland forest vegetation (Triska and Cromack, Jr., 1980; Harmon et al, 1986; Naiman, 1991). The following description of the study area is to ensure that the reader can appraise the distinctiveness, or lack thereof, of channels, therefore the channel classification system, in the Clayoquot region. 2.2 Clayoquot Sound Area Description1 2.2.1 Climate The Clayoquot Sound Region is subject to dominantly onshore airflow and orographic lifting. Because the temperature of the Pacific Ocean changes very slowly, it moderates the temperature of the region year round, resulting in mild winters and cool summers (CSP, 1995). In addition, evaporation from the ocean surface is a major source of atmospheric 1 Sections 2.1.1 Climate and 2.1.2 Terrain are based on largely two references, Ryder (1978) and Harcombe and Oswald (1990). water vapor (Trewartha and Horn, 1980). Moisture laden westerly winds are forced to ascend when they encounter the Vancouver Island mountains and, as the air masses rise and cool, heavy precipitation falls onto the windward side of the mountain range. The Clayoquot Region is particularly wet in late autumn, winter, and early spring when almost continuous cyclonic storms bring heavy and prolonged precipitation as well as strong winds. In the spring, however, the westerlies weaken, rainstorms become less frequent and less severe, and high pressure systems prevail though there is no markedly dry season (CSP, 1995). Many areas along the coast, particularly those at high elevations, receive more than 2500 mm of precipitation annually (Schaefer, 1978). In general, total precipitation increases with distance from the outer coast (Ucluelet and Tofmo) to the western limit or windward side of the Vancouver Island mountains (Tofino Inlet) then decreases to the east or leeward side of the main mountain range (Port Alberni) (figure 2-1). Many of the study sites are located on the western slopes of the Vancouver Island mountains which is the highest precipitation zone in the region. Long-term measurements of precipitation in this area do not exist; however, data from recently established rain gauges near Tofino Inlet indicate that precipitation is approximately double that of the outer coast (Ucluelet and Tofino) (M. Church, pers. comm., 1998). Snowfall accounts for only a small proportion of total annual precipitation near sea level, albeit snow becomes more prevalent at higher elevations. However, "the elevation at which freezing occurs fluctuates widely: rain often falls at high elevations and snow sometimes falls at sea level" (CSP, 1995). 11 -4 -» 'B. c o E c E 1400 1200 -1000 -800 ->• 600 H 400 200 0 -Tofinolnlet — A - - Tofino —T— Ucluelet -•- EstevanPoint • CarnationCreek PortAlberni Figure 2-1: Mean monthly precipitation and temperature. Values are 20 - 30 yr averages except those for Carnation Creek (9 yrs) and Tofino Inlet (1 -2 yrs). Data sources: Environment Canada and Chapman Geoscience (Tofino Inlet). 12 The temperature moderating effect of the Pacific Ocean is most obvious along the coast while the range of temperatures increases farther inland (figure 2-1). Nonetheless, mean annual temperatures in the Coast Mountains and Islands Region, which encompasses Clayoquot Sound, are the highest in Canada at greater than 10 ° C (Schaefer, 1978). The mean freeze-free period is approximately 250 days along the coast and 150 days inland. Prolonged periods of subzero temperatures are uncommon (CSP, 1995); therefore, the soil is frost free with the exception of the immediate surface layers (Schaefer, 1978). 2.2.2 Terrain The Clayoquot Sound Region includes sections of two divergent physiographic subdivisions: the Estevan Coastal Plain and the Vancouver Island Mountains. The Estevan Coastal Plain is characterized by hummocky to almost flat land that is subdivided into numerous islands and peninsulas by inlets, channels, and Kennedy Lake. Steep, rocky hills interrupt the continuity of the plain further. The Vancouver Island Mountains in the region have valley sides commonly steeper than 60%, and are highly dissected and extremely rugged, with sharp ridgetops that exceed heights of 1000 m. Mountain peaks in the region may reach 1300 m which is somewhat lower than the highest peaks (2000 m) in central Vancouver Island. Summits closest to the coast descend to approximately 500 m (CSP, 1995). Major landforms and details of the landscape are determined by rock structure. Coarse crystalline metamorphic and intrusive rocks are the most widespread rock types in the Clayoquot area (CSP, 1995). Such rock types are generally associated with weather resistant (Jungen and Lewis, 1978), steep, and rugged terrain. Joints and faults constitute lines of 13 weakness where erosional processes create gullies and major topographic depressions. In the Clayoquot region, faults typically extend from northwest to southeast, roughly parallel to the coastline. Occupying relatively small areas are older volcanic rocks and sedimentary rocks, including limestone (CSP, 1995). The resistance of sedimentary rocks to weathering varies depending upon its composition and other factors determined during genesis. Limestones generally form upstanding, relatively resistant topography. Similarly, the topography underlain by volcanic rocks resembles that of sedimentary rocks of comparable mineral structure. Geomorphic processes during postglacial time have been controlled in part by the effects of the Fraser Glaciation, the last glaciation ending approximately 12,000 years ago. Surficial material in the Clayoquot Region, therefore, is a consequence of glaciation. Glacial till, a compact mixture of heterogeneous particles deposited directly by melting ice, is the most common glacial material. Most gentle to moderately steep slopes in the area are covered by till but, on steeper slopes, loose till is susceptible to mass wasting in the prevailing wet climate. Consequently, bedrock outcrops or consolidated material are exposed and colluvial deposits such as debris-flow fans and talus cones are created. In addition to hillslope processes, weathering and fluvial and marine processes have created landforms such as gullies, sea cliffs, alluvial fans, river terraces, and beaches, that we see today (CSP, 1995). Ferro-Humic Podzols are the dominant soil in the till and thin colluvial materials of the upper mountain slopes as well as in the marine and glaciofluvial coarse textured sediments of the well-drained, coastal plain sites. Because excess moisture and poor drainage are common, 14 much of the soil landscape is subject to continuous seepage; therefore, leaching of clay, organic matter, iron, and aluminum from upper to lower mineral horizons is intense (Jungen and Lewis, 1978; Meidinger and Pojar, 1991). In unstable areas, soil development is hindered by the slow creep of surficial materials. Vegetation on this highly productive forest land stabilizes the soil, reducing creep and erosion until a major disturbance reduces vegetative cover (Jungen and Lewis, 1978). Folisols, shallow organic soils overlying bedrock, occupy 20% - 40% of the Clayoquot Sound landscape (CSP, 1995). They are most widespread at high elevations and on the windward outer coast where forest productivity is relatively low (Jungen and Lewis, 1978). 2.2.3 Vegetation According to the biogeoclimatic system (Krajina, 1965), land units may be classified on the basis of climate, soil, and vegetation (Meidinger and Pojar, 1991). The Coastal Western Hemlock zone includes forests located at elevations less than 900 m which also are recognized as "coastal temperate rainforest". Western hemlock (Tsuga heterophylla ) is ubiquitous in this zone while western red cedar {Thuja plicata), amabilis fir (Abies amabilis), yellow cedar (Chamaecyparis nootkatensis), Sitka spruce (Picea sitchensis), Douglas-fir (Pseudotsuga menziesii), and red alder (Alnus rubra) occur under suitable conditions. Between 900 m and 1200 m above sea level, forests are part of the Mountain Hemlock zone and flora at elevations greater than 1200 m are located in the Alpine Tundra zone. Mountain hemlock (Tsuga mertensiana) is the dominant tree species in the Mountain Hemlock zone where yellow cedar and amabilis fir also are common (Meidinger and Pojar, 1991; CSP, 1995). The severe climate and lack of surficial material suitable for supporting tree growth prevalent in the Alpine Tundra zone does not permit tree species to become established. Isolated pockets of soil may support herbaceous plants such as red mountain-heather (Phyllodoce empetriformis), Merten's cassiope {Cassiope mertensiand), saxifrages (Family Saxifragaceae), and lichens (Harcombe and Oswald, 1990). Except where logging has occurred, Clayoquot Sound has an almost continuous, structurally complex, old-growth forest which is characterized by uneven canopies with gaps and all-aged stands (CSP, 1995). Microclimatic variables such as wind speed, short-wave radiation (light), relative humidity, soil moisture (Chen et al, 1995), diurnal changes in air and soil temperature, and seasonal patterns of snow accumulation and melt are affected by physical characteristics of the forest canopy. Specifically, forest and understory type, density, and structure provide various microhabitats for many plant and animal species: non-vascular, vascular, invertebrate and vertebrate organisms alike. Some important functions of non-vascular flora include water retention, nutrient cycling, and mycorrhizal production (CSP, 1995). 2.2.4 Hydroriparian Ecosystem and Hydrology The network of stream channels draining the watersheds of the Clayoquot area range in size from tiny rills at the highest elevations to rivers in major valleys. The aquatic component of the hydroriparian system is difficult to separate from the terrestrial component in the most headward portions of the drainage basin because often streams begin as seepage from saturated soil (CSP, 1995). Here, distinct channel beds and banks are not distinguishable. Also, the influence of riparian vegetation on the channel itself, and vice-versa, unify the two 16 systems. Riparian vegetation provides shade, leaf litter, woody debris, and bank stability to stream channels. The aquatic system deposits sediments on and erodes adjacent land and influences microclimate, soil moisture and nutrient conditions, providing critical support to semi aquatic animals. (Gregory et al, 1991; CSP, 1995; Bunnell et al, 1995). Rainstorms, particularly in winter, deliver large volumes of water onto the steep mountain slopes of the Clayoquot region. While some water is intercepted and taken up by vegetation, most rainwater is either absorbed into the soil or delivered as subsurface runoff to tributaries (Hetherington, 1987; CSP, 1995). Intense winter rain generates the highest stream discharges, particularly when it falls onto melting snow (CSP, 1995) although Hetherington (1987) noted that significant rain-ori-snow events are infrequent in the nearby Carnation Creek watershed. A consequence of the steep terrain and seasonally hyperhumid environment of Clayoquot Sound is frequent, high gradient, largely seasonal or ephemeral headwater channels. 2.3 Classification of the Hydroriparian System: Stream Environment In response to the British Columbia government's decision (April 13, 1993) regarding land use in Clayoquot Sound, the Scientific Panel for Sustainable Forest Practices in Clayoquot Sound was appointed and charged with scientifically reviewing forest practices and recommending more sustainable procedures. The Panel described the hydroriparian ecosystem as the "skeleton and circulation system of the ecological landscape" and recommended that the entire hydroriparian zone be designated a special management zone. A 17 classification for lotic systems was developed to provide a basis for management of the terrestrial surface adjacent to stream channels, to the limit of riparian influence (CSP, 1995). The reach scale is the level at which application of the hydroriparian classification was intended. A reach is a length of stream channel with homogeneous morphological, sedimentological, and hydrological features (Hogan and Church, 1989). Its length can be measured in meters to tens of meters in small, steep streams, or perhaps hundreds of meters or more in fifth-order and larger streams (Frissell et al, 1986). The hydroriparian system classification delineates classes according to four physical criteria: (1) bed material, (2) channel gradient, (3) channel entrenchment, and (4) bankfull width (figure 2-2). The Panel contended that these classification criteria influence the hydroriparian ecosystem tremendously. In addition, they allow users to determine channel classes during terrain mapping, a practical provision to support operational use. The most basic division of lotic systems separates channels depending on their ability to adjust their shape and gradient. Alluvial channels flow through their own deposits and, therefore, are able to adjust their form in response to changing conditions. Conversely, the morphology of non-alluvial channels is controlled by the material forming their bed and banks. Channel gradient determines important aspects of fluvial processes and morphology. For instance, gradient is directly related to both bed material load and clast size and inversely related to water discharge (in alluvial channels) (Schumm, 1977). Morphologic characteristics such as step frequency and pool length also are influenced by channel gradient (Church, 1992). Entrenched channels are those confined within fluvially eroded gullies or valleys of some depth (CSP, 1995). Valley-wall confinement restricts lateral channel A alluvial channels B non-alluvial channels 1 gradient <8% channel width - i <3m - ii 3-30m - iii > 30m 2 gradient >8% - i <3m - ii 3-30m - iii > 30m 1 gradient <8% a non-entrenched - i <3m ii 3-30m L iii >30m 2 gradient 8-20% a WOM-entrenched - i <3m - ii 3-30m - iii > 30m 3 gradient > 20% - b entrenched I— b entrenched -a MOH-entrenched _ perennial/ seasonal _ ephemeral b entrenched - i <3m - ii 3-30m - iii >30m - i <3m - ii 3-30m - iii > 30m Figure 2-2: Classification of the stream hydroriparian system (CSP, 1995). 19 migration, maximizes channel response to increased discharge by limiting overbank flow, and allows direct colluvial inputs into the channel which makes them prone to periodic disturbance from hillslope failures (Montgomery and Buffington, 1997). Channel width is a convenient measure of stream size, "for it more than any other easily measured characteristic of the channel is correlated with flow parameters such as average annual discharge and discharges having specific recurrence intervals" (Dunne and Leopold, 1978). Channel width is indicative of certain physical and biological stream processes (Vannote et al, 1980; CSP, 1995). Typically, channel size (width) increases with distance downstream along with discharge, depth, stream power, and velocity while bed material size, slope, and flow resistance index are inversely related to channel size (Knighton, 1987) hence characteristically decrease downstream. The river continuum concept (Vannote et al, 1980) asserts that heterotrophic processes, largely driven by allochthonous resources from riparian vegetation, are dominant in highly shaded headwater streams. Further downstream, where channel width increases, the riparian canopy opens, light levels rise, and autotrophic processes become increasingly important. Although the hydroriparian classification system was designed to be relatively straightforward, to avoid uncertainty and misinterpretation amid its users (CSP, 1995), the distinction within bed material and entrenchment categories often is not obvious. A bedrock reach lacks a continuous bed of active alluvium and indisputably is non-alluvial. 'Bedrock', in this context, can include non-lithified, non-alluvial lag material and armour material, as well as bedrock outcrops. Channels clogged with large boulders are common in areas formerly shaped by glacial processes (Abrahams et al, 1995), like Clayoquot Sound. This non-lithified, non-alluvial or colluvial lag material remains after channel degradation. On the other hand, an armour layer is created by fluvial processes. Armour material is composed of relatively coarse, tightly compacted grains which have been separated from finer particles during those events when flow discharge is capable of transporting most of the particles present. Gradually, the bed degrades, fine particles are less abundant, the dominant particles become coarser and transport rates decrease. Finally, transport of finer material that may underlie the armour is prevented. While bed armouring is very common in gravel-bed rivers (Andrews and Parker, 1987), it may be less important in channels with coarser substrates. Commonly, ambiguity arises when patches of transportable alluvium overlie non-alluvial beds. Under these circumstances, Montgomery et al (1996) considered reaches to be non-alluvial provided that the length of channel bed covered by alluvium did not extend beyond one channel width. Similarly, reaches were considered alluvial i f individual bedrock outcrops were no longer than one channel width. Roughness configurations of alluvial channels vary primarily with slope and position within the channel network (Montgomery and Buffington, 1997). For example, clasts may range from boulders in headwater streams to sand grains in lowland rivers. The frequency with which bed particles are moved depends upon the competence of the channel (Baker and Ritter, 1975), which is the largest grain size that can be transported under the prevailing hydraulic conditions. Fine substrates of sandy alluvial beds exhibit motion at moderate flow stages which occur once or twice each year, on average, (Baker, 1977) whereas coarse bed material is entrained during infrequent, extreme flood events (Howard, 1987) with recurrence intervals of approximately 50 years (Grant et al, 1990). "This is mainly because of the high 21 response threshold required to scour bouldery alluvium..." (Baker 1977). Shallow and ephemeral flow in many headwater channels is generally not capable of transporting colluvial sediment introduced to the channel, resulting in substantial storage of this material (Benda, 1990; Whiting and Bradley, 1993; Abrahams et al, 1995; Montgomery and Buffington, 1997). Such flow may rework a small proportion of the accumulated material, but it does not govern deposition, sorting, or transport of the substrate. Episodic movement by debris flows may account for most of the sediment transport in steep headwater channels (Benda, 1990; Montgomery and Buffington, 1997). The term alluvial, therefore, is inappropriate to describe small, boulder-bed channels and should be reserved for streams that frequently move their bed particles. In recognition of this, Montgomery and Buffington (1997) assigned the term colluvial to describe channels which are positioned at the extreme tips of drainage networks and which demonstrate sporadic fluvial transport. In this thesis, the term semi-alluvial is used to describe those channels exhibiting weak fluvial transport signaled by obviously structured clasts and woody debris. An entrenched channel is one that, as the result of downward fluvial erosion, is continuously confined within banks sufficiently high that overflow may not occur (CSP, 1995). Interpretation of the phrase "sufficiently high to prevent overflow" requires judgment which may perplex the inexperienced user of the classification system. In practice, entrenchment may be determined by the position of terrestrial vegetation in relation to the stream bed. On the sidewalls of a channel, the often abrupt change from vegetated to non-vegetated banks represents the upper limit of water flow which occurs frequently enough to prevent plant growth (Wood-Smith and Buffington, 1996). This may be considered the channel 'limit' or 'edge'. When rooted terrestrial vegetation and the stream bed are in juxtaposition, potential for bank overflow is indicated; therefore, by definition, the channel is not entrenched. The presence of a 'valley flat' adjacent to the stream bed, as depicted in figure 2-3-A, is part of the floodable surface and, therefore, is an example of non-entrenchment. Terrestrial vegetation established on the sidewalls of a channel suggests little potential for bank overflow. The channel, then, is likely entrenched (figure 2-3). A moss covered underside of channel spanning L W D was considered additional evidence to support entrenched assignments. Conversely, evidence of scour on the underside of channel spanning L W D supported non-entrenched designations. Because channel sidewalls are often discontinuous, entrenchment status changes every few tens of meters. The user, therefore, must determine whether or not the majority of the channel is entrenched. This may be accomplished objectively through simple measurements or more subjectively by relying on one's overall impression. 2.4 Study Reach Descriptions Study reaches are small, steep, and in pristine state. "Small" is a relative term and is used here to imply that channel scale is comparable to the scale of individual bed elements. For example, individual boulders (>256 mm) may span a small channel while several cobbles (>64 mm) may be required to bridge the entire width. The depth of small channels is typically 0.1-1.0 times the diameter of the largest grains and so the relative roughness (the ratio of grain diameter to water depth, D/d) is usually greater than 1.0 (Church, 1992). A 25 cm step in each metre of reach length is capable of controlling slope in small streams with 23 Figure 2-3: Determining (non-) entrenchment status by the position of terrestrial vegetation in relation to the stream bed. (A) stream bed and terrestrial vegetation are at the same level suggesting that bank overflow is likely. (B) Terrestrial vegetation established on sidewalls is evidence of channel entrenchment. (C) Terrestrial vegetation on the underside of L W D also is evidence of entrenchment. 24 gradients < 25%. A step of this height is comparable with the diameter of a small boulder which suggests that individual bed elements control not only roughness but also local gradient. Steeper gradients may be attributed to larger boulders, aggregate structures, or LWD. This definition of small stream is less appropriate for channels with bedrock, gravel, or sand beds. Therefore, channel width often is used as a surrogate measure of stream size. According to the stream hydroriparian system, a channel is small if it is < 3 m wide. Study reach width ranged from 1.3 to 4.9 meters. In general, a steep stream is one in which the dominant channel geomorphic unit sequence is the step-pool sequence, as opposed to the pool-riffle sequence which is dominant in low to moderate gradient streams (Heede, 1981; Whitakker and Jaeggi, 1989; Chin, 1989; Grant et al, 1990; Church, 1992). A low gradient channel is one with a bed slope of < 2% (Grant et al, 1990; Church, 1992). The boundary between moderate and high gradient channels, however, is not clear and may range from > 2% to 8% (Chin, 1989; Grant et al, 1990;Church, 1992; Abrahams et al, 1995; CSP, 1995). Study reach gradients range from 10% to 54%. While study reach widths intersect the boundary between small and intermediate stream sizes, reach gradients are high indicating headwater positions in the drainage network hence small streams. Study reaches were bordered by old growth forest, but natural disturbance emanating from upstream, such as landslide activity, was not assessed. Study reaches were assigned a stream hydroriparian class according to bed material, gradient, entrenchment, and width. Furthermore, study channels may be arranged into either temporary or permanent flow regime categories although this distinction is not formally incorporated into the stream hydroriparian system. Temporary flow sub-categories include ephemeral and seasonal (summer-dry) regimes. Ephemeral streams carry storm runoff derived from saturation seepage or from overland flow during and immediately following fairly intense rainstorm events (Goudie et al, 1994, CSP, 1995). Seasonal streams flow throughout the year with the exception of the summer dry period when water may be retained in depressions or in permanent trickles only (Dieterich, 1992; CSP, 1995). Permanent or perennial streams have regular base flow derived from either springs or persistent seepage through the banks (CSP, 1995). Since flow regime is defined by annual persistence of water, frequent observations throughout a typical year are required to determine flow regime definitively. A total of 17 reaches, distributed amongst six areas in the Clayoquot Sound region (Tofino Creek, Deer Bay, Nahmint River, Toquart River, Marion Creek and Maitland Mountain), were surveyed (figure 2-4). The number of study streams remained relatively small because the objective of the research was to establish the physical and ecological distinctiveness, or lack thereof, of specified stream types. For these purposes, stream morphology and benthic communities must be described in substantial detail. Hence, required labor is much greater compared to that required for reconnaissance. Study reach characteristics and classes are summarized in table 2-1. Since a major objective of this research is to determine the efficacy with which the CSP classification delineates physically distinct channels, the information included in table 2-1 is done so with the intent to familiarize the reader with the characteristics of each study reach as well as the CSP class to which it belongs. Subsequent analyses are based on groups of channels as reported in table 2-1. However, in section 2.3,1 asserted that the term semi-alluvial more accurately describes the substrates of many of the study reaches. I investigate this assertion through additional analyses discussed in a ensuing section, at which point, I will introduce a revised classification which incorporates a semi-alluvial distinction. Table 2-1: Summary of study reach characteristics and class. site substrate banks width (m) slope (%) flow regime class DB1 non-alluvial non-entrenched 1.8 ' 25 perennial B3ai DB2 non-alluvial entrenched 2.7 22 seasonal B3bi DB3 alluvial non-entrenched 2.1 18 seasonal A2i DB5 non-alluvial non-entrenched 2.8 28 perennial B3ai D B f non-alluvial entrenched 2.5 24 seasonal B3bi MRN1 non-alluvial non-entrenched 1.3 -30 perennial B3ai MRN2 non-alluvial entrenched 4.91 54 seasonal(?) B3bii MTD1 non-alluvial non-entrenched 4.2 48 ephemeral(?) B3aii NMT4 alluvial non-entrenched 2.0 19 ephemeral A2i NMT5 alluvial non-entrenched 3.4 20 ephemeral A2ii TCI non-alluvial non-entrenched 2.4 42 seasonal(?) B3ai T C l a non-alluvial non-entrenched 2.5 15 seasonal(?) B2ai T C l b non-alluvial non-entrenched 2.4 10 perennial B2ai TC3 non-alluvial non-entrenched 2.4 43 seasonal B3ai TC43 non-alluvial non-entrenched 3.8 13 perennial(?) B2aii T Q l non-alluvial entrenched 2.7 50 perennial B3bi TQ3 non-alluvial non-entrenched 3.2 41 ephemeral(?) B3aii 1 approximate width only Figure 2-4: Map of the study area and positions of study reaches (indicated by •). 28 Chapter 3: Channel Geomorphic Unit Representation in Stream Hydroriparian Classes 3.1 Introduction The basis of this chapter is that stream reaches may be characterized by the morphological units within them. Variability in morphological units influences not only channel morphology but also stream dynamics, such as flow hydraulics and sediment transport rates (Grant et al, 1990). The organization, structure and dynamics of the physical system, in turn, affect aspects of lotic communities such as nutrient uptake, algal and fish abundance and invertebrate production and diversity. Many research and management objectives, therefore, are best addressed at the spatial scale of channel units (Hawkins et al, 1993). For example, physical habitat evaluations are often an important tool for assessing the effects of human activities on a stream and its biota (Bisson et al, 1981; Roper and Scarnecchia, 1995; Bisson and Montgomery, 1996; Wood-Smith and Buffington, 1996). Channel geomorphic units are quasi-discrete areas of relatively homogeneous bed topography, water surface slope, depth and velocity (Frissell et al, 1986; Hawkins et al, 1993) that are typically one or more channel widths in length (Grant et al, 1990). Collectively, geomorphic units are various types of topographical lows and highs (Hogan and Church, 1989). When geomorphic units with similar bed slope, side slope, bank material, and width occur in succession, they constitute a reach. Pool-riffle sequences dominate low to moderate gradient streams and step-pool series are common in high gradient streams (Leopold et al, 1964; Heede, 1981; Whittaker and Jaeggi, 1982; O'Neill and Abrahams, 29 1984; Chin, 1989; Grant et al, 1990; Church, 1992). Pools are topographic depressions with lower energy gradients, finer grain sizes, and water deeper and slower than the reach average. Riffles are topographic highs which have higher energy gradients, coarser bed elements and water shallower and faster than the reach average (Richards, 1976b; Sullivan, 1986; Goudie et al, 1994). Steps are channel spanning accumulations of boulders and cobbles that alternate with small secondary pools, which are filled with less coarse material, to create a longitudinal profile resembling a staircase (Chin, 1989). Grant et al (1990), however, acknowledged a broader range of bedforms in high gradient, boulder-bed, mountain streams. In addition to pools, riffles and steps, they characterized rapids and cascades on the basis of roughness and flow characteristics. Their nomenclature was adopted when possible and their descriptions, in part, have been incorporated into section 3.3.1 "Geomorphic Unit Descriptions". During high magnitude flood events, water may move large cobbles and boulders against a keystone, creating a boulder step which, subsequently, traps additional particles (Sullivan, 1986; Church, 1992). The function of such slope adjustments is to counteract steep gradients that otherwise would cause excessive erosion (Heede, 1981). The mechanism of formation, however, remains poorly understood (Grant et al, 1990). The antidune theory of step-pool origin in mountain streams is acknowledged but controversial (Whittaker and Jaeggi, 1982; Abrahams et al, 1995). Antidunes are waves in sand beds that are formed in association with a standing water wave, where the crest of the bed undulation is in phase with that of the water surface. Whittaker and Jaeggi (1982) reported that, in flume experiments, high intensity flow events transformed an initially plane bed of cobbles (slope > 7.5%) into a bed of transverse ribs and concluded that the process was basically the same as that which produces antidunes. 30 Development of a coarse armour layer accompanied step formation resulting in a large increase in resistance to flow (Whittaker and Jaeggi, 1982) which has been interpreted to indicate that step-pool structure attains maximum stability (Abrahams et al, 1995). The maximum friction factor theory (MFF) suggests that natural channels adjust their form to maximize flow resistance (Abrahams et al, 1995) and is inconsistent with the antidune theory of origin. Step-pools form when Froude numbers fall below the range of those usually associated with antidunes (F r = 1). Froude number is a dimensionless ratio of inertial to gravity forces in flowing water (Goudie et al, 1994): Fr = vl{<gd) (1) where v is velocity, g is acceleration due to gravity, and d is flow depth (Jarrett, 1984). In addition, the M F F model does not require that the clasts forming the steps be submerged, a situation shown to exist (Wohl and Grodek, 1994 referenced in Abrahams et al, 1995), whereas the antidune model does. While M F F theory does not clarify the mechanism forming step-pool structure it does add to the controversy surrounding the antidune theory of origin. Water velocity in a channel depends on the depth and surface slope of the water and inversely on the boundary resistance estimated by a roughness coefficient. This relation is expressed by both the Chezy Formula: v = C(RS)'/2 (2) and the Manning relation: 31 v = (l/n)(R2/3S"2) (3) where C is the Chezy resistance factor, R is the hydraulic radius (approximated by the mean depth), S is the energy gradient (approximated by the water surface slope), and n is the Manning resistance coefficient (Dunne and Leopold, 1978; Gore, 1996). Primarily based on flume experiments, standard hydraulic theory suggests that when slope increases, higher velocity results (Jarrett, 1984). While stream hydraulics was not a primary focus of this study, water velocity was investigated because, from the position of stream animals, current is the most significant feature of running water (Hynes, 1970). 3.2 Methods 3.2.1 Data Collection During field reconnaissance, more than 60 stream reaches were classified according to the stream hydroriparian system. Seventeen reaches representing eight classes (A2i, A2ii , B2ai, B2aii, B3ai, B3aii, B3bi, and B3bii) were selected for the study. Using the descriptions of Grant et al (1990) as a guide, channel geomorphic units were classified during reach surveys into one of eight groups: pool, glide, riffle, bedrock chute, rapid, boulder cascade, bedrock cascade and fall. A combination of bed slope and dominant bed material type and organization helped define the units, which are thoroughly described in section 3.3.1. Substrate dominance and organization were estimated visually. To describe grain size distributions accurately no individual stone should exceed 0.1% of a sample. Relaxing this criterion results in a reduction in the assurance that the distribution has been well characterized (Rood and Church, 1994). Sampling according to these stringent criteria would 32 be impossible logistically in the study reaches. However, in situ measurements of the population of framework clasts, those that define channel planform and breaks, would be possible given adequate time. After each channel unit was identified its width, length, current velocity2, and bed slope were measured and recorded on scaled diagrams (figure 3-1). The units themselves were represented by unique symbols. Active channel width of each morphological unit was measured with a 2 m metal retractable or a 30 m cloth measuring tape. The active channel width is that portion of the channel where flow prevents the establishment of terrestrial vegetation (Wood-Smith and Buffington, 1996) and, therefore, is a measurement extending from the lower vegetation limit on one bank to that of the other bank. One to several width readings were taken depending upon the length and the width variability of the channel unit. A weighted average active channel width for the entire surveyed reach was calculated. Channel unit length was determined with the aid of a hip chain. When discharge permitted, water surface velocity was measured as the duration taken by a cork to float downstream a specified distance, depending upon the length of an unobstructed path. Velocity measurements were repeated a minimum of three times, averaged, and converted to units expressed in cm/sec. Water velocity was difficult, at best, and often impossible to measure in those channel units with large roughness elements, for instance, boulder cascades. Furthermore, no simple formula is available to estimate mean flow velocity from observed surface velocity owing to the variability of velocity profiles as channel roughness changes 2 Velocity was measured in those units sampled for benthic macroinvertebrates only. 33 LEGEND tree * boulder stump L W D fall cascade bedrock riffle pool Scale 1:100 <xoy. ted, Figure 3-1: A stylized example of a scaled diagram produced during reach surveys. Trees and stumps were recorded when their roots were incorporated into the channel bed and/or banks. Major pieces of L W D were recorded when any portion was within the active channel. 34 with depth of flow. Although Jarrett (1984) developed equations to predict Manning's n, flow velocity, and discharge of high-gradient channels, the upper limit on slope (4%) is much lower than the gradients of the study reaches. Channel unit bed slope was measured over its entire length with a clinometer, provided that this was possible within the constraints of the instrument and human maneuverability. Weighted average bed slope was calculated in order to determine the gradient of the entire reach. Additional techniques were used to calculate the overall slope gradient when the study reach was near the limits of a gradient category (i.e. near 8% or 20%). These methods were: (1) the difference between upstream and downstream altimeter readings, (2) proportion of stream area with gradient < 20% and > 20%, and (3) rise over run determined by longitudinal profiles prepared using survey information. Study reaches near gradient category limits and their bed slopes, as determined by each of the four methods, are summarized in table 3-1. The rise over run method was considered more reliable than either the altimeter or the proportion method because the scale of the pocket altimeter was marked at 10 m intervals, thus limiting precision, and the proportion method disregarded gradient magnitude. For example, i f two channel units had gradients of 1% and 19%, they were both considered as < 20%. The rise over run method tended to verify the decision of the weighted average technique. In general, 30-40 m of stream length were surveyed per metre stream width to ensure inclusion of several successive channel units. In larger channels with low to moderate gradients channel units iterate every five to seven channel widths (Leopold et al, 1964; 35 Richards, 1976a; Keller and Melhorn, 1978) and every three to four channel widths in smaller, steep streams (Church, 1992). Unit spacing appears to be related to the size of the channel (Keller and Melhorn, 1978; Chin, 1989) as well as the gradient (Heede, 1981; Church, 1992). Channel unit length in a stream with a single metre width is of the order of three to four metres; hence 30-40 metre surveys captured ten repeating channel units. Table 3-1: Study reach gradients as determined by four methods. site altimeter weighted average % area > 20% rise/run DB1 N / A 22% 44% 25% DB2 22.2m 21% 40% 22% D B f N / A 22% 38% 24% MRN1 36.6m 25% 45% 26% 3.2.2 Data Analysis The focus of the analyses is on the relative area of geomorphic units among groups of channels. The relative proportion of stream area in various channel units is considered to be a meaningful indication of in-stream habitat condition (Ralph et al, 1994) because the availability of specific units influences the biotic community (Bisson et al, 1981; Roper and Scarnecchia, 1995). It is assumed, for now, that geomorphic units represent distinctive habitats. The validity of this assumption will be investigated in Chapter Four. First, from width and length measurements, the area of each channel unit was calculated. Second, areas of a given type of channel unit within a single stream were summed. Third, the area of each channel unit type was expressed as a proportion of stream area. Fourth, streams were combined into groups based on channel classification criteria (bed material, gradient, 36 entrenchment, and width) and the mean relative proportion of area in each channel unit type was compared in order to investigate the success of each criterion as a distinguishing characteristic. A t-test for independent samples with separate variance estimates was employed to evaluate the difference in the mean relative proportion of area (hereafter referred to as 'mean area') in each channel unit type between two groups. For example, alluvial channels were compared to non-alluvial channels. Regrouped data and separate analyses were used to compare reaches with gradients < 20% and > 20%, reaches that are entrenched and non-entrenched, and reaches that are <3 m wide and > 3 m wide. Fifth, channels were grouped according to the stream hydroriparian classification system. Finally, t-tests were performed on all combinations of pairs in order to investigate the difference in mean area in each channel unit type between two stream hydroriparian classes. Because the standard deviation varied directly as the mean, the data were log 1 0 (x+1) transformed which brought about equality of variances but did not bring about normality (see page 52) which are two assumptions of a t-test. When the means and standard deviations are correlated, the performance of the test statistic deteriorates, thus increasing the probability of a Type I error (Cochran and Cox, 1950). T-tests with separate variance estimates were necessary because when the variances of two groups are widely different, and the number of observations in the groups are unequal, the variance of the difference for the dependent variable should not be estimated from the pooled within group variances. The statistical significance of the difference may not be detected by a t-test computed in this manner (Hicks, 1993). The deliberate measurements of selected physical variables technically do not qualify as proper scientific experiments as there is no effective control over the subject of interest. However, statistical experiments with successful control through replication, prescreening, and blocking of experimental units and ex post facto separation of concomitant information may be accepted as proper field experiments (Church, 1984). Having established that this observational study may qualify as an experiment, it may be classified as a "discovering" experiment which serves as a means to secure acceptable scientific evidence of a relationship. In discovering experiments, p < 0.1 is an acceptable level of detection of "meaningful" differences between groups; however, such exploratory results require independent confirmation. As in any experiment, higher p-levels (i.e. p < 0.01) indicate more definitive relationships; however, subsequent verification via an independent study is required nevertheless (Flueck, 1978). 3.3 Results 3.3.1 Geomorphic Unit Descriptions Fall Falls are vertical steps typically created by boulders or woody debris oriented transverse to the channel that may or may not span the entire width; however, they are distinct from upstream and downstream units (Grant et al, 1990) (figures 3-2-D and 3-3-A). In the study reaches, mean fall height is 0.88 m and, by definition, flow is in the supercritical regime (i.e. Froude number > 1). Because falls are vertical and have no run, height measurements were 38 Figure 3-2: Definition diagrams (longitudinal and plan views) for channel geomorphic units: (A) boulder cascade, (B) rapid, (C) riffle, (D) fall and pool, and (E) glide. 39 D Figure 3-3: Channel units in various study reaches. (A) Fall: water plunges over small woody debris stabilized by boulders; (B) bedrock cascade: water flows directly on steeply inclined bedrock outcrop that is free of alluvium; (C) boulder cascade: flow tumbles over stable, moss-covered boulders into pocket pools; and (D) rapid: step-pool sequence is obvious. 40 Figure 3-3 (continued): Channel units in various study reaches. (E) Chute: water, mainly in the subcritical regime, flows directly on the bedrock outcrop where some colluvial/alluvial material has been deposited; (F) riffle: relatively small, uniformly distributed bed elements are submerged; and (G) pool: small woody debris dam retains tranquil flow. 41 substituted for that of length. Fall area, used in subsequent analyses, was calculated as height x width. The vertical incline of falls suggests that their gradients approach vertical. Bedrock Cascade Bedrock cascades are non-alluvial rock outcrops characterized by either a notched or smooth appearance, depending upon the nature of the parent material (figure 3-3-B). Water flows directly on the bedrock surface. Bedrock reaches are generally steeper than alluvial reaches of comparable size (Howard, 1987; Montgomery et al, 1996) (figure 3-4) and are free from alluvium owing to high transport capacities associated with steep gradients (Montgomery and Buffington, 1997). The mean gradient of bedrock cascades was 49% (table 3-2). Table 3-2: Channel geomorphic unit bed slope summary. channel mean bed slope median bed slope geomorphic unit (%) (±SE) bed slope (%) range (%) sample size fall - W —> OO - » oo 80 bedrock cascade 48.99 (3.22) 50 14-80 33 boulder cascade 44.98 (2.05) 47 15-80 53 rapid 20.38 (0.86) 20 9-35 55 bedrock chute 12.4 (2.41) 10.5 2-29 10 riffle 8.74 (0.58) 8 -3-25 61 glide 6.40(1.57) 6 1 - 10 5 pool 3.24 (0.46) 3 -14-20 98 1 The symbol -> oo indicates that bed slope approaches vertical. Boulder Cascade Boulder cascades are characterized by tightly imbricated (Chin, 1989) boulders which are partially emergent at low and intermediate flow; therefore, the relative roughness (D/d) is 42 120 -20 if channel geomorphic unit Figure 3-4: Box plot of channel geomorphic unit vs. bed slope. The box extents indicate the 25th and 75th percentiles, horizontal line inside box indicates the 50th percentile, capped bars indicate 10th and 90th percentiles, and symbols mark data points outside the 10th and 90th percentiles. 43 high (> 1) (Grant et al, 1990; Church, 1992; Montgomery and Buffington, 1997). Boulder cascades, in general, are non-alluvial; therefore, they exhibit no apparent lateral or longitudinal organization. However, boulder cascades may be arranged to achieve maximum stability (M. Church, pers. comm., 1997). Interspersed amongst the boulders are pocket pools which are not channel spanning, are less than one channel width in length, and are areas where relatively smaller grains accumulate. In the study reaches, mean slope gradient of boulder cascades is 45% (figures 3-2-A, 3-3-C, 3-4, and table 3-2). Rapid Rapids are consistent with step-pools frequently treated in the literature (Whittaker and Jaeggi, 1982; Chin, 1989; Abrahams et al, 1995); however, step-pool is a phrase commonly applied at the reach scale. Rapids are characterized by bed elements arranged into ribs which are oriented transverse to the channel and are partially emergent at low and intermediate flow. Ribs, or steps, are separated by secondary pools which are not channel spanning, are less than one channel width in length, and are areas where relatively smaller clasts accumulate (Chin, 1989; Grant et al, 1990; Montgomery and Buffington, 1997). Relative roughness (D/d) is high (>1) at the ribs and lower in the secondary pools (< 1) (figures 3-2-B, 3-3-D, and 3-4). Although ribs composed of alluvial boulders and cobbles are most common, steps also may be controlled by lithology in bedrock channels or by large woody debris in forested channels (Chin, 1989). Rapids are distinguished from boulder cascades by the size and organization of bed elements, by the size and spacing of pool areas, and by the characteristic gradient. The clasts of rapids are more organized and less uniformly sized than the coarser, less ordered clasts of boulder cascades. Secondary pools are more regularly spaced and larger in area than pocket pools. The bed slope of boulder cascades typically is greater than that of rapids. The mean slope gradient of rapids in the study reaches is 20% (table 3-2). Chute Bedrock chutes are non-alluvial rock outcrops characterized by either a notched or smooth appearance, depending upon the nature of the bed material (figures 3-3-E and 3-4). Water flows directly on the bedrock surface. The mean bed slope of chutes in the study reaches is 12% (table 3-2). Chutes are distinguished from bedrock cascades primarily by gradient. Hawkins et al (1993) suggest that chutes occur where bedrock has narrow, steep slots and sheets occur where shallow water flows uniformly over smooth bedrock of variable gradient. Chutes, as defined in this study, are not inconsistent with both chutes and sheets of Hawkins, however, unit gradient receives greater emphasis. Bedrock chutes were fairly uncommon, comprising only 2.5% of enumerated channel geomorphic units. Riffle Riffles are composed of primarily uniformly distributed, alluvial gravel and cobbles (figures 3-2-C, 3-3-F, and 3-4). The relatively small grain size (2.0mm to 256mm diameter) contributes to low relative roughness (D/d < 1). The mean riffle bed slope in the study reaches is 9% (table 3-2). 45 Glide Glides are transitional units that possess attributes common to both pools and riffles (Bisson et al, 1981; Sullivan, 1986). Church (1992) described them as 'extended riffles' or relict pools that have been filled by sediment. The substratum of glides is not uniform, resembling that of pools, while the depth and flow are more similar to those of riffles (figure 3-2-E and 3-4). Glides may be common at pool-riffle and/or pool-rapid breaks (Bisson et al, 1981; Church, 1992). The mean gradient of glides in the study channels is 6% (table 3-2). Glides were uncommon, comprising only 1.3% of enumerated channel units; therefore they were removed from the data set. Because they are sufficiently distinct from other units, glides are mentioned here but they will not be treated beyond this description. Primary Pool Primary pools are topographic depressions with a non-uniform substratum; therefore the larger clasts may be partially emergent at low flow (figures 3-2-D, 3-3-G, and 3-4). Pools created by plunging water, such as a fall, generally are deepest where the fall strikes the bed and forms a small depression or plunge hole, and are most shallow where alluvium accumulates at their downstream ends (Grant et al, 1990). Such characteristics depend upon the agent of pool formation. For example, obstructions retain water in dammed pools; therefore, a plunge hole and shallow end do not exist. However, characteristic of both plunge and dammed pools is a downstream control. Pools in the study reaches had a mean bed slope gradient of 3% (table 3-2). When water was present, the water surface slope was frequently, but not always flat, hence the main longitudinal current often was apparent. Water surface slope, however, was not measured formally. 46 Relative flow velocity of channel units is presented in figure 3-5-A. The progression of units from high to low velocity closely approximates the progression of units from high to low gradient (figure 3-4) suggesting that slope influences water velocity. Figure 3-5-B shows that velocity is significantly correlated with but is not closely determined by bed slope. 3.3.2 Channel Class Distinctions Hydroriparian System Classification Criteria The variation between channels < 3 m wide and channels > 3 m wide, with respect to the mean areas of seven geomorphic units (bedrock cascade, boulder cascade, chute, fall, pool, rapid, riffle), was assessed through the use of t-tests. Similarly, channels were rearranged to investigate the differences among entrenched and non-entrenched channels, alluvial and non-alluvial channels, and channels with gradients < 20% and >20%. The data are summarized in table 3-3. The data show that reaches with active channel widths < 3 m wide were significantly different from reaches with active channel widths > 3 m wide with respect to mean pool area only (table 3-3-A). Therefore, for the purposes of further analyses, the width distinction was disregarded and width classes were merged thus decreasing eight channel classes to four. This merger increased sample size, or replicates, thereby providing an estimate of error variability for all channel classes by which the significance of hypothesis tests may be judged (Snedecor and Cochran, 1967). Figure 3-5: (A) Box plot of channel unit vs. water velocity (see figure 3-4 explanation of box plot). (B) Scatter plot of water velocity vs. slope. The correlation coefficient is significant (p < .0005). 48 Table 3-3: Comparisons of mean area in geomorphic units among paired groups of reaches divided by a (A) width, (B) entrenchment, (C) bed material, and (D) gradient criterion. degrees of dependent variable mean area t-value freedom p-level A. independent variable: width <3 m >3 m fall 4.8% 3.3% 0.478 11 0.642 bedrock cascade 12.9% 20.0% 0.307 6 0.770 boulder cascade 18.7% 19.8% 0.244 7 0.814 rapid 17.5% 26.6% -1.139 9 0.284 chute 3.4% 1.0% 0.520 12 0.612 riffle .21.3% 16.3% 0.815 5 0.452 pool 20.7% 8.9% 2.269 5 0.073* B. independent variable: entrenchment entrenched non-ent. fall 4.7% 4.2% 0.352 4 0.743 bedrock cascade 20.3% 13.3% 0.304 5 0.773 boulder cascade 24.6% 17.3% 0.735 5 0.495 rapid 18.9% 20.6% 0.415 8 0.689 chute 1.1% 3.2% -0.250 9 0.808 riffle 10.1% 22.9% -1.847 4 0.138 pool 20.3% 16.3% 0.162 4 0.879 C. independent variable: bed material alluvial non-alluvial fall 4.0% 4.4% -0.267 4 0.781 bedrock cascade 0 18.1% 5.149 13 0.001*** boulder cascade 0 23.1% 5.745 13 0.001*** rapid 40.9% 15.7% -2.779 7 0.027** chute 0 3.3% 2.772 13 0.016** riffle 37.7% 16.0% -3.399 6 0.015** pool 13.1% 18.1% 0.515 3 0.641 D. independent variable: gradient < 20 % > 20 % fall 3.3% 4.8% -1.557 8 0.158 bedrock cascade 6.4% 19.6% -1.513 11 0.157 boulder cascade 3.6% 27.4% -2.584 13 0.023** rapid 26.0% 17.0% -0.244 6 0.816 chute 2.4% 2.9% -0.102 10 0.921 riffle 33.5% 12.4% 4.487 15 0.001*** pool 21.0% 15.2% 0.834 10 0.424 1. Asterisks indicate p < 0.10 (*),p < 0.05 (**), and p < 0.01 (***). 2. Because glide area was eliminated from the analyses, columns do not sum to 100%. In addition, stream area covered by LWD was eliminated from the analyses. Entrenchment did not affect mean channel unit area (table 3-3-B). However, because sample sizes are small (4 entrenched channels and 13 non-entrenched channels), this result gives only a weak confirmation of the null hypothesis. Because bed material and gradient of the non-entrenched streams are more variable than those of the entrenched streams, non-entrenched, non-alluvial channels were compared to entrenched, non-alluvial channels to eliminate some of the variability between groups. Nevertheless, t-tests failed to show that the mean area of any channel unit varied significantly (p > .21) between entrenched and restricted non-entrenched groups. A more reliable test still would be to compare channels that are different in terms of their entrenchment status only, while other variables are constant (i.e. B3a vs. B3b channels). The results of this comparison will be discussed on a subsequent page. The data show that alluvial and non-alluvial categories varied significantly with respect to mean area in bedrock and boulder cascades rapids, chutes, and riffles (table 3-3-C). One might expect mean area in channel units with contrasting bed material to differ between alluvial and non-alluvial channels, for example, rock units (bedrock cascades and chutes), large clast units (boulder cascades) and small clast units (riffles). Rapids, intermediate clast units, do not concur with such reasoning. It is reasonable to expect that mean area in channel units with steep bed slopes will differ from mean area in channel units with mild bed slopes when gradient classes are compared. Mean boulder cascade and riffle area were significantly different between gradient classes (table 3-3-D); however, mean areas of falls, bedrock cascades, and pools were not different between gradient categories. In general, large sample sizes are required to detect differences, hence significant results, in highly variable data. Because the bedrock cascade data set is both small and variable (X=.06 and a/Vn=.04 for gradients < 20% and X=.20 and cWn=.07 for gradients > 20%), a type II error, non-rejection of H 0 when H , is true, may have been committed. Neither falls nor pools were effective discriminators amongst gradient categories, or, for that matter, amongst any group, suggesting they are ubiquitous units not affected by channel gradient within the range of the study reaches. The results indicate that mean area of geomorphic units was not influenced by channel gradient to the degree that would be expected, considering that unit descriptions are based largely on bed slope. This suggests that gradient divisions other than 20% may be critical. Because the mean bed slope of rapids is 20%, they are split in the assessment of gradient classes; therefore, 15% and 25% gradient divisions were chosen for investigation. However, 25% falls just outside the 50 t h percentile of rapids and 15% bisects the 50 t h percentile of chutes (figure 3-4). Figure 3-4 indicates that geomorphic unit bisection is unavoidable as a result of the wide and overlapping gradient ranges of channel units; however, a 25% gradient division interferes with fewer units than any other reasonable boundary. The results of the comparisons are summarized in table 3-4. The data show that riffle area was significantly higher in channels with gradient < 15% than in channels with gradient > 15%. Riffle area also was significantly different in channels with 51 Table 3-4: Comparisons of mean area in geomorphic units among (A) reaches with gradients greater than and less than 15% gradient and (B) reaches greater than and less than 25% gradient. dependent degrees of variable mean area t-value freedom p-level A . independent variable: 15% gradient division < 15 % > 15 % fall 4.7% 4.2% -.688 2 .563 bedrock cascade 8.1% 16.4% -.730 3 .518 boulder cascade 2.3% 22.6% -1.923 4 .127 rapid 25.0% 19.1% -.099 2 .930 chute 4.9% 2.3% .917 3 .427 riffle 29.8% 17.7% 2.800 11 .017** pool 20.3% 16.6% .073 2 .948 B. independent variable: 25% gradient division < 25 % > 25 % fall 4.1% 4.5% -.760 15 .459 bedrock cascade 8.2% 22.6% -.896 13 .387 boulder cascade 7.4% 32.1% -1.479 13 .163 rapid 22.9% 17.2% -.305 13 .765 chute 3.8% 1.5% .823 15 .424 riffle 28.8% 9.7% 4.049 12 .002*** pool 22.4% 11.4% 1.781 14 .097* 1. Asterisks indicate p < 0.10 (*),p < 0.05 (**), and p < 0.01 (***). 2. Because glide area was eliminated from the analyses, columns do not sum to 100%. In addition, stream area covered by L W D was eliminated from the analyses. Table 3-5: Relative proportions of channel class area in geomorphic units. Channel Bedrock Boulder Class Fall Cascade Cascade Rapid Chute Riffle Pool A2 4.0% 0 .03% 40.9% 0 37.7% 13.1% B2a 1.4% 12.8% 7.1% 11.0% 4.9% 29.2% 28.9% B3a 4.8% 19.2% 29.0% 16.0% 3.8% 13.8% 12.2% B3b 4.7% 20.3% 24.6% 18.9% 1.1% 10.1% 20.3% 1. Bold typeface indicates the (co)dominant units in each class. 52 gradients above and below 25%, as was pool area. In general, neither a 15% nor a 25% gradient division successfully separated channels on the basis of their high gradient units. A 25% gradient division, however, successfully distinguished channels on the basis of their low gradient units. Channel Classes Rapid and riffle area exceed that of other units in class A2 channels. In class B2a channels, riffles and pools exhibit a clear dominance over other channel units. Boulder cascade area outweighs that of other channel units in both B3a and B3b channels (table 3-5). While channel unit proportions of class area exhibit clear dominance, these numbers do not indicate the variability of channels within a class which will have a major impact on the outcome of significance tests. Figure 3-6 illustrates the mean, standard error, and standard deviation of class area in falls, bedrock and boulder cascades, rapids, chutes, riffles, and pools. To investigate the significance of the differences amongst channel classes, with respect to the mean area in each channel unit, t-tests were performed on all combinations of pairs and are reported in table 3-6. Because the variables were not distributed normally, even after they were log 1 0 (x+1) transformed, Mann-Whitney U-tests were applied to all combinations of pairs. The Mann-Whitney U test is a non-parametric and less robust equivalent of the parametric t-test. The results of the U-tests, also reported in table 3-6, were not appreciably different from the t-tests, suggesting that the results are meaningful indeed. The data show that the variation in channel classes A2 and B2a, with respect to mean area in each channel unit, was not significant with the exception of rapids when the outlier in TC43 53 B2a B3a B3b Figure 3-6: Box plots of channel class vs. proportion of class area in (A) falls, (B) bedrock cascades, (C) boulder cascades, (D) rapids, (E) chutes (F) riffles, and (G) pools. The box extents indicate the standard error of the mean, horizontal line inside the box indicates the mean, and capped bars indicate the standard deviation. 54 Table 3-6: T-test and Mann-Whitney U-test results of pairwise comparisons among channel classes with respect to mean area in geomorphic units. t-value p-level (separate (two-tailed variance critical dependent variable estimates) df region) U-value p-level A. Class A2 vs. Class B2a fall -1.696 4 0.165 1.000 0.248 bedrock cascade 1.979 2 0.186 1.500 0.190 boulder cascade 1.903 2 0.197 2.000 0.275 rapid1 -2.005 2 0.183 1.000 0.127 rapid2 10.674 2 0.009*** 0.000 0.083* chute 1.699 2 0.232 1.500 0.190 riffle -0.865 3 0.451 2.000 0.275 pool 1.685 2 0.234 0.000 0.050** B. Class A2 vs. Class B3a fall 0.426 4 0.692 10.000 0.909 bedrock cascade 3.759 6 0.009*** 3.000 0.087* boulder cascade 4.421 6 0.004*** 2.000 0.053* rapid -2.252 6 0.065* 2.000 0.053* chute 1.534 6 0.176 7.500 0.494 riffle -3.703 5 0.014** 1.000 0.030** pool 0.043 3 0.968 10.000 0.909 C. Class A2 vs. Class B3b fall 0.060 5 0.954 6.000 1.000 bedrock cascade1 2.243 3 0.111 1.500 0.112 bedrock cascade2 1.780 2 0.217 1.500 0.190 boulder cascade 2.827 3 0.066* 2.000 0.157 rapid -1.582 5 0.175 2.000 0.157 chute 1.711 3 0.186 3.000 0.289 riffle -2.900 4 0.044** 0.000 0.034** pool 0.427 5 0.687 5.000 0.724 1 outlier included in data set 2 outlier eliminated from data set Asterisks indicate p < 0.10 (*),p < 0.05 (**), and p < 0.01 (***). 55 Table 3-6 (continued): T-test and Mann-Whitney U-test results of pairwise comparisons among channel classes with respect to mean area in geomorphic units. t-value p-level (separate (two-tailed variance critical dependent variable estimates) df region) U-value p-level D. Class B2a vs. Class B3a fall 2.216 3 0.113 1.000 0.079* bedrock cascade 0.275 4 0.797 8.000 0.569 boulder cascade 1.027 4 0.362 7.500 0.494 rapid1 1.107 2 0.383 7.000 0.425 rapid2 7.282 6 0.001*** 0.000 0.040** chute -0.578 4 0.594 7.500 0.494 riffle -3.557 7 0.009*** 1.000 0.030** pool -3.342 6 0.016** 0.000 0.017** E. Class B2a vs. Class B3b fall 1.575 5 0.176 3.000 0.643 bedrock cascade1 0.170 5 0.987 5.500 0.860 bedrock cascade2 -0.583 4 0.591 2.500 0.383 boulder cascade 0.771 5 0.476 2.500 0.216 rapid1 1.225 3 0.308 3.000 0.289 rapid2 5.897 3 0.010*** 0.000 0.064* chute -0.838 3 0.464 4.000 0.480 riffle -2.600 4 0.060* 1.000 0.077* pool -1.242 3 0.302 3.000 0.289 F. Class B3a vs. Class B3b fall 0.277 4 0.796 14.000 1.000 bedrock cascade1 0.277 6 0.791 12.000 0.705 bedrock cascade2 1.107 5 0.319 5.000 0.210 boulder cascade 0.103 6 0.921 13.500 0.925 rapid -0.327 6 0.755 12.000 0.705 chute 0.286 9 0.781 13.000 0.850 riffle 1.043 4 0.356 9.000 0.345 pool -0.516 5 0.628 11.000 0.571 1 outlier included in data set 2 outlier eliminated from data set Asterisks indicate p < 0.10 (*),p < 0.05 (**), and p < 0.01 (***). was eliminated (table 3-6-A and figure 3-7-B). This result suggests that bed material does not adequately distinguish these two classes; therefore, the suitability of simply alluvial and non-alluvial distinctions was questioned once again. This result, coupled with impressions from field observations discussed in section 2.2, prompted a re-classification of the study channels which excluded the alluvial distinction and included a semi-alluvial criterion (table 3-7). A t-test showed that mean area in only two channel units, boulder cascades and riffles, was different between channels with semi-alluvial and non-alluvial substrates (p = .072 and .029, respectively). The number of channel units with a statistically significant effect on the outcome is appreciably smaller than that of comparisons between alluvial and non-alluvial categories (table 3-3-C). Table 3-7: Reclassification of study reaches to recognise a semi-alluvial category. stream original (CSP) bed material assignment revised bed material assignment DB 1 non-alluvial non-alluvial D B 2 non-alluvial semi-alluvial DB 3 alluvial semi-alluvial DB 5 non-alluvial semi-alluvial DB f non-alluvial non-alluvial M R N 1 non-alluvial semi-alluvial M R N 2 non-alluvial non-alluvial M T D 1 non-alluvial non-alluvial N M T 4 alluvial semi-alluvial N M T 5 alluvial semi-alluvial TC 1 non-alluvial semi-alluvial TC l a non-alluvial semi-alluvial TC lb non-alluvial semi-alluvial TC 3 non-alluvial non-alluvial TC 43 non-alluvial semi-alluvial TQ 1 non-alluvial non-alluvial TQ3 non-alluvial non-alluvial 57 a % o o c o t i o O h 2 O h ^ / ^ Jt ^ * / / / / # / 3 o 1.0-0.84 DB 5 MRN 1 DB 1 TC 3 TC 1 —•— TQ3 —•-• MTD 1 DB f —•— TQ 1 DB 2 MRN 2 5> 8? Figure 3-7: Scatterplots of channel geomorphic units vs. relative proportion of reach area for (A) class A2 channels, (B) class B2a channels, (C) class B3a channels, and (D) class B3b channels. 58 Channel classes A2 and B3a differed significantly with respect to mean area in bedrock cascades, boulder cascades, rapids and riffles (table 3-6-B). The mean area in bedrock and boulder cascades in class B3a channels is high but low in riffles and intermediate in rapids, while the mean area in bedrock and boulder cascades in class Al channels is low but high in riffles and rapids (figure 3-7, and table 3-5). In general, high gradient channel geomorphic units are dominant in high gradient channels with larger bed elements, and low gradient channel units are dominant in lower gradient channels with smaller bed elements. Similar results may be expected when class A2 channels and class B3b channels are compared because the major distinctions between these classes are bed material and gradient also, as well as entrenchment. However, results showed that when channels were separated into entrenched and non-entrenched categories, they were not significantly different (table 3-6-C). In the same way, B3a and B3b channels, two classes distinguished only by their entrenchment designation, offered non-significant results when compared (table 3-6-F). Therefore, it was assumed that little of the variation between A2 and B3a channels could be explained by entrenchment. A2 and B3b channel classes differed significantly with respect to mean area in boulder cascades and riffles (table 3-6-C). Although bedrock cascades are steep, non-alluvial units, and are absent from A2 channels altogether, bedrock cascades were not significant whether or not the M R N 2 outlier was included in the data set. In B3b channels, bedrock cascade data is highly variable, ranging from 0% to over 60% of reach area (figure 3-7). A larger sample size would be required to detect significant differences between two groups that are variable within themselves. 59 The data showed that channel classes B2a and B3a differ significantly with respect to mean area in rapids (TC 43 outlier excluded), riffles, and pools (table 3-6-D). Similarly, the differences in mean area in rapids (TC 43 outlier excluded) and riffles was statistically significant when B2a and B3b channels were compared (table 3-6-E). These differences may be explained by gradient since this is the primary distinction between the classes in question. The assumption regarding entrenchment was discussed earlier and applies here. Rapids are an intermediate channel unit occupying lower gradient portions of high gradient streams and the higher gradient portions of lower gradient streams. However, when the outlier was eliminated, rapid area occupied 0% of class B2a channels. Although class area in rapids is not great in B3a and B3b channels (16.0% and 18.9%, respectively) it is significantly greater than 0%. Previous results confirmed that riffles cover a significantly greater area in lower gradient classes than in higher gradient classes. Pool area is significantly higher in B2a channels than in B3a channels. Gradient differences are the most logical explanation; however, previous results showed that pools were not significantly different between gradient classes. For this particular data set, pool area is precise in B2a channels and evenly distributed over a small range in B3a channels (figure 3-7); therefore, the variance within groups is sufficiently low and the group means are sufficiently different. 3.4 Discussion 3.4.1 Geomorphic Units A broad range of bed slope (figure 3-4), velocity (figure 3-5) and substratum characteristics associated with a single category of geomorphic units implies substantial variability within one type. For instance, consider one unit within the range of rapids but at the lower limit, and another at the higher limit. The former, while still a rapid, may resemble a riffle while the latter may resemble a boulder cascade which suggests that boundaries separating geomorphic units often are parts of a continuum rather than distinct borders. Geomorphic units within these "gray areas", therefore, often are subject to misnomers, a source of error which undoubtedly influenced the results to some degree. Headward, non-alluvial channels may have much poorer morphological unit distinctiveness compared to downstream, properly alluvial reaches because the stream flow is less able to impose characteristic alluvial organisation along the channel. Riffles, sites with the most abundant gravel accumulation, may be the least exceptional (compared to the alluvial sites). Although bed slope ranges of channel unit types overlap (figure 3-4), they have distinct means (and medians) (table 3-2). The average bed slopes of morphological units in the study reaches are much steeper than those with otherwise similar descriptions reported in Bisson et al (1981), Grant et al (1990), Wood-Smith and Buffmgton (1996), and Montgomery and Buffington (1997). Gradient ranges of units described in these works are summarized in table 3-8. The streams studied by the aforementioned researchers had lower gradients and/or greater magnitudes; therefore, owing to scale differences, channel unit descriptions in the current study are similar but appearances may be somewhat dissimilar. Geomorphic units with bed slopes steeper than their counterparts in larger streams, which are generally located further downstream, are probably typical owing to the higher gradients of small, headwater channels in general. 61 Table 3-8: Gradient ranges of geomorphic units from previous studies. geomorphic unit pool glide riffle rapid (step-pool) cascade steps Bisson et al (1981) ? ? <4% >4% N / A N / A Grant et al (1990) Wood-Smith & Buffington (1996) Montgomery & Buffington (1997)1 0 - 1 % ? see riffles N / A < 2% N / A > 1 % - 2% 2% - 4% 1 % - 3% > 2% - 3% N / A 4% 5%-7% >4% >10% > 10% N / A N / A 1 gradients were measured at the reach scale as opposed to the channel unit scale. Clearly, there is a correlation between slope and channel geomorphic units which rank, from highest to lowest gradient, as follows: fall, bedrock cascade, boulder cascade, rapid, bedrock chute, riffle, glide, and pool. Hawkins et al (1993) similarly ranked the gradients of channel units as follows: fall, cascade, chute, rapid, riffle, run. [Runs are often considered synonymous with glides (Bisson, 1981; Church, 1992)]. Montgomery and Buffington (1997) found that distinct reach types in mountain drainage basins occupied particular positions through the channel network from headwaters to mouth. Bedrock reaches occur at locally steep locations throughout the channel networks (Howard, 1987; Montgomery and Buffington, 1997) and have greater slopes than either alluvial (cascade, step-pool, and pool-riffle) or colluvial (sensu Montgomery and Buffington, 1997) reaches for a given drainage area. Resident bed material is determined by slope as well as by upstream and adjacent hillslope sources (Knighton, 1987); therefore, substrate is a diagnostic variable. Montgomery and Buffington (1997) suggest that relative roughness (D/d) and bed slope together differentiate alluvial reach types, thus providing a reasonable stratification of channel morphology. In this study, channel units were described on the basis of dominant bed material and its organisation as well as bed slope. Because gradient ranges within a unit were large, substrate characteristics were very important discriminators. Keller and Melhorn (1978) found that channel units in bedrock streams were more difficult to identify than those in alluvial channels, perhaps owing to substrate homogeneity. Dominant substrate and bed slope are determined with relative ease and their measurements require little equipment. Because classifiers exert tremendous energy simply accessing these very steep and densely vegetated channels, ease of implementation is an important practical consideration; therefore, the aforementioned variables seem to be reasonable criteria according to which channel units may be classified. However, it has been suggested that these criteria are of limited value when identifying pools and riffles because their discriminating power varies with discharge and/or with previous discharge regimes (O'Neill and Abrahams, 1984). While this may be the case in gravel-bed channels, it is unlikely that dominant substrate characteristics and bed slope change appreciably, i f at all, from year to year in semi- and non-alluvial channels. Classification (CSP, 1995) of the hydroriparian system may be accomplished through air photo interpretation; only ground-truthing requires field visits. Field inspection, however, is imperative in order to characterize channel morphology at the geomorphic unit level of resolution. Flow independent criteria are important in these headward channels (Richards, 1976b; Wood-Smith and Buffington, 1996) since many are seasonal or ephemeral and often have little or no surface flow to aid in channel unit identification. Typically, geomorphic units are identified during low flow conditions because channel units are manifest during low flow, although 63 they are formed during the largest flows (Grant et al, 1990). The proportion of stream area occupied by different habitat types changes with flow; therefore, bias is introduced if geomorphic units are identified while the river stage fluctuates. For example, Hogan and Church (1989) found that two similar streams comprised of pool-riffle sequences at low flow became more riffle-like as discharge increased. A study by Roper and Scarnecchia (1995) confirms that discharge may affect the consistency with which individuals classify habitat types. They found that trained observers tended to classify more pools as riffles because of the depth and velocity equalization concurrent with increased discharge. Results from the same study suggest that trained observers were less successful in consistently classifying riffles where the reach gradient was greater than 4%. Presumably, this is because 4% approaches the upper gradient limit of riffles, beyond which rapids tend to form (Bisson et al, 1981; Grant et al, 1990; Wood-Smith and Buffington, 1996; Montgomery and Buffington, 1997). Grant et al (1990) classified channel units according to visual estimates of relative roughness (D/d), degree of step development, and percentage of low flow area in the supercritical regime. Bisson et al (1981) distinguished channel units according to location within the channel, pattern of water flow, and nature of controlling structures. Wood-Smith and Buffington (1996) classified channel units on the basis of bed elevation relative to that of the active channel margin, bed gradient, morphology and hydraulic characteristics. A taxonomy of morphological features is required to abate inconsistent nomenclature associated with the wide range of channel units in steep streams (Grant et al, 1990) and to provide a standard frame of reference that facilitates communication among researchers and managers (Hawkins 64 et al, 1993). Standardized nomenclature and accurate descriptions of structurally and functionally distinct morphological features is required i f successful classification of channel units is to be accomplished; therefore, universal classification criteria must be adopted. But inconsistent nomenclature suggests that no single classification system is universal in its design and application. This may be due, in part, to the variability in aquatic systems and the landscape in which they are situated, different classification objectives, and practical issues such as cost, ease of implementation, and background of classifiers. 3.4.2 Channel Classes Within the size range of the study reaches, width did not effectively discriminate groups of reaches on the basis of channel units, other than pools, despite the fact that study reach widths fall both above and below the 3 m width division. In light of the river continuum concept (Vannote et al, 1980), it was assumed that the biological processes among streams were comparable because all study reaches are highly shaded. The degree to which riparian vegetation wil l provide shade to a stream is a function of the height and cover characteristics of the canopy, the influence of topographic shading, and the width and solar aspect of the stream segment (Oikos and Johnson, 1996). When the shortest riparian canopy in the study (18 m) is combined with the widest study reach (4.9 m), the resulting shading class is high; therefore, all other combinations also must result in high shading classes. Omission of the width criterion, therefore, was justified on morphological and ecological grounds. Channel width also may influence the degree to which the hillslope contributes colluvium directly to the channel. Valley flats may trap hillslope material and prevent it from entering the channel (Whiting and Bradley, 1993). Since the study reaches lack floodplains, the potential for direct input from the neighbouring terrestrial environment is similar among all streams, all things being equal. Notwithstanding, the likelihood of lateral inputs is greater for entrenched channels than for non-entrenched channels. Several significance tests failed to show that groups of entrenched and non-entrenched channels could be distinguished on the basis of channel unit area suggesting that entrenchment does not influence reach organisation in channels of this nature. However, entrenchment may influence the morphology of large as well as disturbed channels to a greater degree. In some large rivers, slope adjustments are attained by meandering habits, however, valley sidewalls exert some control over lateral migration (Dunne and Leopold, 1978; Church, 1992) so entrenchment may be an important influence on channel morphology. In small channels, slope adjustments are attained by stepped longitudinal profiles (Heede, 1981) rather than meandering, therefore, entrenchment is less important. Surface materials on the sidewalls of entrenched channels can be unstable and erodible (Province of B C , 1995). A vegetative cover not only protects the soil surface directly, but also binds the soil (Thorne, 1990), maintains a permeable surface and reduces surface runoff, thus minimizing erosion (Muller and Oberlander, 1984). Removal of the forest from channel sidewalls may increase erosion and sediment washing into the channel which, in turn, may influence its organisation. Studies have indicated that the gradients of entrenched channels are generally higher than those of non-entrenched but otherwise comparable channels. For example, Harris (1988) found that V-shaped valleys have significantly steeper average gradients than other valley 66 types in the eastern Sierra Nevada, California. Sullivan et al (1987) reported that channels constrained between valley walls have more high-gradient units (i.e. boulder cascades). The Province of B.C. (1995) also acknowledges that gullies have very steep gradients. My data indicate that the two steepest reaches of the study are entrenched; however, the average gradients of the remaining two entrenched reaches are moderate in comparison to others. Results of a t-test showed that there is no difference between gradients of entrenched and non-entrenched channels (p < .31). When proportions of channel unit area were investigated, the revised semi-alluvial group was less distinct than the original alluvial category compared to non-alluvial channels. The alluvial group was represented by three channels which, after reclassification, occupied the lower range of semi-alluvial. The typical grain size of the substrate in alluvial channels was smaller than the semi-alluvial average; therefore, fluvial sorting is apt to be more frequent if the slopes of alluvial channels are not considerably lower. However, fluvial sorting in channels originally classified as alluvial remains infrequent so, according to the definition of alluvial (p. 17), such a designation is not justifiable. A semi-alluvial distinction created a single, somewhat more variable group of channels with greater grain size and fluvial transport ranges. In general, boulder cascades were considered non-alluvial; therefore, their dominance in non-alluvial channels is reasonable. Since riffles were considered semi-alluvial, their greater abundance in semi-alluvial channels is understandable. In that case, one would expect that semi-alluvial rapids and non-alluvial bedrock cascades and chutes would also be different amongst generally semi- and non-alluvial reaches. The 20% gradient division was most successful at separating channels on the basis of both high and low gradient channel units. The 25% gradient division successfully separated channels on the basis of low gradient channel units. This may be attributed to the large gradient ranges in channel units. In almost every case 15%, 20%, and 25% fall within the gradient range. Notably, gradient ranges of intermediate to low gradient channel units (rapids, chutes, riffles, and pools) are much more narrow than gradient ranges of high gradient channel units (bedrock and boulder cascades). This suggests that cascades are more apt to be located in reaches with gradients lower than the mean bed slope of cascades whereas rapids, chutes, and riffles are not as commonly located in higher gradient reaches. The lack of variation in channel classes with respect to high gradient units may be attributed to this incident. Rapids and riffles most successfully discriminated channel classes. Four comparisons, all of which involved class A2 and B2a channels, yielded statistically significant results for both rapids and riffles. Boulder cascades were moderately successful at distinguishing channel classes; two comparisons yielded statistically significant results (table 3-6). As predicted, in general, rapids and riffles were dominant in lower gradient, semi-alluvial channels whereas boulder cascades were most abundant in higher gradient, non-alluvial channels or channels at the upper limits of the semi-alluvial range. This result is apt to be typical of most hillside channels in the region. Bedrock cascades and chutes did not discriminate channel classes well. The sample sizes of these morphological units are smaller and more variable within a class (table 3-4, figures 3-6, 68 and 3-7) than any other unit which, together, confound trends. Likewise, pools and falls were not effective discriminators of channel classes. Intuitively, one might expect pool and fall area to be greater in relatively low and high gradient channels, respectively; however, this was not the case. Pool formation may result from acceleration and convergence of flow in channel units immediately upstream (Grant et al, 1990). For example, immediately downstream of falls are zones of intense energy dissipation owing to scour and turbulence which often result in pool formation (Beschta and Platts, 1986; Heede, 1981). Rapids and cascades also tend to cause flowlines to converge downstream; thus high-velocity flow is concentrated and, where focused toward the channel bed or resistant channel margins, promotes scour (Grant et al, 1990). Alternately, pools often terminate in cascades or steps which suggests that large boulders or logs act as weirs creating backwater pools. The size, frequency, distribution, and quality of pools in a stream depend upon the mechanisms of formation and other characteristics such as size of channel substrates, erodibilty of banks, size of obstruction and depth of flow (Beschta and Platts, 1986). So perhaps the key to pool formation is the channel's potential to scour which is partially dependent upon falls and other high gradient units which, in turn, are dependent on the valley gradient. In general, quantification of forest channel morphology is challenging owing to large variability within and between forest streams (Wood-Smith and Buffington, 1996). Sample sizes, therefore, must be large to confirm either the null or alternate hypotheses. Comparisons of stream classes in pristine forest ecosystems, require sufficient replication so that within class variability may be distinguished from between class variability. Each class was represented by a minimum of three and a maximum of seven streams; however, because 69 within class variability is high, distinctions between classes may become more clear with additional representatives. ( Chapter 4: Variation in Benthic Macroinvertebrate Abundance and Community Structure Among Habitats and Streams 70 4.1 Introduction The premise of this chapter is that the physical habitat, together with the pool of species available for colonization determine the structure, function, and other aspects of the organization and development of aquatic communities (Frissell et al, 1986; Barmuta, 1989; Dunson and Travis, 1991; Death, 1995; Grubaugh et al, 1996; Crowl et al, 1997). From the perspective of organisms, habitat is a mosaic of spatially nested microhabitats, each offering different environmental conditions (Minshall, 1984; Angradi, 1996). "...The number of species able to live in any one habitat is proportional to the number of potential microhabitats in the environment" (Mackay and Kalff, 1969). Microhabitat, or patches, may be determined by many interacting factors, including substratum conditions, topography, current patterns, organisms, and disturbance (Pringle et al, 1988). Different patches are usually in close enough proximity to one another that mobile stream organisms can select the environment that provides the most suitable conditions (Dunson and Travis, 1991; Hawkins et al, 1993). The variation among habitats, however, is generally greater than the variation among patches. Studies have demonstrated that species exhibit habitat "preference" (Cummins and Lauff, 1969; Minshall, 1984; Barmuta, 1989). Substrate characteristics (i.e. size, stability, and heterogeneity) and current velocity are two important aspects of habitat that influence invertebrate communities (Minshall and Minshall, 1977; Huryn and Wallace, 1987; Wohl et al, 1995; Angradi, 1996). Controlled experiments have led to the conclusion that "large" substrate particles are more productive than "small" particles (Minshall, 1984); however, distribution patterns in natural streams are not so easily generalized (see Pennak and Van Gerpen, 1947; Cummins and Lauff, 1969; Mackay and Kalff, 1969; Ward, 1975; Benke, 1984; Huryn and Wallace, 1987; Wohl et al, 1995; Grubaugh et al, 1996; Angradi, 1996). Substrate stability denotes the degree of its resistance to movement (Minshall, 1984). Unstable environments host mobile species with good colonizing skills while stable environments attract more sedentary species. It is the intermediate sites that support a larger, more diverse fauna (Connell, 1978; Ward and Stanford, 1983; Scarsbrook and Townsend, 1993). A heterogeneous mixture of bed particles provides a variety of microhabitats and, therefore, encourages a more diverse, but not necessarily more abundant, fauna. Heterogeneity of particle size is greatest in streams dominated by gravel/cobble mixtures (Minshall, 1984). Current velocity exerts a strong influence on the distribution of stream-dwelling organisms because of their inherent reliance on flowing water to deliver food, to meet respiratory requirements, and to escape unfavourable conditions (Hynes, 1970; Minshall, 1984). In erosional habitats, the dominant taxa require fast, turbulent water for food harvesting or physiological requirements and, consequently, possess special morphological and behavioural adaptations for attachment (hooks and silk and sticky secretions), clinging (long, curved tarsal claws, dorso-ventral flattening of the body, streamlining, and ballast), or current avoidance (site selection, burrowing) (Hynes, 1970; Merritt and Cummins, 1978; Barmuta, 1989; Hauer and Resh, 1996). In depositional habitats, fine sediments are deposited and resident organisms may be well adapted for burrowing and poorly adapted to clinging or 72 inhabiting interstitial spaces (Barmuta, 1989; Hauer and Resh, 1996). Water velocity decreases near channel boundaries (Dunne and Leopold, 1978; Ward, 1992). Small and dorso-ventrally compressed organisms can take advantage of extremely low current velocity in the near-boundary sub-layer (Hynes, 1970). Also, dead water zones occur downstream of roughness elements, within clumps of moss and detrital aggregates, in interstitial spaces, and in the hyporheic zone (Ward, 1992). Water temperature and chemistry, flow regime, and shade (Hynes, 1970; Minshall and Minshall, 1977; Minshall, 1984; Death, 1995) are additional factors that affect invertebrate occurrence and distribution but they work on broader scales than the former variables. Water temperature, within a physiographic region, is influenced by its source (i.e. groundwater, storm runoff, meltwater), elevation (Carlson et al, 1990), and proximity to ocean. Also, channels may be subject to varying degrees of solar radiation inputs owing to differences in exposure, topographic shading, and canopy characteristics, thus affecting both shade and temperature. Temperature influences specific life-history parameters such as egg maturation, fecundity, dormancy, larval growth and maturation, voltinism, and emergence timing (Waters, 1979; Sweeney, 1984; Ward, 1992). Differences among invertebrate communities may not be attributed to spatial variation alone but to the appearance/disappearance of species as one after the other completes its development. Shade is correlated with the occurrence and/or abundance of certain species, however, it also is correlated with primary production, temperature, and leaf litter inputs (Carlson et al, 1990). Therefore, it is difficult to ascertain whether the effects of shade are direct or coincidental (Hynes, 1970). Many aquatic insect species are negatively phototactic, seeking cover under and between rocks and other objects 73 during the day (Ward, 1992). It is in this regard that microhabitat characteristics, particularly substrate, influence light levels. The spatial and temporal patterns of solute transport and transfer are a function of phenomena and processes occurring at the watershed scale such as lithology, terrestrial biogenic processes, and precipitation (Harper, 1981; Webster and Ehrman, 1996). The chemical composition of water, therefore, varies from region to region. Calcium and magnesium are the dominant cations in most running waters which are described as the "bicarbonate" type. Lotic systems with different major ions are usually limited to arid regions (Hynes, 1970). Precipitation, the primary source of river water, delivers ions derived from sea spray, air pollution, and dust to surface waters (Hynes, 1970; Kitano, 1975; Webster and Ehrman, 1996). Water from a spring may be de-oxygenated and iron-depositing and/or mineral-rich owing to the dissolution of the rock in which it is contained (Hynes, 1970; Ward, 1992). When water flows along its course it is subject to ionic exchange through dissolution of soil and rock matter and adsorption and fixation by soil materials (Kitano, 1975; Webster and Ehrman, 1996). Flow regime of a channel may be a primary determinant of the community structure and life-cycle patterns of its inhabitants (Ward, 1992). Precipitation volume and frequency and soil type, aggregation, and ability to be infiltrated regulate flow regime, for they determine the proportions of overland and subsurface flow (Williams and Hynes, 1977; Williams, 1987). Drought is a very common natural disturbance (Resh et al, 1988) which represents a significant obstacle for resident organisms; therefore, animals occupying habitats with 74 temporary flow must be properly equipped to survive such conditions (Williams and Hynes, 1977). Cyclic seasonal drying represents a predictable, long-term feature of the system that may impose constraints of an evolutionary nature on the biota (Resh et al, 1988; Brussock and Brown, 1991). Flow frequencies of many ephemeral streams which experience seemingly erratic flow occurrences also may be predictable over a period of years (Williams, 1987). Flow-dependent species obviously cannot survive metabolically-active life history stages during the dry season; therefore they have adapted as specialists (Brown and Brussock, 1991) by adjusting their life cycle timing (egg deposition, hatching, rate of development, and adult emergence), by producing drought resistant eggs, by developing colonization abilities, and by tolerating hyporheic habitats (Pennak and Van Gerpen, 1947; Delucchi, 1988; Resh et al, 1988). These adaptations allow "typical" stream fauna to inhabit less ideal habitats thereby reducing the differences in community structure between temporary and permanent streams (Delucchi, 1988). A high proportion of the reaches in this study are seasonal or ephemeral (table 2-1); therefore, judging by previous work, flow regime is apt to be a very important factor influencing the organization of the macroinvertebrate community. The occurrence and detailed distribution of invertebrates in lotic systems may be a result of a number of interrelated abiotic factors (Hynes, 1970; Minshall, 1984; Ormerod, 1988; Ward, 1992). Biotic processes such as competition and predation also influence community structure; however, they are beyond the scope of this thesis. Because substrate characteristics and current velocity are intimately connected, as well as other variables important to animal distribution such as food retention (Mackay and Kalff, 1969; Williams, 1980), it is difficult to separate the influences of various factors (Minshall, 1984). Controlled experiments are 75 required to unequivocally separate these effects and to assess which processes are ultimately responsible (Ormerod, 1988). For instance, Minshall and Minshall (1977) aimed to describe the microdistribution of invertebrates in response to the substratum and to control or otherwise account for current velocity and food resources. It is convenient, however, to isolate influential factors to clarify discussions. The purpose of this chapter is to examine the influence of six habitat types on benthic macroinvertebrate abundance and community structure in thirteen small streams in the Clayoquot Sound region. Habitat types are synonymous with channel units described in Chapter Three. By considering the biology of the dominant taxa, within the context of their environment, the resultant conditions of different habitats may become more apparent than through physical characterisation alone. 4.1.1 Invertebrate Biology The remainder of this chapter is an attempt to interpret inter-habitat variation in macroinvertebrate abundance and community structure in light of the physical habitat, which is the templet for morphological and behavioural adaptations. At the outset, it is important to establish the value of anecdotal accounts of habitat preference at the familial and higher taxonomic levels of resolution. Within a single genus, the range of habitat preference is wide; therefore, the larger the taxonomic group, the more difficult it is to generalize the habitat (Edmunds et al, 1976; Minshall and Minshall, 1977). Assigning individuals to discrete groups, such as families, can result in an oversimplification which may lead to inaccurate characterisation of community trophic relationships and bioenergetics (Milhuc, 76 1997). Species identifications, therefore, provide a more detailed and accurate assessment of physical or biotic influence not evident in family abundance (Minshall, 1968; Hynes, 1970; Minshall and Minshall, 1977; Williams, 1980; Wohl, et al, 1995). On the other hand, the familial level of taxonomic identification may be a more realistic endeavor. Difficulty in identifying some groups to the species level, for example, Order Diptera and Class Oligochaeta, may cause the number of identified species to be lower than that of higher taxonomic levels (Bournaud et al, 1996). Distributional patterns of species are a function of dispersal mechanisms, environmental tolerances and historical factors; therefore, species are not likely to be pandemic in a broad geographical area. A less precise taxonomic designation, therefore, may be adequate in large study areas (Corkum, 1989). Bournaud et al (1996) studied the macroinvertebrate communities in the Rhone River at both the species and family levels to establish congruity of patterns. Auspiciously, they concluded that the two levels of taxonomic identification provided nearly the same information almost everywhere. Similarly, Zamora-Munoz and Alba-Tercedor (1996) concluded that both family- and species-level identification of macroinvertebrates provides similar information about water quality. Descriptions of the ten commonest taxa, which are the focus of subsequent analyses, follow. These accounts are meant to provide the reader with background information regarding invertebrate habits and habitat preferences. Interpretations of the results are based on both characteristics of the physical habitat and invertebrate biology; therefore, it is important that they precede the results. 77 Order Ephemeroptera (mayflies) Mayfly larvae reside in all types of oxygen rich freshwater environments (Pennak, 1978) and have adapted to a wide range of current speeds and exposures (Pennak and Van Gerpen, 1947). In a northern Colorado stream, Ephemeroptera larvae were found in large numbers on gravel, cobble, and bedrock; however, substrate preference may vary among families and also among species (Pennak and Van Gerpen, 1947; Ward, 1975). Although very few life histories are completely known (Edmunds et al, 1976; Pennak, 1978), in temperate areas, univoltism is the most common life-cycle strategy among mayflies (Hynes, 1970; Edmunds et al, 1976). Life cycle details for a particular species varies according to temperature which is influenced largely by latitude and altitude (Edmunds et al, 1976; Pennak, 1978). Long emergence periods may be common in the Clayoquot region where the winters are relatively mild (Merritt and Cummins, 1996). Three Ephemeroptera families were among the ten most abundant taxa in the study, Heptageniidae, Leptophlebiidae, and Baetidae. Heptageniidae (flatheaded mayflies) The heptageniids are a diverse and abundant family (15 and 132 North American genera and species, respectively) (Edmunds and Allen, 1987). Their primary feeding mechanisms are characterized by gathering fine particulate organic matter (FPOM) from the water, deposits, or surface films (collectors-gatherers) and by scraping periphyton from organic or mineral surfaces (scrapers) (Merritt and Cummins, 1996). Larvae are distinctly flattened and body lengths range from 5 to 20 mm (Edmunds et al, 1976). Morphological adaptations such as long, curved tarsal claws, ventral flattening, and enlarged ventral gills oriented forwards (Merritt and Cummins, 1996), thus increasing the area of marginal contact and preventing 78 water from flowing underneath the body (Hynes, 1970), allow animals to cling tightly to the submerged substrate of lotic-erosional habitats (dingers) (Merritt and Cummins, 1996). They occur on rocks, wood, vegetation, or in leaf packs (Edmunds and Allen, 1987) which indicates that habitat distributions are not consistent within the family (Ormerod, 1988). Many heptageniids are microhabitat specialists that prefer to live between and beneath stones and to avoid exposed surfaces in fast water (Hynes, 1970; Edmunds et al, 1976; and Pennak, 1978). Heptageniids primarily are characterised by univoltinism; however, some species produce two or more generations per year (Merritt and Cummins, 1996). Adult emergence ranges from April to October (Edmunds et al, 1976). Leptophlebiidae (prong-gilled mayflies) There are approximately 9 genera and 72 North American leptophlebiid species, most of which can be categorized into the collectors-gatherers or scrapers functional groups (Edmunds and Allen, 1987; Merritt and Cummins, 1996). Leptophlebiid mature larval bodies typically are depressed and range from 4 to 15 mm in length (Edmunds and Allen, 1987). Most commonly, their modes of existence, or habits, include clinging, occupying substrate surfaces such as fine sediment or floating leaves (sprawling), and propelling themselves through the current with quick, darting motion (swimming) (Pennak, 1978; Merritt and Cummins, 1996). Larvae inhabit both fast and slow water lotic environments and occupy a variety of substrates including rocks, leaf packs, and woody debris. Some leptophlebiid species avoid direct exposure to strong currents by moving away or by burrowing into the substrate (Hynes, 1970; Edmunds and Allen, 1987). Adult emergence may begin in May and continue to as late as November (Edmunds et al, 1976). 79 Baetidae (small minnow mayflies) Among the most common and universal of stream animals are the baetids (18 and 121 North American genera and species, respectively) (Merritt and Cummins, 1996) which belong to the collectors-gatherers and scrapers functional groups, like the two previous families (Hynes, 1970; Edmunds et al, 1976; Edmunds and Allen, 1987; Merritt and Cummins, 1996). Baetids are dingers and swimmers. Their fusiform body (mature larvae body length: 3 to 12 mm) offers least resistance to fluids and their long cerci act as vanes to turn their heads into the current (Hynes, 1970). The Baetidae have been associated with bedrock and swift currents (Pennak and Van Gerpen, 1947; Minshall and Minshall, 1977). Life histories of the baetids are not well documented; however, like most mayflies of temperate regions, each year one generation emerges, probably during the summer months (Edmunds et al, 1976). Several Baetis species, however, may emerge throughout the year (Merritt and Cummins, 1996; Abell, 1984). Order Diptera (flies, mosquitoes, and midges) Order Diptera is one of the largest in the Class Insecta; however, few families have aquatic larval stages (Pennak, 1978). Nevertheless, Diptera contain more aquatic species than are found among any of the completely aquatic orders such as Ephemeroptera, Plecoptera, and Trichoptera. The families Simuliidae and Chironomidae are among the few strictly confined to aquatic environments (Merritt and Cummins, 1996). "Aquatic Diptera larvae show greater variability in structure and habitat than any other order of aquatic insects" (Pennak, 1978); therefore, Diptera are renowned for causing identification difficulties (Bournaud et al, 1996). Family Simuliidae (black/lies) The family Simuliidae is one of the most homogeneous and easily recognized families in the order Diptera. There are approximately 143 North American species (Merritt and Cummins, 1996). The larvae (body length: 3 to 15 mm) (Pennak, 1978, Merritt and Cummins, 1996) collect fine particulate organic matter (FPOM) by filtering it from flowing water with their cephalic (labral) fans (collector-filterers) (Merritt and Cummins, 1978). Simuliids, which are among the most adapted of all animals to life in running water, inhabit lotic-erosional environments and often occur in very fast currents (Pennak and Van Gerpen, 1947; Hynes, 1970; Minshall; and Minshall, 1977; Pennak, 1978; Merritt and Cummins, 1996). Simuliids produce tangled mats of silk on substratum to which they attach themselves by a circlet of numerous radiating rows of minute hooks on their posterior ends (Hynes, 1970; Pennak, 1978; Merritt and Cummins, 1996). Consequently, their cylindrical bodies trail out into the current. Members of the family Simuliidae are often associated with bedrock (Pennak and Van Gerpen, 1947; Ward, 1975). There may be as many as six generations per year in temperate regions, depending on the species and the habitat (Merritt and Cummins, 1996). Family Chironomidae (midges) The family Chironomidae is markedly complex (Pennak, 1978) and is probably the dominant family of aquatic Diptera (Clifford, 1991). "The number of chironomid species present in most systems often accounts for at least 50% of the total macroinvertebrate species diversity (richness component)" (Merritt and Cummins, 1996). The number of North American species in this family has been estimated at 530 to 1172. The range is wide because most groups are unfamiliar; therefore, the number of described species bears little resemblance to the actual 81 numbers per genus. Most species belong to one of two functional groups: (1) collectors -gatherers and filterers and (2) predators - engulfers and piercers. Engulfers attack prey and ingest other animals in whole of in part. Piercers also are carnivorous predators that attack prey and pierce tissues and cells from which fluids are sucked (Merritt and Cummins, 1996). Members of the family Chironomidae inhabit essentially all types of aquatic habitats (Clifford, 1991) and their modes of existence primarily include sprawling and burrowing into sediments or plant tissue (Merritt and Cummins, 1996). Observational studies have shown that various lotic species favour fine sediment (i.e. sand) and detritus, into which they burrow, and the protected surfaces and small crevices of course gravel and cobbles (Pennak and Van Gerpen, 1947; Mackay and Kalff, 1969; Minshall and Minshall, 1977). Chironomid larvae range from 2 to 30 mm in body length and are elongated, cylindrical and slender in form (Pennak, 1978). The life histories are difficult to generalize but most chironomids may have one to several generations annually, and some may require two years to complete a life cycle (Hynes, 1970; Pennak, 1978; Clifford, 1991). In any event, Mackay and Kalff (1969) concluded that Chironomidae were an important component of the invertebrate community in a small Quebec stream during all seasons. Order Plecoptera (stoneflies) Plecoptera larvae are typically associated with unpolluted lotic waters where there is an abundance of oxygen (Pennak, 1978; Baumann, 1987). As their name suggests, stonefly larvae are major contributors to benthic communities associated with cobbles (Mackay and Kalff, 1969). Since stonefly larvae are sluggish (Pennak, 1978), they favour undersides and interstices of gravel, cobble, and even boulder substrates (Pennak and Van Gerpen, 1947; 82 Abell, 1984). Neither exposed bedrock nor shifting sands are suitable for most species in this order (Pennak and Van Gerpen, 1947). Family Nemouridae (brown stone/lies) There are approximately 65 members of the Nemouridae in North America. They belong to the shredders-facultative detritivores and collectors-gatherers functional groups because they chew coarse particulate organic matter (CPOM), their dominant food, and collect decomposing fine particulate organic matter (FPOM) (Merritt and Cummins, 1996). Nemourid habits are generally characterized as sprawling and clinging. Body length of mature larvae ranges from 5 to 20 mm (Baumann, 1987). The cerci of nemourids probably serve a similar function as those of baetids (Hynes, 1970). Plecoptera nymphs occur in detritus, leaf packs, algal mats, and under stones in every kind of unpolluted lotic habitat. Many species, however, are quite specific in their ecological preferences. Little accurate information is available on the life histories of the Plecoptera (Pennak, 1978). Only about 15% of the species have been studied in this respect (Merritt and Cummins, 1978). Most nemourids have one generation a year but some have two or three year life spans (Clifford, 1991; Merritt and Cummins, 1996). A succession of species emerge throughout the year (Merritt and Cummins, 1978; Baumann, 1987), a fairly unique characteristic in the insect world (Pennak, 1978). Family Chloroperlidae (yellow or green stone/lies) In North America there are approximately 13 and 77 chloroperlid genera and species respectivley. Mature chloroperlid larvae are chiefly engulfers (carnivorous predators), 83 feeding on other aquatic insects such as mayfly nymphs and Diptera larvae (Pennak, 1978) but omnivorous when they are young, generally collecting (gathering) and scraping food (Merritt and Cummins, 1996; Baumann, 1987). Mature larvae body length ranges from 10 to 40 mm. Their cerci are short, generally less than 3A of the length of the abdomen (Merritt and Cummins, 1996); therefore probably do not act as vanes in the current. Instead, the Chloroperlidae of the Northern Hemisphere have taken to burrowing down into the substratum to avoid changes in current. Their long and narrow bodies (Hynes, 1970), short legs and cerci enable them move through the substrate in a snakelike manner (Baumann, 1987). One to two years are required for chloroperlids to complete their life cycle. Emergence occurs in the spring through the autumn (Merritt and Cummins, 1978). Order Trichoptera (caddisflies) Family Lepidostomatidae Family Lepidostomatidae has approximately 80 North American species in only two genera. Lepidostoma, the only genus known to occur in the west, accounts for 75 of the species (Merritt and Cummins, 1996). Considerable structural and behavioural diversity exists among the larvae of the Nearctic Lepidostoma spp. Lepidostoma, a shredder-detritivore, (Wiggins, 1987) is adapted for living on vascular hydrophytes or detritus with modifications for moving vertically on stem type surfaces (climbers). Other modes of existence expressed in this genus are sprawling and clinging (Merritt and Cummins, 1996). Mackay and Kalff (1969) found that Lepidostoma spp. contributed to high numbers in leaf packs in late summer in a small Quebec stream. Some may be found in temporary streams (Wiggins, 1987). Most North American caddisfly species, including Lepidostoma spp., have univoltine life cycles 84 (Merritt and Cummins, 1996). Trichoptera larvae tend to appear while older larvae or pupae are still present suggesting that generations overlap (Hynes, 1970). Class Oligochaeta (aquatic earthworms) - Order Haplotaxida Family Enchytraeidae Enchytraeids are mainly terrestrial oligochaetes common in saturated debris along margins of streams and lakes (Pennak, 1978) and some aquatic or semi-aquatic habitats (Brinkhurst, 1986; Clifford, 1991). Body length ranges from 10 to 30 mm (Pennak, 1978). Enchytraeids live in interstitial spaces and may occur in considerable numbers (Hynes, 1970). Class Turbellaria (flatworms) Order Tricladida The Tricladida is the major order of freshwater turbellarians consisting of approximately 43 North American species. Triclads are predators and scavengers of invertebrates (Clifford, 1991). Body length is generally greater than 5 mm (Pennak, 1978). Their flattened forms allow them to conceal themselves under stones in running water (Hynes, 1970). Some flatworms depend upon the near constant conditions within the groundwater for their survival (Wright et al, 1984); therefore, they may burrow down into the substratum of temporary streams during dry periods (Hynes, 1970). 85 4.2 Methods 4.2.1 Data Collection Stratified random benthic samples were extracted, using a kick and sweep method. Within a stream, three representatives from each geomorphic unit type were randomly selected, where three or more of each unit type existed. To illustrate, suppose a reach was composed of eight pools, two riffles, six boulder cascades and four rapids. Because there are only two riffles, both would be selected. Three pools, boulder cascades, and rapids also would be selected. Random selection involved assigning a number to each and every unit (i.e. pool 1, pool 2, rapid 1, rapid 2, riffle 1, riffle 2, etc.) recorded on the scaled diagrams produced during field surveys (figure 3-1). Numbered units then were drawn from a hat. Selected geomorphic units were sampled according to the following protocol. A one meter transect in the centre of the channel was estimated. A dip net was placed on the bed at the downstream end of the transect. The substrate was vigorously disturbed for two minutes and displaced matter was carried downstream, by the current, into the dip net. The net contents were placed into a plastic sampling jar and preserved in a neutralized, 5-10% formaldehyde solution to await processing. In addition, current velocity (refer to section 3.2.1), dominant vegetation, shading, and dominant substrate were measured and/or estimated at each sampled site. Dominant riparian shrubs and trees adjacent to the each sampling site were identified according to the descriptions of Pojar and MacKinnon (1994). Each sampling site was assigned a shading class (L, M , or H) according to the methods described in the WRP Riparian Assessment Field Guide Draft (June 1996). A shading class was determined according to the height of the 86 dominant riparian canopy and the channel width. Topographic shading and large canopy gaps were accounted for by increasing the assigned class by one and reducing the assigned class by one, respectively. Dominant substrate measurements were discussed in section 3.2.1. Five (DB1, DB5, MRN1, TCI , and TQl) of the thirteen streams surveyed in the summer of 1996 had sufficient water to allow kick sampling. When discharge increased, four (DB1, DB5, MRN1, and TQl ) of the original five channels were resampled, in addition to DB3, MTD1, NMT4, and NMT5, in October and November of the same year. The four channels surveyed in spring/summer 1997 (DBf, TClb , TC43, and MRN2) were sampled at this time as well as four previously sampled channels (DB1, DB5, MRN1, and TQl) . In the laboratory, samples were washed through nested 1 mm and 256 jum sieves and the size fractions were separated and stored in 95% ethanol. Invertebrates were removed from the > 1 mm portion, identified and counted under a dissecting microscope. The remaining organic and inorganic material was retained for assessment of sample processing efficiency (i.e. quality control). The 256 jum - 1 mm fractions were retained but not processed further. The Insecta, Decapoda, and Oligochaeta were identified at the familial taxonomic level when possible. Don Stacey (Royal Ontario Museum) and Dr. Art Borkent (Royal British Columbia Museum and American Museum of Natural History) verified Oligochaeta and Diptera identifications, respectively. Scarce invertebrates, such as the Gastropoda and Nematoda were identified at the class taxonomic level. The Turbellaria and Hydracarina were identified at the order taxonomic level. Taxonomic works referenced were Wiggins (1996) for Trichoptera (caddisflies), Merritt and Cummins (1996) for all other insects, Brinkhurst (1986) for the Oligochaeta, Pennak (1989) for the Decapoda and Turbellaria and Thorp and Covich (1991) for the Nematoda and Hydracarina. 4.2.2 Data Analyses Fixed effects, unbalanced three-factor analysis of variance (ANOVA) models (procedure G L M , SAS, 1990) were employed to examine variation among streams, habitats and sampling dates in total macroinvertebrate abundance and abundance of the ten most common taxa. A posteriori comparisons were made using the Bonferroni method, a multiple range test. Macroinvertebrate abundance data were logarithmically transformed [log1 0(x+l)] (Minshall and Minshall, 1977) to correct for heteroscedasticity (Angradi, 1996). Principal Components Analysis (PCA) is a multivariate ordination technique used to reveal any intrinsic pattern in sites on the basis of taxonomic or environmental data. Similar sites appear close together in ordination space while divergent sites occur far apart. Ordination, perhaps, is the most important numerical technique available for seeking patterns in community structure because it detects patterns as arrangements of units on axes (Doledec and Chessel, 1994). Internal structures and major sources of variation within the physical and biological systems were explored through the application of correlation matrix P C A (factor analysis, Statsoft, 1994) (Barkham and Norris, 1970). Taxa comprising < 1% of the total data set were excluded in order to eliminate rare animals (Zumora-Munoz and Alba-Tercedor, 1996). Macroinvertebrate data were normalized by applying a log 1 0 (x+1) transformation (Minshall and Minshall, 1977) while environmental data were not transformed. To relate the physical and the biological systems, their principal components were tested for correlation. In addition, the variables of each system were tested for correlation with the principal components of the opposite system in order to examine relations between biological principal components and the physical variables (Barkham and Norris, 1970). Ordination of macroinvertebrate data and interpretation of these data in relation to environmental parameters is known as indirect gradient analysis (Doledec and Chessel, 1994; Jongman et al, 1995). Fixed effects, unbalanced two-factor A N O V A models were used to investigate variation in factor scores of biological PC 1 and physical PC 1 as a result of stream and habitat effects. Essentially, this helped determine how significantly stream and habitat influenced the ordination of taxonomic and environmental data. A n analysis of covariance (ANCOVA) model was used to investigate the effects of stream and habitat on the dependent variable (biological PC 1 factor scores) while removing the effect of the concomitant variable (physical PC 1 factor scores). A posteriori comparisons were made using the Bonferroni method. 89 4.3 Results 4.3.1 Quality Control Ten percent of the benthic samples were randomly selected for quality control which involved sorting through the material greater than 1 mm in size remaining after the animals were extracted and removing any invertebrates that were overlooked the first time. Percent success rate was calculated for these samples and is reported in table 4-1. Mean success rate was 84% and an estimate of the variation in the population of all samples, represented by the coefficient of variation, was 4%. The unusually low success rate (44%) for DB5 bedrock cascade 6 was attributed to the generally small size of the missed organisms. This outcome probably is not typical of samples collected from bedrock cascades which usually were free of excess organic and inorganic matter. Detritus and moss, in particular, conceal organisms making them difficult to detect. 4.3.2 Macroinvertebrate Abundance In total, 185 benthic samples were collected over the three sampling dates. More than 13,100 specimens representing approximately 55 families, 16 orders and 8 classes were identified (appendix A ) . The ten most abundant families and orders are listed in table 4-2. Some macroinvertebrate species are habitat specialists with very narrow microhabitat requirements, for example, the Simuliidae (Hauer and Resh, 1996). Other species are habitat generalists which tolerate a wide range of environmental conditions and, consequently, their distribution is more ubiquitous, for example, the Chironomidae. Habitat which fulfills the primary requirements of certain organisms are more apt to host distinctive communities whereas those that satisfy secondary or tertiary needs of organisms (i.e. habitat may provide adequate 90 Table 4-1: Insect data quality control summary. habitat unit & sampling invertebrates invertebrates specimen success site date missed (#) found (#) total (#) rate DB 5 bedrock 6 96/10/22 5 4 9 44 M T D 1 boulder 2 96/10/24 4 20 24 83 M R N 1 boulder 3 96/08/19 2 6 8 75 TC 1 pool 1 96/08/21 3 25 28 89 M T D 1 pool 2 96/10/24 0 5 5 100 M R N 1 pool 3 96/10/25 1 4 5 80 TQ 1 pool 3 96/08/20 2 4 6 67 DB 3 pool 4 96/10/22 1 7 8 87 DB f pool 10 97/06/06 16 51 67 76 DB 3 rapid 1 96/10/21 2 5 7 71 DB f rapid 1 97/06/06 23 98 121 81 DB 5 riffle 1 97/06/09 4 29 33 88 N M T 4 riffle 1 96/11/08 0 11 11 100 DB 3 riffle 2 96/10/21 2 37 39 95 N M T 4 riffle 2 96/11/08 2 37 39 95 M R N 1 riffle 3 96/10/25 5 112 117 96 M R N 1 riffle 4 97/06/07 51 183 235 78 TC 43 riffle 4 97/06/09 26 326 352 93 TQ 1 riffle 4 97/06/08 58 481 540 89 Table 4-2: Total number of individuals in the ten most abundant taxa and their proportions of the grand total (streams, habitats, and dates combined). family order total % of total Heptageniidae Ephemeroptera (mayflies) 2819 21.41% Nemouridae Plecoptera (stoneflies) 2321 17.63% Chironomidae (midges) Diptera (flies, mosquitoes, midges) 1490 11.32% Leptophlebiidae Ephemeroptera 822 6.24% Enchytraeidae Haplotaxida (segmented worms) 675 5.13% Baetidae Ephemeroptera 552 4.19% Chloroperlidae Plecoptera 544 4.13% ~ Tricladida (flatworms) 511 3.88% Simuliidae (blackflies) Diptera 405 3.08% Lepidostomatidae Trichoptera (caddisfiies) 350 2.66% combined total 10489 79.67% 91 conditions for the prey of a predator or a secondary predator) may have less distinctive communities (Williams, 1980). Therefore, the ten most common taxa were chosen for further investigation in order to examine both habitat specialists and generalists. Less common taxa could have been included in the A N O V A models, for example, 18 taxa were included in a principal components analysis discussed in section 4.3.3. The intention of this study, however, is not to explore individual taxa but to investigate differences in macroinvertebrate abundance among habitats which can be accomplished without considering each taxon individually. Total abundance, based on the average number of animals per sample calculated from pooled faunal data from 13 streams and 3 sampling dates, was highest on riffles, followed by rapids, pools, boulder cascades, chutes and lowest on bedrock cascades (figure 4-1-A). Total abundance, based on pooled faunal data from 6 habitat types and 3 sampling dates, was greatest in TC 43, followed by TQ1, TCI , MRN2, DB1, MRN1, TClb , DBf, DB5, DB3, NMT5, NMT4, and lowest in MTD1 (figure 4-1-B). Mean animal abundance per sample in TC43 and TQ1 exceeded mean abundance per sample even in riffles, the most prolific habitat type (figure 4-1-A and -B). In addition, the uncharacteristically productive nature of TQ1 and TC43, relative to the other study streams, became apparent in the results of initial analyses. These two streams dominated the results, therefore, were removed from A N O V A models. Furthermore, TCI was eliminated from analyses because very few samples were extracted during August 1996 only which appeared to confuse rather than to clarify patterns. Figure 4-2 depicts inter-habitat variation in total macroinvertebrate abundance for each stream, excluding TCI , TC43 and TQ1. Figure 4-1: Mean invertebrate abundance (+1 SE) expressed as: (A) animals per sample for each habitat type pooled across streams and sampling dates and (B) animals per sample for each study stream pooled across habitat types and sampling dates. 50 20 •a c 3 10 I DBl DB3 MTD1 NMT4 NMT5 V/ HZ TClb boulder cascade bedrock cascade li-ij chute ^ pool rapid I riffle Figure 4-2: Inter-habitat variation in mean (+1 SE) total invertebrate abundance for each stream (excluding TCI, TC43, and TQ1). Data pooled across sampling dates. A n A N O V A procedure showed that total invertebrate abundance was significantly different among habitats (p = .0006) (table 4-3). Based on mean abundance, riffles ranked first followed by rapids, chutes, pools, bedrock, and boulder cascades. Habitat rank order is different compared to that displayed in figure 4-1 which is a consequence of TC43 and T Q l exclusion. Pool, bedrock and boulder cascade abundance were significantly different from riffles while only pools and boulder cascades were different from rapids. The bedrock cascade rank suggests it too should be different from rapids; however, degrees of freedom prevent this outcome from appearing. While the response of organisms to micro-environment conditions varies, these results indicate that the majority prefer gravel/cobble with abundant habitable surface area and interstitial spaces, intermediate stability, and relatively swift, but not extreme, current whereas the minority favor large, highly stable particles that are more exposed to the full force of the current. Although the substrate of chutes primarily is composed of bedrock, their relatively low to moderate gradient allows for stable alluvium which increases refugia, therefore, colonization potential. Differences among neither streams nor sampling dates were significant (table 4-3); however, Bonferroni's multiple range test indicated that total invertebrate abundance in MTD1 was different from six of the other 9 streams and also that August 1996 and June 1997 were different from October 1996. Bonferroni multiple comparison results are reported in table 4-4 for all A N O V A models. 95 Table 4-3: Results of three-factor A N O V A models for effects of stream, habitat, date, and interactions on total invertebrate abundance and on individual taxon abundance for the ten most common taxa in the study. Dependent variable stream F habitat F date F stream x habitat F stream x date F habitatx date F stream x habitatx date F total abundance 1.65 4.86*** 2.31 1.40 1.51 0.54 0.85 Heptageniidae 4.60*** 5.61*** 4.19** 1.90** 1.44 0.39 1.03 Nemouridae 4 g9*** 5.07*** 2.53* 1.28 0.63 0.64 1.95* Chironomidae 1.83* 2.16* 5 45* * * 1.36 2.75** 0.61 1.14 Leptophlebiidae g 27*** 5.28*** 1.61 1.08 2.02* 0.66 1.66 Enchytraeidae 5.82*** 3 41*** 0.26 1 79** 1.76 0.40 0.54 Baetidae 2.55** 1.90 5 71*** 0.60 1.95 0.53 0.68 Chloroperlidae 2 77*** 5.35*** 5.82*** 1.02 0.84 0.94 1.05 0. Tricladida 3.40*** 1.07 0.11 0.95 1.26 0.69 0.39 Simuliidae 2.57** 5.55*** 3.38** 1.10 0.89 1.11 0.75 Lepidostomatidae 2.44** 2.28* 1.16 1.56* 1.75 2.51** 1.22 1. Asterisks indicate p < 0.10 (*), p < 0.05 (**), and p < 0.01 (***). 2. Degrees of freedom for main effects and interactions in all models are as follows: stream = 9, habitat = 5, date = 2, streamxhabitat = 25, streamxdate = 4, habitatxdate = 8, streamxhabitatxdate = 9. Error degrees of freedom for each model = 83. Degrees of freedom for incomplete experimental designs are peculiar by nature owing to cells which lack observations. The results show that the effects of stream (p = .0001), habitat (p = .0002), date (p = .0185), and streamxhabitat interactions (p = .0158) on mean Heptageniidae abundance were statistically significant (table 4-3). In general, abundance was similar amongst perennial and some seasonal streams (excluding MRN1) and amongst ephemeral and other seasonal streams; however, it was different between the two groups (table 4-4). Heptageniidae were most abundant in riffles followed by chutes, rapids, pools, bedrock cascades and least abundant in boulder cascades. These results do not depart appreciably from those of the total abundance model, not surprisingly, because heptageniids account for more than 20% of total abundance (table 4-2). Abundance was highest in June, followed by August and lowest during October. Each date was different from the other signifying that emergence events, 96 Table 4-4: Multiple pairwise comparisons of means amongst streams, habitats, and sampling dates for total invertebrate abundance and individual taxon abundance for the ten most abundant taxa in the study. dependent variable stream comparisons habitat comparisons date comparisons total density Heptageniidae Nemouridae Chironomidae Leptophlebidae Enchytraeidae Baetidae Chloroperlidae O. Tricladida Simulidae Lepidostomatidae MRN1 M T D 1 T C l b M T D I MRN2M T D 1 DBf 1™ 1 DBJM™ DB5MTO1 D B 3 N M T 5 N M T 4 ^fXDlMRN'TClb MRN2 DBf DB1DB5 jyj-^jy2MKN1 NMT5 NMT4 MTD1 DB3 j-jg^MRNl NMT5 NMT4MTD1 DB3 'pQ j^NMTS NMT4 MTD1 DB3 D B j N M T 5 NMT4 MTD1 DB3 J)gfNMT5 NMT4 MTD1 DB3 |^jpj^jMRN2 DB5 l\f]^f^MRN2 D B 5 TClb DB1 DBf J^^ff4'Mm2 D B 5 T Clb DB1 DBf J^ff£)yMRN2 DBS TClb DBI DBf £)JJ3M R N 2 DB5 TClb DBI DBf MRN 1 D B 1 N M T 5 N M T 4 M T D 1 DBf° B 1 N M T 4 M T D 1 M R N 2 N M T 4 M T D 1 p g g N ^ M T D ! j j g j T C l b j - j g j MRNI DBf jVM7'5MRN1 NMT4Mmi DBf MRN2 DB5 M7Z)7MRN1 D B f M R N 2 D B 5 J)gfOB5 MTD1 NMT4 TCHb^™' NMT4 MRN1 M T D 1 NMT4 NMTS DBJ. DB5 D B f DB3 A/7Z)7 D B f T C l b M K N 1 A'M7'4 D B f T C l b M R N 1 M R N l T C l b D B 5 D B f D B 1 MRN2 N M T S NMT4 DB3 MTD1 TClbVmW pjg^MRNl MTD1 J)gfMRNl UB1 M R N 1 MRN2 M l N T M/r5 M R N 1 NMT4mtll~DBiiasi ^y^j^MRN2 DBS TClb MRNI DBI JJJJ^MIUvC DBS TClb MRNI DBI ^y^yjMRN2 DBS TClb MRNI DBI J^JJ'DlMKtiX D B 1 DBf°B1 MRN2N M T 4 D B 3 N M T 5 DB5 N M T 4 D B 3 NMTS j Q j b N M T 4 DB3 NMT5 Jyjpjsf J NMT4 DB3 NMT5 MTD1 jyQ JNMT4 DB3 NMT5 MTD1 DBf j y j j y ^ 2 D B 5 D B 1 MRN' NMT4 MTD I NMTS DB3 j-jg^MRNl NMT4 MTD1 NMT5 DB3 •JQ j-jg^MRN2 MRNI NMT4 MD1 NMT5 DB3 p j g JMRN2 MTD1 DB3 J y J R N l ^ ™ 2 D B f D B 5 NMT4MBN2DB{DB5 A/7Z)/M R N 2 D B f D B 5 D B I NMT5] MRN2 DBf DBS DB3 MRN2 DBf DBS DBI MRN1 N M T 5 M T D 1 N M T 4 D B 1 M T D 1 N M T 4 TClb DB3 DB5 DBf MRN2 NMT5MM] MTDlMmi DB1 NMT4MSN,DB[ MRN1 P B 1 D B f T C l b D B 3 N M T 4 D B 5 M T D 1 NMT5 MRN2 DB1 M R N 1 D B f ™ 1 T C l b M R N 1 DB3 M R N 1 NMT4MKNl DBS*""" MTDl™™1 T C l b D B 1 M R N 1 D B 3 M T D 1 N M T 5 N M T 4 DBf*1™1 D B 3 MTD1 NMT5 N M T 4 J J J Q J j y ^ j JJBjJClb M R N l T C l b D B f DB3 T C l b D B f MTD 7T C l b D B f NMT5TClb DBf NMT4TC,hDEf p g j DBS MTDl NMT4 T C l b MRNI DBf NMT5 DB3 DB5 D B 1 MTDl™ NMT4Dm R i P B c B o R a P B o C P R i R a j o A o 0 A J g e R i g 0 R i Ra Ri p B c B o C B o Ra B o P w J A 0 A J 0 0 J A g e R i B o R i C R a R i B o P B c C R a P B c B o R i J ° A O J pRi Ra B g R i Ra Ri Ra Be C P Bo J ° A ° 0 J A Ri c B c P c B e RaB e Bo A J 0 J A 0 0 A ' C R i p BeRi p R a R a C P B c R i P B e B o C R a 0 J A J ° A A O J pRa Ri B g R a Ri g e R i Ra Bo P C RjBe R a B e jA O A J O Q J A g 0 B c pBe R j P Ra C Bc Bo pRi p^gRi C R i BeRi BoR i Ra Ri Bo C Be P C r a R i B o P B e P R a C P R i c B o c P C B e R a J°A° 0 , A A JO J°A° 0 , A Ri P C Ra Bo Be J A O 1. Streams, habitats and dates are listed in descending series based on arithmetic means of log,0 (x+1) transformed data. 2. Superscripts indicate which pairs are significantly different (p < 0.1) according to Bonferroni's multiple range test. 3. Be=bedrock, Bo=boulder, Ra=rapid, C=chute, Ri=riffle, and P=pool. 4. Stream abbreviations in italic, bold, and underlined letters represent ephemeral, seasonal, and perennial flow regimes, respectively. Font style does not apply to superscripts. 97 severe spates, and/or mortality occurred during periods between sample extraction. A plot of means (figure 4-3) details the significant streamxhabitat interaction which appears to be a consequence of prolific bedrock cascades in D B f and uncharacteristically low productivity in DB1 riffles. Mean gradient of D B f bedrock cascades is half the study average (25% and 50%, respectively); therefore, sediment transport potential is lower and stable alluvium more likely. This, in addition to the notched nature of D B f bedrock, increases invertebrate retreat. Riffle alluvium in DB 1 was not more than a few grains thick owing to the dominantly bedrock bed, thus decreasing particle surface area and interstitial spaces compared to typical riffles. Stream, habitat, and date had significant effects on abundance of the Nemouridae (p = .0001, .0004, and .0856, respectively) (table 4-3). In general, nemourid abundance was greatest in perennial and seasonal channels and lowest in ephemeral channels (table 4-4). However, nemourid abundance in DB1, a perennial stream, was more similar to ephemeral channels than the seasonal/perennial group which also may be a consequence of the dominantly bedrock bed. Based on mean abundance, riffles, chutes, and rapids hosted similar populations of Nemouridae which were different from those of boulder cascades, pools, and bedrock cascades, in general. This suggests that Nemouridae prefer habitats with fairly swift currents but with micro-habitat providing retreat from the full force of the flow. These preferences are similar to those of the heptageniids and the entire community in general; however, habitat ranks do not concur among models. Nemourid density was highest in June, followed by August and October but mean density differed between June and October only, 98 - • - riffle rapid chute -A- pool -+• bedrock - • - boulder DB 1 DB 3 DB 5 D B f MRN 1 MRN 2 MTD 1 NMT4 NMT 5 TC lb Figure 4-3: Plots of means illustrating details of stream x habitat interactions among the Heptageniidae. In the legend, habitat types are listed in descending series based on total invertebrate abundance. Mean abundance scale of this plot, and all other interaction plots, is based on log 1 0(x+l) transformed data. - O - Jun'97 - D - Oct'96 DB 1 DB 3 DB 5 DB f MRN 1 MRN 2 MTD 1 NMT 4 NMT 5 TC lb A u g ' 9 6 Figure 4-5: Plots of means illustrating details of stream x date interactions among the Chironomidae. In the legend, sampling dates are listed in descending series based on total invertebrate abundance. 99 signifying that slow but steady emergence continued throughout the season. The effect of streamxhabitatxdate interactions on mean nemourid density also was significant (p = .0560) because each habitat did not have the same rank in each stream and during each sampling date (table 4-3 and figure 4-4). Stream, habitat and date had significant effects on chironomid abundance (p = .0746, .0665, and .0060, respectively) (table 4-3). The strength of stream and habitat effects are somewhat weaker compared to date which may be a reflection of the ability of chironomids, a widespread group, to reside in all types of aquatic environments. Chironomids occupy ephemeral streams (Abell, 1984); however, abundance appears to be higher in channels with perennial and seasonal flow. Although A N O V A results indicate that chironomid abundance differed significantly among habitats, Bonferroni multiple comparisons suggest the opposite (table 4-4). With respect to mean Chironomidae abundance values, habitats do not rank differently compared to those based on total animal abundance with the exception of bedrock cascades which move up the ranks two positions. This indicates that chironomids are more able to tolerate bedrock substrates than the other organisms, perhaps because their small bodies allow them to inhabit the low velocity near-boundary sub-layer, alleviating difficulties caused by current (Hynes, 1970). Chironomid abundance was highest in June, followed closely by August but significantly lower in October suggesting that adult emergence took place during late summer and/or early autumn or, perhaps, that the Chironomidae vacated their space in response to the autumn spates (table 4-4). Chironomid abundance in DB1 was higher in August than in June; therefore, the streamxdate interaction was significant 1.5 0.5 0 \ a c o A ft i t • i i i ( • <aM«BZZQHH' , QQQQggi -SSy § § S z z H riffle • s l S z g H rapid CP A • HP • A S2SgzH chute pool cQfflfflSzzP*-"1-^  bedrock — c*i in ^  ~ — i n x> «MMSzzaf - 'H ;~ QQQaggi-ssB SSSzzH boulder -O- June'97 ••- October'96 - A - August'96 Figure 4-4: Plot of means illustrating details of stream x habitat x date interactions among the Nemouridae. In the legend, dates are listed in descending series based on total invertebrate abundance. Because habitats types were not paralleled across streams and because low flow prevented sampling at some locations, habitats types were not sampled with equal intensity. 101 (p = .0333) (figure 4-5). DB1 pools had an anomalously high chironomid population: 66% of the specimens from all DB1 pool samples were chironomids. This large, highly localised chironomid population may be composed of a species unique to DB 1. Stream and habitat had highly significant effects (p = .0001 and .0003, respectively) on abundance of the Leptophlebiidae. Because the Leptophlebiidae, and less abundant taxa, comprise a much smaller proportion of total abundance, their distribution among habitats diverges from that typically seen in the total abundance, Heptageniidae, Nemouridae, and Chironomidae models. Multiple comparisons among streams show that Leptophlebiidae abundance was highest in MRN1. MRN1 is a unique channel because it is spring-fed and extremely small (1.3m wide). The subsurface origin of the channel's base flow may influence the chemical composition of the water. Because channel depth, discharge, and flow velocity vary directly with channel width (Dunne and Leopold, 1978), MRN1 is shallow, discharge is low and current velocity is slow relative to other study reaches. At least some Leptophlebiidae species prefer low velocity environments (Minshall and Minshall, 1977). The Leptophlebiidae were most abundant in riffles and pools, the current velocities of which are low compared to other habitats (figure 3-5), and least abundant on chutes and bedrock cascades which supports the idea that Leptophlebiidae prefer to avoid the full force of the current. Also, the bedrock substratum of chutes and cascades may deter the Leptophlebiidae which are known to burrow to escape current hazards (Edmunds and Allen, 1987). The effect of date was not significant on leptophlebiid abundance; however, Bonferroni multiple comparisons indicate that each date was different from the other (table 4-4). A weak 102 streamxdate interaction (p = 0.0985) effect, caused by a slightly higher abundance of Leptophlebiidae in MRN1 in October than in June, may be a result of sampling error or simply a random incident (figure 4-6). A N O V A results showed that stream and habitat had a significant effect on the abundance of the Enchytraeidae (p = .0001, .0073, respectively). Abundance was highest in ephemeral channels which were significantly different from all perennial channels; a result of the terrestrial to semi-aquatic nature of enchytraeid worms. The highest abundance of enchytraeids occurred in rapids and riffles, probably because, in ephemeral streams, these are the predominant habitats (table 3-5). The date effect was not significant on Enchytraeidae abundance; however, multiple comparisons indicate that each date was different from the other (table 4-3). A plot of means (figure 4-7) shows that habitat forced a change in enchytraeid abundance that differed from one stream to the next, thus signifying an interaction between habitat and stream (p = .0258). For example, chutes and bedrock cascades and chutes alone in DB5 and DBf, respectively, were more productive compared to those in other streams owing, perhaps, to the occurrence of alluvium on the bedrock surfaces and consequent refuge. In addition, uncharacteristically high abundance in MRN1 boulder cascades may be a result of smaller boulder size common in MRN1. Small grain size influences not only the frequency and characteristics of interstitial spaces but also one's ability to thoroughly disturb the substrate when sampling. 103 1.5 Figure 4-6: Plot of means illustrating details of stream x habitat interactions among the Leptophlebiidae. In the legend, dates are listed in descending series based on total invertebrate abundance. 2 1.5 o Figure 4-7: Plot of means illustrating details of stream x habitat interactions among the Enchytraeidae. In the legend, habitat types are listed in descending series based on total invertebrate abundance. 104 Mean abundance of the Baetidae was significantly different among streams (p = .0121) and among sampling dates (p = .0047) (table 4-3). In general, abundance was lowest in ephemeral channels; however, Abell (1984) reports that highly mobile, short-lived insects such as Baetis spp. can occupy even ephemeral channels. Mean baetid abundance was not significantly different among habitats (p = .1023) suggesting that baetids can exploit many different environments. Many Baetidae, however, have adapted to tolerate high velocity currents which accounts for high abundance on bedrock cascades and chutes and low abundance in pools (table 4-4). The Baetidae were most abundant in June, followed by August and least abundant in October. Each date was significantly different from the others suggesting that major emergence events, perhaps by different species, occurred between each sampling period. A N O V A results show that mean abundance of the Chloroperlidae was significantly different among streams (p = .0005), habitats (p = .0003), and sampling dates (p = .0043) (table 4-3). Essentially, flow regime influenced distribution of the Chloroperlidae; abundance was higher in perennial and seasonal streams than in those with ephemeral flow. Chloroperlid abundance in riffles was higher than, and significantly different from all other habitats (table 4-4). Rather than moving on, many Chloroperlidae move through the substratum to avoid changes in current (Hynes, 1970; Baumann, 1987). Riffles are most conducive to this behavioural adaptation. Chloroperlid abundance was significantly lower in October compared to June and August (table 4-4) indicating that adult emergence occurred in late summer/early autumn. 105 The results show that mean abundance of the Tricladida was significantly different among streams (p = .0014) (table 4-3). Bonferroni's multiple range tests indicated that MRN1, the stream with the highest mean triclad abundance, differed significantly from all other streams excluding NMT5 and MRN2, the two streams with the second and third highest mean triclad abundance (table 4-4). Like the Leptophlebiidae, perhaps the Tricladida are responding to the unique characteristics of M R N 1. A N O V A results also show that neither habitat (p = .3821) nor sampling date (p = .8944) had a significant effect on triclad abundance suggesting that the Tricladida exploit a variety of habitats throughout the annual cycle. A N O V A results show that mean abundance of the Simuliidae was significantly different among streams (p = .0116), habitats (p = .0002), and sampling dates (p = .0388) (table 4-3). Bonferroni's multiple range tests show that TClb , the stream with the highest mean simuliid abundance, was significantly different from all streams excluding DBf, DB5, and MRN2, the streams with the second, third and fourth highest simuliid abundance, respectively, (table 4-4). The substrata, predominantly bedrock and boulders, of these four streams are highly stable and the flow velocity, in general, is swift although this varies considerably with season. Simuliid preference for rapid, smooth flow and highly stable substrates (Hauer and Resh, 1996) explains, in part, their abundance at these sites. In addition, high simuliid abundance on chutes and low abundance in pools reflect these preferences. October was significantly different from all other sampling dates, with respect to simuliid abundance which suggests that emergence occurred in late summer/early fall. 106 Stream (p = .0164) had a significant effect on mean abundance of the Lepidostomatidae (table 4-3). Lepidostomatid abundance was highest in DB1 and was significantly different from three sites with very low abundance, DB5, MTD1, and NMT4. The remaining sites were similar (table 4-4). A N O V A results also indicate that abundance of the Lepidostomatidae was significantly different among habitats (p = .0544); however, Bonferroni multiple comparisons suggested that one habitat was not different from the other (table 4-4). The effect of sampling date was not significant (p = .3187). The effects of streamxhabitat (p = .0705) and habitatxdate (p = .0170) interactions were significant. A plot of means (figure 4-8), illustrates that habitat forced a slightly different change in lepidostomatid abundance from one stream to the next. Because Bonferroni's multiple range test failed to distinguish habitats on the basis of Lepidostomatidae abundance, the interaction is difficult to interpret. Also, the strength of this interaction was fairly weak. The Lepidostomatidae are known to occur in leaf detritus (Wiggins, 1987). Distribution of the Lepidostomatidae may be a reflection of detritus availability rather than habitat type preference. The habitatxdate interaction may be a result of staggered life histories of related species which allows more species to utilize the resources of the environment and to avoid competition (figure 4-8) (Mackay and Kalff, 1969). Low macroinvertebrate abundance in streams with ephemeral flow regimes was conspicuously contrasted by much higher abundance in channels with seasonal or perennial flow. Water, more than any other component of the environment, controls the distribution of aquatic organisms. Therefore, macroinvertebrate response to various habitat types may be 107 0.8 0.4 Figure 4-8: Plots of means illustrating details of (A) stream x date interactions and (B) habitat x date interactions among the Lepidostomatidae. Habitat types and dates are listed in descending series, in the legends, based on total invertebrate abundance. 108 different in ephemeral channels, compared to perennial streams, as a result of severe flow intermittency. This circumstance may confound observed habitat preferences; therefore, channels with ephemeral flow regimes were separated from the others and were analysed independently. Results of an A N O V A applied to macroinvertebrate data collected from ephemeral streams showed that habitat was not a significant factor influencing total macroinvertebrate abundance (p = .1056) or abundance of the Heptageniidae (p = .1477), the Nemouridae (p = .3135), the Chironomidae (p = .9676), the Leptophlebidae (p = .9382), the Enchytraeidae (p = .2832), the Baetidae (p = .9005), the Chloroperlidae (p = .9500), the Tricladida (p = .5945), and the Lepidostomatidae (p = .3261). The Simuliidae were not represented in any of the ephemeral channels; therefore, there was no variance among habitats with respect to simuliid abundance. However, the significance of habitat effects remained unchanged when an A N O V A was applied to macroinvertebrate data collected from seasonal and perennial streams. This outcome may be a result of the generally low abundance and even distribution of macroinvertebrates among habitat types in ephemeral streams. The stream effects on abundance of the Chironomidae and the Lepidostomatidae were lost (p = .2444 and .1074, respectively), perhaps because chironomid and lepidostomatid abundance was similar in all streams except MTD1 and NMT4. Abundance of the Chironomidae and the Lepidostomatidae was comparatively low in the two aforementioned ephemeral streams (table 4-4). The effect of sampling date on total macroinvertebrate abundance was significant, although weak (p = .0942), following removal of ephemeral streams from the data set. The original non-significant result (table 4-3) may be a consequence of the Enchytraeidae, the abundance of which is not influenced by season. Ephemeral channels were sampled during October only, a time during which the Enchytraeid 109 worms were prevalent, thus contributing to higher overall abundance on that date. However, elimination of ephemeral streams from the data set caused overall macroinvertebrate numbers to decline in October, but not in June or August, thereby increasing the difference in total abundance among sampling dates. The effect of sampling date on the Nemouridae was insignificant (p = .1074) when ephemeral streams were eliminated from the data set. The low abundance of the Nemouridae in ephemeral streams, which were sampled in October only, suggests that the Nemouridae were scarce in October in general. However, the paucity of Nemouridae may be independent of sampling date because when these ephemeral streams were removed from the data set, abundance was similar throughout the sampling dates. Thus the original significant result (table 4-3) may have been a consequence of inconsistent flow conditions rather than phenological progression. The non-significant result corresponds with the seasonal succession of emerging Nemouridae species throughout most of the year (Merritt and Cummins, 1996). The elimination of ephemeral streams from the data set also influenced two way interactions. Stream x habitat interactions exerted statistically significant yet weak (p = .0523 and .0947, respectively) influences on total macroinvertebrate and Chironomidae abundance whereas the significant effect of such an interaction on the Leptophlebiidae was lost (p - .2304) after manipulation of the macroinvertebrate data. Stream x habitat interactions may be a result of inconsistent habitat characteristics among streams. Examples have been discussed and may include riffles comprised of thin gravel layers, boulder cascades consisting of relatively small clasts, or alluvial material upon bedrock cascades. In addition, abundance of the 110 Enchytraeidae was affected by a weak stream x date interaction (p = .0925) which apparently was caused by relatively high abundance in DB3, a seasonal stream with similar numbers of enchytraeid worms compared to the ephemeral streams (table 4-4). A N O V A models were employed to test the effects of stream, habitat, and date, as well as the effects of two-way and three-way interactions, on total macroinvertebrate abundance and abundance of the ten commonest taxa (table 4-3). However, because hypotheses regarding three-way interactions were not formulated, statistical examination of such an interaction was not required, therefore, it was omitted from each A N O V A model. This modification allowed examination of the influence of the three-way interaction on the other sources of variation. Consequently, mean abundance estimates associated with the main and two-way interaction effects theoretically were improved. Elimination of the three-way interactions increased experimental error degrees of freedom thereby narrowing confidence intervals surrounding mean abundance estimates. When tables 4-3 and 4-5 are cross-referenced, it is apparent that the signficance probability (or p-value) of the stream effect on total macroinvertebrate abundance and abundance of the Lepidostomatidae decreased following removal of the three-way interactions from the A N O V A models. Similarly, the signficance probability of the habitat effect on abundance of the Baetidae and the Lepidostomatidae decreased. Also, the p-value of the date effect on total abundance decreased following removal of the three-way interactions from the A N O V A models (table 4-5). A small p-value means a strong rejection of H 0 . The signficance I l l probability of the stream effect on baetid abundance increased following removal of the three-way interactions from the A N O V A models. In the same way, the p-value of the date effect on abundance of the Nemouridae increased. A large p-value means that H 0 fails to be rejected. The significance probabilities of the main and interaction effects on mean abundance were not altered enough to change the level at which they were significant originally (marked by asterisks in tables 4-3 and 4-5) in the majority of the A N O V A models following elimination of the three-way interactions. Subsequent discussions are based on the original A N O V A models. Table 4-5: Results of three-factor A N O V A models for effects of stream, habitat, date, and two way interactions on total invertebrate abundance and on individual taxon abundance for the ten most common taxa in the study. Dependent variable stream x stream x habitatx stream F habitat F dateF habitat F date F date F total abundance 1.80* 4.86*** 2.60* 1.44 1.58 0.55 Heptageniidae 3.64*** 4.65*** 3.17** 1.84** 1.38 0.25 Nemouridae 417*** 3 7^*** 2.32 1.04 0.66 0.49 Chironomidae 1.94* 2.12* 5 99*** 1.44 2.66** 0.49 Leptophlebiidae 7 22*** 3 44*** 1.87 0.93 2.02* 0.76 Enchytraeidae 5.31*** 4.50*** 0.26 1 9 ! * * 1.78 0.46 Baetidae 1.88* 1.99* 4 go*** 0.57 1.94 0.50 Chloroperlidae 3 23* * * 3.57*** 5.50*** 1.09 1.15 1.14 0. Tricladida 3 37*** 1.11 0.05 1.12 1.27 0.68 Simuliidae 2.60** 3 <;g*** 4.55** 1.06 1.18 1.52 Lepidostomatidae 2 77*** 2.36** 1.46 1.49* 1.89 2.36** 1. Asterisks indicate p < 0.10 (*), p < 0.05 (**), and p < 0.01 (***). 2. Degrees of freedom for main effects and two-way interactions in all models are as follows: stream = 9, habitat = 5, date = 2, streamxhabitat = 25, streamxdate - 4, habitatxdate = 8, streamxhabitatxdate = 9. Error degrees of freedom for each model = 92. 112 4.3.3 Macroinvertebrate Community Structure Dominant Taxa Table 4-6 lists, by stream, the most abundant taxa in each habitat type. On bedrock cascades, families Baetidae and Chironomidae were among the five most abundant taxa, based on proportion of mean abundance, in 5/5 streams sampled and families Heptageniidae and Simuliidae were among the five most abundant taxa in 3/5 streams. In boulder cascades, family Nemouridae and order Tricladida were among the five most abundant taxa in 6/7 and 4/7 streams sampled, respectively. Families Heptageniidae, Enchytraeidae, and Chironomidae were also among the most abundant taxa in boulder cascades but in only 3/7 channels. In rapids and chutes, Families Nemouridae, Heptageniidae, and Enchytraeidae were among the five most abundant taxa in more study reaches than any other taxa. In riffles, Families Nemouridae, Heptageniidae and Chironomidae were among the five most abundant taxa in more than 7/11 study reaches. In pools, Family Chironomidae was among the five most abundant taxa in 7/12 streams sampled, Family Heptageniidae in 6/12 study reaches and Families Leptophlebiidae and Enchytraeidae in 5/12 streams. Because the Heptageniidae, Nemouridae, and Chironomidae were far more abundant (table 4-2) and more ubiquitous than other taxa, other families provided a means to distinguish communities among habitats. The Physical System The results of a principal components analysis (PCA) processed on the array of 95 sites by four environmental variables are presented in figure 4-9-A and -B. Principal components (PC) 1 and 2 account for 40.42% and 24.76% of the total variance, respectively. The eigenvalues of both principal components 3 and 4 were less than 1 (figure 4-9-A inset), Table 4-6: Proportional abundance of most common taxa in each habitat type and each stream. Bedrock Cascade % Boulder Cascade % Rapid % Chute % Riffle % Pool % DB 1 Brachycentridae 16 Heptageniidae 30 Leptophlebiidae 12 Chironomidae 66 Baetidae 15 Nemouridae 13 Polycentropodidae 10 Heptageniidae 7 Chironomidae 12 Order Tricladida 9 Heptageniidae 10 Lepidostomatidae 5 Heptageniidae 9 n/a n/a Chloroperlidae 8 Chironomidae 10 Leptophlebiidae 5 Glossosomatidae 9 Hydropsychidae 5 Chloroperlidae 9 Chloroperlidae 2 Cumulative % 61 Cumulative % 65 Cumulative % 51 Cumulative % 85 DB 3 n/a Nemouridae 33 Enchytraeidae 48 Enchytraeidae 47 Enchytraeidae 25 Enchytraeidae 13 Chironomidae 30 Nemouridae 17 Chloroperlidae 18 Tipulidae 13 Ceratopogonidae 8 Chloroperlidae 6 Tipulidae 11 n/a Cumulative % 59 Class Nemotoda Cumulative % 3 89 n/a Order Tricladida Chironomidae Cumulative % 6 5 81 Cumulative % 54 DB 5 Baetidae 25 Heptageniidae 32 Heptageniidae 24 Simulidae 48 Heptageniidae 22 Leptophlebiidae 16 Chironomidae 20 Nemouridae 30 Nemouridae 20 Nemouridae 18 Nemouridae 20 Heptageniidae 15 Enchytraeidae 15 Glossosomatidae 14 Leptophlebiidae 9 Heptageniidae 8 Leptophlebiidae 8 Ameletidae 13 Tipulidae 15 Leptophlebiidae 1 Chironomidae 8 Baetidae 6 Chloroperlidae 8 Nemouridae 11 Simulidae 10 Enchytraeidae 7 Enchytraeidae 4 Enchytraeidae 5 Baetidae 5 Baetidae 11 Cumulative % 85 Cumulative % 90 Cumulative % 65 Cumulative % 85 Cumulative % 63 Cumulative % 66 DBf Nemouridae 26 Nemouridae 20 Nemouridae 19 Chironomidae 23 Nemouridae 20 Chironomidae 21 Simulidae 15 Chironomidae 19 Chironomidae 16 Simulidae 16 Heptageniidae 14 Polycentropodidae 17 Chironomidae 12 Baetidae 12 Tipulidae 11 Nemouridae 13 Chironomidae 11 Leptophlebiidae 13 Heptageniidae 10 Simulidae 9 Heptageniidae 8 Enchytraeidae 12 Enchytraeidae 9 Enchytraeidae 7 Baetidae 7 Order Tricladida 7 Rhyacophilidae 7 Rhyacophilidae 8 Leuctridae 1 Heptageniiidae 7 Cumulative % 70 Cumulative % 67 Cumulative % 61 Cumulative % 72 Cumulative % 61 Nemouridae 7 Cumulative % 72 1. Those habitats marked n/a were not present in the specified streams. 2. Italics indicate most abundant taxa excluding the top three of the study (Heptegniidae, Nemouridae, and Chironomidae). 113 Table 4-6 (continued): Proportional abundance of most common taxa in each habitat type and each stream. Bedrock Cascade % Boulder Cascade % Rapid % Chute Riffle % Pool % MRN 1 Leptophlebiidae 21 Nemouridae 30 Leptophlebiidae 25 Leptophlebiidae 35 Order Tricladida 13 Leptophlebiidae 76 Nemouridae 24 Nemouridae 20 Philopotamidae 8 Order Tricladida 9 Chironomidae 8 Chloroperlidae 17 n/a Hydroptilidae 8 Chironomidae 7 n/a Chloroperlidae 7 Leuctridae 8 Nemouridae 8 Chloroperlidae 7 Order Tricladida 6 Order Tricladida 4 Cumulative % 58 Cumulative % 69 Cumulative % 70 Cumulative % 84 MRN 2 Baetidae 56 Heptageniidae 24 Nemouridae 23 Heptageniidae 25 Heptageniidae 33 Chironomidae 14 Baetidae 18 Nemouridae 17 Chironomidae 11 Nemouridae 12 Heptageniidae 15 Ameletidae Cumulative % 100 n/a Baetidae 7 Order Tricladida 15 n/a Chironomidae 6 Simulidae 3 Cumulative % 71 Baetidae 6 Cumulative % 60 Cumulative % 62 MTD 1 Enchytraeidae 70 Enchytraeidae 52 mite 42 Enchytraeidae 44 Order Collembola 7 Nemouridae 9 Enchytraeidae 39 mite 33 Staphylinidae 10 Order Hemiptera 9 Order Collembola 7 Chironomidae 11 n/a Cumulative % 87 Cumulative % 100 n/a Class Nemotoda 5 Stratiomyidae 11 Tipulidae 4 Cumulative % 99 Cumulative % 97 NMT 4 Enchytraeidae 67 Enchytraeidae 50 Enchytaeidae 75 Class Gastropoda 6 Class Gastropoda 12 Order Collembola 25 mite 6 Class Oligochaeta 8 Cumulative % 100 n/a n/a n/a Order Tricladida 5 Staphylinidae 6 Tipulide 4 Hydroptilidae 6 Order Collembola 4 Cumulative % 82 Cumulative % 86 1. Those habitats marked n/a were not present in the specified streams. 2. Italics indicate most abundant taxa excluding the top three of the study (Heptegniidae, Nemouridae, and Chironomidae). 114 Table 4-6 (continued): Proportional abundance of most common taxa in each habitat type and each stream. Bedrock Cascade % Boulder Cascade % Rapid % Chute % Riffle % Pool % NMT 5 Enchytraeidae 28 Order Tricladida 27 Enchytraeidae 55 Nemouridae 14 Chironomidae 18 Chironomidae 27 Order Tricladida 8 Enchytraeidae 18 Order Tricladida 6 Chironomidae 7 Tipulidae 18 Tipulidae 6 Chloroperlidae 6 Heptageniidae 9 Class Gastropoda 3 Cumulative % 63 Lepidostomatidae 9 Carabidae 3 Cumulative % 99 Cumulative % 100 TC 1 Brachycentridae 50 Leptophlebiidae 30 Chironomidae 14 Polycentropodidae 13 Enchytraeidae 14 Leuctridae 13 Lumbriculidae 7 n/a n/a n/a n/a Tubificidae 8 Stratiomyidae 7 Chironomidae 8 Cumulative % 92 Cumulative % 72 TC lb Nemouridae 19 Simulidae 87 Heptageniidae 28 Polycentropodidae 23 Baetidae 19 Heptageniidae 8 Chironomidae 11 Leptophlebiidae 19 Chironomidae 14 Nemouridae 4 Nemouridae 9 Leuctridae 18 Simulidae 10 n/a n/a Leuctridae 2 Leuctridae 9 Chironomidae 15 Tipulidae 9 Cumulative % 100 Chloroperlidae 9 Enchytraeidae 4 Cumulative % 71 Cumulative % 66 Cumulative % 79 TC43 Heptageniidae 43 Heptageniidae 56 Hepatageniidae 66 Heptageniidae 52 Heptagneiidae 51 Baetidae 22 Nemouridae 25 Nemouridae 12 Nemouridae 33 Nemouridae 20 Nemouridae 18 Baetidae 8 Baetidae Chironomidae 3 Order Tricladida 7 n/a Order Tricladida 5 Enchytraeidae 4 Order Tricladida 5 Baetidae 3 Leptophlebiidae 4 Chironomidae 4 Chironomidae 2 Uenoidae 3 Enchytraeidae 2 Enchytraeidae 3 Cumulative % 92 Cumulative % 95 Cumulative % 94 Cumulative % 93 Cumulative % 85 1. Those habitats marked n/a were not present in the specified streams. 2. Italics indicate most abundant taxa excluding the top three of the study (Heptegniidae, Nemouridae, and Chironomidae). 115 Table 4-6 (continued): Proportional abundance of most common taxa in each habitat type and each stream. Bedrock Cascade % Boulder Cascade % Rapid % Chute % Riffle % Pool % TQ1 Nemouridae 26 Heptageniidae 30 Nemouridae 27 Heptageniidae 13 Nemouridae 22 Heptageniidae 23 n/a Order Tricladida 11 n/a n/a Lepidostomatidae 9 Chloroperlidae 9 Lepidostomatidae 10 Chloroperlidae 1 Order Tricladida 6 Chironomidae 9 Rhyacophilidae 4 Chironomidae 5 Cumulative % 69 Cumulative % 72 Cumulative % 70 1. Those habitats marked n/a were not present in the specified streams. 2. Italics indicate most abundant taxa excluding the top three of the study (Heptegniidae, Nemouridae, and Chironomidae). 116 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 -2 -1 0 1 2 3 4 < • substrate (-.65) gradient (+.79) velocity (+.73) PCI (40%) Figure 4-9: Results of a PCA processed on the array of 95 sites by 4 environmental variables. (A) Plot of factor loadings with an eigenvalue diagram inset. (B) Plot of factor scores. Variables significantly correlated with each axis are shown outside plot (correlation coefficients in parentheses). Symbols represent habitat types and fills represent streams. V= bedrock cascade, 0 = boulder cascade,0 = chute, •= pool,0= rapid, A= riffle, •= DB1, •= DB3, •= DB5,1= DBf, •= MRN1, •= MRN2, •= MTD 1,H = NMT4, M= NMT5,E3= TClb,H= TC43, andl=TQl. therefore, were not included, a decision based on the Kaiser criterion (Statistica, 1994). To reveal physical variables with strong influences on ordinations, each variable was correlated with PC 1 and 2 (table 4-7). The effects of gradient and velocity increase to the right of the origin and are significantly correlated with PC 1 (r = .79 and .73, respectively). PC 1, therefore, can be said to separate habitats along a gradient from depositional (pools) on the left to erosional (cascades) on the right (figure 4-9-B). The effect of substrate, on the other hand, decreases to the right of the origin and also is significantly correlated with PC 1 (r = -.65). PC 1, therefore, is said to separate habitats along a gradient from alluvial (pools, riffles) on the left to non-alluvial (cascades) on the right. Shade is highly and significantly correlated (r = .96) with PC 2; consequently, shaded sites appear near the top of the ordination diagram, and vice-versa. A high shade level is not associated with any one stream or habitat type but with a specific site, depending upon topographic shading and canopy gaps; therefore, a pattern among sites is not obvious. Substrate and velocity are also significantly correlated with PC 2 but the correlations are much weaker (r = .19 and .23, respectively). Ordination of streams and habitats illustrates adjacent and overlapping clusters of chute and bedrock cascades suggesting that these habitats are similar with regard to the measured environmental variables. Close examination of figure 4-9-B reveals that bedrock cascades are displaced further to the right of the origin than chutes; a reflection of higher bed gradient, current velocity and sediment transport capacity. Factor scores of rapids also group; however, less distinctively than chutes and bedrock cascades. Pools, riffles and boulder cascades, which are spread throughout the ordination space, do not form distinguishable groups. This Table 4-7: Correlation matrix of taxa, environmental variables, and principal components. hep nem chi lepl enc bae chl tri sim lep2 leu tip rhy pol pup hyd mite lum vel gra sub sha biol bio2 physl phys2 hep 1.00 nem 75*** 1.00 '" chi 34*** 1.00 lepl 2g*** 23*** 43*** 1.00 enc -.08 .07 23** -.09 1.00 i bae .68*** 51*** 35*** .08 -.02 1.00 chl 3g*** 53*** 37*** .50*** .02 .06 1.00 tri 39*** 5 j *** 40*** .25** .14 32*** 45*** 1.00 sim .19* .26** 34*** .12 -.02 27*** .04 -.00 1.00 lep2 32*** 30*** 55*** 39*** -.05 .11 .62*** 45*** .02 1.00 leu .22** 35*** 4g*** 53*** -.06 -.02 73*** 31*** 23** .60*** 1.00 i tip .10 45*** 43*** .16 .06 45*** 39*** .25** .24** 41 *** 1.00 rhy 42*** 52*** .56*** 27*** .05 3Q*** 47*** 33 *** 42*** 45*** 51*** 49*** 1.00 pol -.08 -.09 .35*** 39*** -.02 -.13 .19* .00 -.04 .30*** 45*** .02 .07 1.00 pup .17* .14 26*** .04 .09 20** .05 .20** .12 .04 -.07 .08 .03 -.04 1.00 hyd 3 J *** 4g*** 32*** .19* -.10 25** 40*** 5 j *** .07 46*** 38*** 37*** 40*** .06 29*** 1.00 mite -.13 -.02 23** 34*** .24** -.14 .16 .09 -.08 30*** .25** .11 .03 33*** -.05 .09 1.00 lum .18* 43*** 32*** 32*** .04 .09 64*** 45*** 22** 45*** 59*** 51*** 43*** .07 -.05 .48* .17 1.00 vel .09 .18* .06 27*** .04 27*** -.12 .08 2i** -.10 -.06 .10 .14 -.08 .07 ;15 -.16 .00 1.00 gra -.14 -.02 .10 -.24** -.04 2g*** 29*** .01 .14 -.17 _ 27*** .03 .08 -.16 .03 1.00 -.11 -.13 .41* 1.00 sub 2g*** .25** .03 .23** .05 .07 .13 .08 -.17 -.02 .03 .12 .06 .01 .13 !08 .07 .06 -.20* -.26* 1.00 sha -.04 -.01 .09 .04 -.11 -.02 -.04 .09 .13 .10 .01 -.01 .03 .06 .05 .11 .14 -.02 .12 .07 -.02 1.00 biol .05 2g*** 58*** 55*** .19* . -.17* 74*** 43*** .03 .68*** g2*** ^2*** 43*** .55*** -.01 41 *** 54*** 64*** -.17 -.26** .10 .06 1.00 bio2 §] *** 7g*** 40*** .05 -.08 82*** .25** 44*** 47*** .22** .11 27*** .57*** 31 *** 3 j *** "44*** _ 32*** 95** 29*** .18* .11 .10 -.09 1.00 physl -.15 -.04 .06 _ 34*** -.02 23** -.25** .00 22** -.12 -.18* .01 .07 -.13 -.01 .02 . -.16 -.09 73*** 79*** -.65*** .05 -.25** .17 1.00 phys2 .05 .08 .11 .05 -.08 .04 -.03 .12 .12 .07 .01 .03 .07 .05 .09 •'15 ; .13 -.00 .23* .06 .19* 96*** .05 .08 .00 1.00 1. Asterisks indicate p< 0.10 (*), p< 0.05 (**), p< 0.01 (***). 2. Heptageniidae=hep, Nemouridae=nem, Chironomidae=chi, Leptophlebiidae=lepl, Enchytraeidae=enc, Baetidae=bae, Chloroperlidae=chl, Order Tricladida=tri, Simuliidae=sim, Lepidostomatidae=lep2, Leuctridae=leu, Tipulidae=tip, Rhyacophilidae=rhy, Polycentropodidae=pol, Diptera pupae=pup, Hydropsychidae=hyd, and Lumbriculidae=lum, vel=velocity, gra=gradient (bed slope), sub=substrate, sha=shade, biol and 2=biological PC 1 and 2, and phys 1 and 2=physical PC 1 and 2. 120 suggests that pools, riffles and boulder cascades are highly variable. A boulder cascade in close proximity to a pool, in ordination space, is testimony of that. The Biological System The results of a PC A processed on the array of 182 sites by 18 taxa are presented in figure 4-10-A. The number of sites included in the biological PC A is almost double that of the physical P C A as a result of multiple sampling dates as well as incomplete velocity and dominant substrate data. Those taxa accounting for > 1% of the macroinvertebrate data set, hence 18 taxa, were included in the P C A analysis. The first, second, and third principal components accounted for 35.27%, 10.83%, and 8.66% of the total variance, respectively. In total, six principal components had eigenvalues greater than 1 (figure 4-10-A inset), therefore, should have been included according to the Kaiser criterion (Statistica, 1994). However, only the first two PCs were chosen in the interest of simplicity. To reveal taxa with strong influences on ordinations, each taxon was correlated with PC 1 and 2. The three taxa correlated most strongly with axis 1 are: Leuctridae (r = .79), Chloroperlidae (r = .74), and Leptophlebiidae (r = .72). The Leptophlebiidae, Chloroperlidae, and Leuctridae are the fourth, seventh and eleventh most abundant taxa in the study, respectively. The three taxa correlated most strongly with axis 2 are: Baetidae (r = .82), Heptageniidae (r = .81) and Nemouridae (r = .71). The Heptageniidae, Nemouridae, and Baetidae rank first, second, and sixth in overall abundance. Additional taxa significantly correlated with each principal component are presented in table 4-7. u 1.0 0.8 -0.6 -0.4 -0.2 -0.0 --0.2 --0.4 Baetidae Heptageniidae Leptophlebiidae Enchytraeidae Hydracarina Polycentropodidae T T 1 1 1 0.2 0.0 0.2 0.4 0.6 0.8 Baetidae (+.82) Heptageniidae (+.81) Nemouridae (+.71) 1.0 PCI (35%) Leuctridae (+.79) Chloroperlidae (+.74) Leptophlebiidae (+.72) Figure 4-10: (A) Plot of factor loadings processed on the array of 182 sites by 18 taxa with an eigenvalue diagram inset. (B) Plot of factor scores. The 3 taxa most highly correlated with each axis are shown outside plot (correlation coefficients enclosed in parentheses). Symbols represent habitat types and fills represent streams. A = bedrock cascade, O = boulder cascade, O = chute, • = pool, O = rapid, V = riffle, • = DB1, • = DB3, • = DB5, • = DBf, • = MRN1, • = MRN2, • = MTD1, B = NMT4, M = NMT5, ED=TClb, ID = TC43, and B = TQ1. ,—. to 122 When streams and habitats were ordinated, the overlap was extensive, thus concealing any patterns that may otherwise be apparent (figure 4-10-B). This indicates that various habitats are similar in terms of the 18 taxa used in the analysis. Mean invertebrate density for each stream-habitat combination was calculated and a PCA was performed on this composite data set in order to reveal broad patterns in community structure among habitats and streams (Angradi, 1996). The results of a P C A processed on the array of 52 sites by 18 taxa are presented in figure 4-11. The first and second axes accounted for 38.70% and 13.47% of the total variance, respectively. To reveal taxa with strong influences on ordinations, each taxon was tested for correlation with both PC 1 and PC 2. The three taxa most strongly correlated with both PC 1 and PC 2 are identical to the non-composite data; however, the correlation coefficients, shown outside the ordination plot (figure 4-11), were higher in the composite data in all cases. Accordingly, the Leuctridae, Chloroperlidae, and Leptophlebiidae are most abundant at the sites on the right side of the ordination diagram and the Baetidae, Heptageniidae, and Nemouridae are most abundant at the sites near the uppermost portions of the plot. The ordination of streams and habitats revealed three groups which were identified through visual examination of the ordination diagram: TC43, bedrock cascades and chutes, and channels with seasonal or ephemeral flow regimes. Manually placed ellipses demonstrate these groups in figure 4-11. The TC43 group is exclusive to this stream and is distant, in ordination space, from the seasonal/ephemeral group which is less exclusive but compact. The bedrock cascade and chute group is neither exclusive nor compact, suggesting that there is variability within the group and similarities with other habitats. 123 Baetidae (+.88) Heptageniidae (+.84) Nemouridae (+.83) 2 -1 -U a. 0 --1 -T • ephemeral channels T 0 PC 1 (39%) Leuctridae (+.85) Chloroperlidae (+.82) Leptophlebidae (+.79) Figure 4-11: Results of a PCA processed on the array of 52 sites by 18 taxa (composite data). Diagram is a plot of factor scores. Symbols represent habitat types and fills represent streams. A = bedrock cascade, 0 = boulder cascade, O = chute, D = p o o l O = rapid, V = riffle, D = DB1, B = DB3, • = DB5, B = DBf, B = MRN1, B = MRN2, B =MTD1 ,B = N M T 4 , ^ = NMT5, H = TCI , ES3 = TClb , B = TC43, and ^ = TQ1. 124 Linking The Systems To relate the physical and biological systems principal components 1 and 2 of the physical system were correlated with principal components 1 and 2 of the biological system (non-composite data) (table 4-7). PC 1 of the physical system was significantly correlated (r = -.25) with PC 1 of the biological system suggesting that the primary faunal gradient is related to bed slope, velocity, and substrate, the primary environmental variables (figure 4-12). Biological principal component 2, however, is not significantly related to either physical component (table 4-7). To proceed further in an explanation of this component, measurement of more environmental factors is required (Barkham and Norris, 1970). To explain the invertebrate variation in terms of the physical system, environmental variables were correlated with biological PCs 1 and 2 (table 4-7). Bed slope (r = -.26) had a significant negative correlation with PC 1 while PC 2 also corresponded to a gradient in bed slope (r = .18) as well as flow velocity (r = .29). In chapter three, current velocity was directly related to water surface slope and depth with the Chezy Formula and the Manning relation and a simple correlation depicted this direct relationship (figure 3-5). The first biological principal component separated sites along a gradient from steep slopes, therefore high current velocity, to the left of the origin to mild slopes and lower current velocities to the right of the origin. Sites were separated along PC 2 by bed slope as well; however, velocity had greater effect than bed slope on the ordination. The sites, therefore, are more apparently ordered along axis 2 according to a gradient in velocity, decreasing from top to bottom, as opposed to a gradient in bed slope. Essentially, abundance of the Chloroperlidae, Leuctridae, and Leptophlebidae, etc. increases with mild bed slopes, low current velocity, and smaller bed-material grain size. Figure 4-12: Scatter plot of biological PC 1 factor scores vs. physical PC 1 factor scores The correlation coefficient is significant (p < 0.05) 126 The larger perennial channels with more water, hence higher current velocities, are near the top of the ordination diagram whereas the smaller seasonal and ephemeral channels are near the bottom of the ordination diagram. In general, the correlations are weak; therefore, the patterns discussed are not prominent on the ordination diagrams. Ordination indicates that the invertebrate communities are responding to a gradient in water persistence as well; however, flow regime was not accounted for formally. Biological PC 1 factor scores varied significantly among both streams and habitats (p = .0001). Physical PC 1 factor scores varied significantly among habitats (p = .0001) but not among streams (p = .1277). Because correlations showed that physical PC 1 varied linearly with biological PC 1 (r = -.25), analysis of covariance (ANCOVA) procedures were used to investigate the effects of stream and habitat on the biological PC 1 factor scores while removing the relationship between the dependent variable (biological PC 1 factor scores) and the concomitant variable (physical PC 1 factor scores). After removing the insignificant effect of physical PC 1 factor scores (F, 5 2 ; p = .8499), streams ( F n > 5 2 ; p = .0001) and habitats (F5 5 2 ; p = .0007) remained significantly different. This is consistent with the results of the correlation between biological and physical P C I . In conclusion, the physical PC 1 factor scores did not add to the amount of variation explained by the biological PC 1 factor scores perhaps because flow regime was not included in the physical P C A but appeared to be the most important variable influencing ordination of the faunal data. A posteriori comparisons, using the Bonferroni method, were employed to determine which streams and which habitats are different with respect to mean biological PC 1 factor scores. 127 The results of the Bonferroni tests indicated that TQ1 was different from TClb , DB3, DBf, DB1, NMT5, NMT4, MTD1, DB5, MRN2, and TC43. Also, MRN1 was different from DBf, DB1, NMT4, MTD1, DB5, MRN2, and TC43, T C l b was different from DB5, and D B f was different from TC43. With respect to mean biological PC 1 factor scores among habitat types, Bonferroni multiple comparisons indicated that riffles were different from rapids, boulder and bedrock cascades, and chutes. In addition, pools were different from both bedrock cascades and chutes. 4.4 Discussion 4.4.1 Inter-habitat Variation This study documents highly significant differences among aquatic habitats with respect to total invertebrate abundance and the abundance of specific taxonomic groups (families Heptageniidae, Nemouridae, Leptophlebiidae, Enchytraeidae, Chloroperlidae, and Simuliidae) which indicates that species within these families may be expressing habitat preferences. The effect of habitat also was significant, but not as strong, on the abundance of families Chironomidae and Lepidostomatidae but was not significant on family Baetidae and order Tricladida (table 4-3). This outcome suggests that either the species present are habitat generalists or that habitat preferences are not consistent among species. Both the Chironomidae and the Baetidae are known for their ubiquitous distribution (Hynes, 1970; Abell, 1984; Clifford, 1991). For instance, in their comparisons of pool and riffle communities, Brown and Brussock (1991) determined that virtually all taxa were more abundant in riffles than in pools, except chironomids which were more equally distributed. 128 Similarly, Williams (1978) found that all animals displayed habitat preference in a rearing channel on the east coast of Vancouver Island with the exception of two chironomid species and one baetid species . Differences in community structure among habitat types were less distinct; only habitats with predominantly bedrock substrata could be definitively distinguished from other habitats (figure 4-10) probably as a result of the Simuliidae and the Baetidae (table 4-4). Riffles and rapids were more productive than any other habitat type (figure 4-1) which agrees with the conclusions of Pennak and Van Gerpen (1947) and Minshall (1984) that bed particles of intermediate size support a more abundant fauna than either very small or very large particles. Furthermore, all taxonomic groups were most abundant in either riffles or rapids except the Baetidae and Simuliidae (table 4-4). Principle Components Analysis showed that the community structures of riffles and rapids were not distinct (figure 4-10) suggesting that the conditions created by these heterogeneous mixtures of bed particles are ideal for co-existence of many species (Stanford and Ward, 1983; Minshall, 1984). Riffles and rapids may be suitable for a variety of animals because they offer diverse flow conditions, for example, erosional conditions occur on the surfaces of riffle gravel and in transverse ribs of rapids whereas secondary pools, interstitial spaces and dead zones upstream of clasts offer retreat from the current. The greatest proportion of interstices whose dimensions are neither too small nor too large for the free movements and activities of organisms occur in coarse gravel (Pennak and Van Gerpen, 1947). Since particle size varies directly with channel slope and current velocity (Dunne and Leopold, 1978), particle stability may be inferred from particle size. As predicted by the intermediate disturbance hypothesis 129 and the habitat templet model (Connell, 1978; Scarsbrook and Townsend, 1993), habitats of intermediate stability and relatively small average grain size, such as riffles and rapids, supported larger and possibly more diverse faunas compared to highly stable sites such as bedrock cascades. Abundance results showed that bedrock cascades and chutes were less productive than any other habitat type (figure 4-1) and that bedrock substratum was avoided by most taxonomic groups excluding the Simuliidae, Baetidae, and Chironomidae (table 4-4). Brown and Brussock (1991) also attributed below average abundance of organisms to the predominance of barren bedrock, in part. P C A results showed that the community structure of bedrock cascades and chutes was more distinctive than any other habitat type (figure 4-10). Owing to high transport capacities, bedrock cascades have virtually no alluvial material on their surfaces therefore provide little protection from the full force of the current. Because collector-fllterers often depend on current (Brussock and Brown, 1991) they are well suited to exploit these conditions, using holdfast structures to maintain position while capturing entrained food resources with catchnets or modified appendages (Grubaugh et al, 1996). For example, the association between Simulium, a black ly genus from the family Simuliidae, and highly stable substrata and areas of rapid flow recurs in the literature (Ward, 1975; Minshall and Minshall, 1977; Logan and Brooker, 1983; Wohl et al, 1995; Hauer and Resh, 1996). The bed slope and, consequently, transport capacity of chutes is less severe; therefore, alluvial material may provide refugia and increase the diversity of microhabitats. The abundance of the Heptageniidae and Nemouridae, for instance, is often higher in chutes than in bedrock cascades, (table 4-4). 130 Faunal data pooled from 13 streams and 3 sampling dates show that mean total abundance was lower in both bedrock cascades and chutes compared to boulder cascades (figure 4-1); however, the total invertebrate abundance A N O V A model indicates that boulder cascades are the least productive habitat type (table 4-4). This discrepancy may be a result of T Q l elimination from the A N O V A model. T Q l is a productive, perennial channel composed predominantly of boulder cascades. In any event, boulder cascade productivity is relatively low, perhaps owing to limited availability of stone surface area compared to smaller, less compact particles (Resh, 1979). In addition, exposed surfaces of boulders, which serve as bedrock (Minshall, 1984) from the perspective of insects, and interstitial spaces between boulders provide a variety of microhabitats; however, water often does not flow over, but around, the boulders and the interstitial spaces may not afford free movement of organisms. Heavy, tightly imbricated boulders are difficult to sample because movement of and disturbance between the boulders often is impossible. The evidence supporting the conclusion that total invertebrate abundance is low in boulder cascades may be an artifact of the sampling technique. Pools often were intermediate units, with regard to animal abundance, except for that of the Baetidae and Simuliidae which rely on swift currents for survival (figure 4-1 and table 4-3). This is consistent with the observations of Brussock and Brown (1991) and Scarsbrook and Townsend (1993) that collector-filterers are more common in riffles than in pools. Pools are usually considered to be depositional areas with accumulations of fine sediment and detritus. In the study reaches, however, deposition does not occur in all pools during periods of low to normal flow. During high, bed forming flow very few pools, i f any at all, are accumulation 131 zones (Brown and Brussock, 1991). Pool substratum composition and characteristics may be a better way to describe a pool and compare it to other habitats. 4.4.3 Inter-stream Variation This study documents highly significant differences among study streams with regard to the abundance of all taxonomic groups investigated except the Chironomidae for which the effect of stream was statistically significant also but less strong (table 4-3). This, once again, may be attributed to the ubiquitous nature of the Chironomidae. Total invertebrate abundance was not significantly different among channels when TQ1 and TC43, the most productive streams, were eliminated from the A N O V A model (figure 4-1 and table 4-3). However, when the aforementioned channels were included, the difference in mean total abundance among streams was highly significant (p = .0001). The highly productive nature of TQ1 may be attributed to its substantial perennial base flow. It is recognized that aquatic fauna tends to become more rich with surface water persistence (Boulton and Suter, 1986). TC43 is a relatively wide channel (table 2-1) with a heterogeneous mixture of gravel, cobble, and boulder substrate. Channels like TC43 represent".. .the zone of change between the higher elevation, steeply-sloped mountain streams and the lower elevation, shallow-sloped alluvial river reach". As a transition point between these two geomorphic conditions, an edge effect may explain the higher biotic diversity observed in TC43 (Vannote et al, 1980; Grubaugh et al, 1996). Since TC 43 was visited only three times within a five week period, I am less familiar with the flow characteristics of this site; however, I suspect the flow regime is perennial. Adjacent forestry activities have opened the canopy and increased exposure to 132 sunlight which may have, subsequently, increased primary (Triska et al, 1983) and invertebrate production (Vannote et al, 1980; Carlson et al, 1990). Mountain streams are generally less productive than lowland channels as a result of lower temperatures, lack of rooted vegetation, and softer waters (Pennak and Van Gerpen, 1947). Therefore, macroinvertebrate densities of TC43 and T Q l , the most productive reaches in this study, presumably are lower compared to animal densities in channels further down the system. To illustrate, Grubaugh et al (1996) examined longitudinal trends in macroinvertebrate abundance along an Appalachian stream continuum and found that as catchment area increased invertebrate abundance increased linearly, increased exponentially, and decreased linearly, in bedrock cascades, cobbles, and pools, respectively. Macroinvertebrate data from streams further down the drainage networks from study reaches are not available, to my knowledge. Even i f data were available, cross study comparisons are difficult because sampling and reporting methods generally are not standardized. A l l taxonomic groups except the Enchytraeidae were more abundant in streams with perennial flow than in those with ephemeral or seasonal flow (table 4-4). Not surprisingly, mean total abundance per sample was lowest in the three ephemeral streams (MTD1, NMT4, and NMT5) compared to the others in the study (figure 4-1). In addition, ordination results indicated that community structure of ephemeral channels, as well as some seasonal streams, was distinct from that of perennial channels (figure 4-10). Enchytraeidae convincingly dominated almost every habitat in all three ephemeral streams (table 4-5) which is consistent with Williams' (1987) claim that certain taxa dominate the temporary water community. In 133 addition, the Coleoptera (beetles), Hemiptera (true bugs) and non-insect groups were more important in ephemeral streams than in other streams (table 4-5). Similarly, Coleoptera, Hemiptera and non-insect taxa were well-represented in the intermittent sections of small streams during dry periods, compared to the perennial reaches (Stehr and Branson, 1938; Harrel and Dorris, 1968; and Wright et al, 1984). The macroinvertebrate community in channels with ephemeral flow appears to be distinctive. However, statistical procedures failed to show that streams with seasonal flow regimes are different from perennial streams with respect to faunal abundance or community structure, aside from that in DB3, a seasonal channel which was grouped with ephemeral streams during ordination. The conclusions of previous studies comparing the invertebrate communities of temporary and permanent waters are inconsistent; therefore, they cannot corroborate or reject these findings. For example, some authors suggest that little faunal overlap occurs (Williams and Hynes, 1977; Wright et al, 1984) and species richness is lower in seasonal streams (Wright et al, 1984) while others suggest that fauna are taxonomically similar in seasonal and perennial streams and that species richness is not damped (Abell, 1984; Boulton and Suter, 1986; Boulton and Lake, 1988; Delucchi, 1988; Dieterich, 1992). Discrepancies between studies may be a result of site permanence, length of the dry period (Delucchi, 1988), or other biotic or abiotic factors. Inconsistent sampling intensity among channels hampered the determination and comparison of taxonomic richness; however, table 4-5 indicates that the five most abundant taxa collected from each habitat type of ephemeral channels constituted a greater proportion of the total 134 abundance compared to that of both seasonal and perennial streams. In many cases, all invertebrates collected from a certain habitat type were represented by fewer than five taxonomic groups suggesting that taxonomic richness was lower in ephemeral channels. This evidence of depressed taxonomic richness in streams with ephemeral flow compared to seasonal and perennial flow is consistent with the findings in some California and Oregon ephemeral streams (Abell, 1984; Dieterich, 1992) and with reports in Resh et al (1988), Brown and Brussock (1991), and Brussock and Brown (1991). Having eliminated TC43 and TQ1 from the A N O V A models, abundance was greatest in MRN1 for more taxonomic groups than any other stream (table 4-4). MRN1 is a spring-fed stream which was assumed to emerge from the ground as a lotic system, although, its source was not located. Chemical composition of the water in MRN1, as a rheocrene, may be different from that of the other study channels, the primary source of which is atmospheric precipitation. However, it was assumed that the water of M R N 1 possessed no adverse chemical conditions, as judged by the faunal composition. Instead, the uniform temperature, constant flow conditions, and resultant stable substrate typical of springbrooks appear to be the primary factors structuring the biotic community (Harper, 1981; Resh et al, 1988; Ward, 1992). In MRN1, moss cover on submerged cobbles, boulders, and woody debris is prevalent, a consequence of the stable substratum which promotes higher aquatic plant development compared to unstable substratum (Harper, 1981; Ward, 1992). High invertebrate abundance in MRN1 may be a result of the moss which is known to support higher animal densities than bare mineral substrates (Mackay and Kalff, 1969; Minshall, 1984). Ward (1992) lists Families Leptophlebiidae, Nemouridae, Chironomidae, and 135 Hydroptilidae, among others, as being frequently encountered near the source of well-oxygenated, non-thermal springbrooks in the northern temperate zone. A l l four families are among the five most abundant taxa in at least one habitat type in MRN1 (table 4-5). 4.4.4 Temporal Variation and Interactions This study demonstrates that six of the eight most abundant insect families (Heptageniidae, Nemouridae, Chironomidae, Baetidae, Chloroperlidae, and Simuliidae) exhibit variation among sampling dates, with respect to abundance (table 4-3), which suggests that the species present respond to environmental variables that change with the seasons, such as temperature, day length, food resources, and discharge. Fauna of seasonal streams likely are influenced most strongly by the onset of channel bed dehydration. Insect populations undergo drastic changes in spatial distribution as a result of not only mortality but also retreat into the hyporheic zone in response to spates. Because many samples collected during the autumn were done so amidst heavy rainstorms which, however, are characterisitic of the environment, abundance estimates made during the autumn may be strongly influenced by discharge (Resh, 1979). Conversely, total invertebrate abundance and the abundance of the Leptophlebiidae, the Lepidostomatidae, and both non-insect groups (Enchytraeidae and Tricladida) did not vary with sampling date indicating that emergence and/or mortality of one species is balanced by the hatching of another. Mackay and Kalff (1969) and Waters (1979) suggest that related insect species, the early instars of which are often indistinguishable, have staggered life histories. The families, therefore, may be represented throughout several seasons. 136 Size-biased sorting efficiency undoubtedly is responsible, in part, for variation in abundance in general but most notably among sampling dates (Pennak and Van Gerpen, 1947). Because certain size classes were selectively sorted, life cycle interpretation may have been influenced (Resh, 1979). For example, those species with early instars present at the time of sampling may have been too small to be captured by the 1 mm sieve; therefore, would have been excluded from the sample. In general, the significance levels of interactions are lower than those of the main effects (table 4-3) which suggests that interactions influence the benthic community less dramatically than habitat, stream, or date alone. Pennak and Van Gerpen (1947) reported that "the seasonal occurrence of maximum and minimum populations for particular forms were not concurrent from one substrate to another but widely variable". This phenomenon signifies habitat x date interactions for which no explanation was offered. Minshall and Minshall (1977) suggest that variation in a species reaction to similar conditions from one time to another is associated with different life cycle stages. Habitat x stream interactions may be attributed to the broad range of bed slope, velocity, and substratum characteristics associated with a single habitat type. Table 3-2 and figure 3-4 illustrate that the bed slope range of each habitat type is very wide. Consequently, current velocity ranges are equally large (figure 3-5). Habitat units near the limit of a range, therefore, may have been misidentified. Substratum characteristics frequently vary among streams; therefore, attributes typical of a specific habitat type in one stream may be quite different from those of another. For example, riffles in DBI were often composed of bedrock. Similarly, the boulders contained in cascades of M R N I were much smaller than the boulders of M T D 1 or T Q l . 137 Interactions also may be attributed to size-biased sorting efficiency. Sieving of very small organisms (<lmm) often is not effective when organic matter, particularly deciduous leaves and moss, is abundant because animals adhere to the detritus and do not pass through the sieve. As a result, the organisms are enumerated when the intention of sieving was to remove them from the sample. Streams, habitats, or dates with abundant organic matter may trap more small insects than those with less detritus. For example, moss was extremely abundant in M R N I , accumulations of detritus were present in interstitial spaces between gravel, cobble and boulders but absent from clear bedrock surfaces, and leaf litter was prevalent in autumn following leaf abscission and preceding strong winter currents and decomposition. Much of the above discussion regarding interactions is largely speculative and, although the explanations seem reasonable, neither the methods nor the data allow confirmation of these ideas. 138 Chapter 5: Conclusions, Limitations, and Final Comments 5.1 Conclusions 5.1.1 Fluvial Geomorphology (1) A simple distinction between alluvial and non-alluvial channels prevented many of the study reaches from being accurately categorized. Non-lithified substrate and non-alluvial lag material may be subject to very infrequent, weak fluvial sorting. The term "semi-alluvial" was introduced to describe channels with constituent material of this nature. (2) The relative proportions of stream area in geomorphic units were influenced by bed material and channel slope more significantly than channel width and entrenchment. Width and entrenchment did not distinguish channels well. (3) In general, high gradient geomorphic units (i.e. bedrock and boulder cascades) were dominant in steep, largely non-alluvial channels. Lower gradient units (i.e. riffles and rapids) were common in semi-alluvial streams with more mild slopes. Therefore, channel classes with opposing bed material and gradient designations (i.e. A2 and B3a or B3b) exhibited notable differences with respect to relative proportions of geomorphic units. (4) Amongst channel classes, relative proportion of area in riffles and rapids was distinctly variable. Inter-class variation in relative proportion of class area in boulder cascades also was apparent but less distinct than in riffles and rapids. Channel classes were not separated well on the basis of class area in bedrock cascades and chutes, presumably as a consequence of huge natural variability. Pools and falls proved to be ineffective 139 discriminators of channel classes because they are uniformly distributed among study reaches. (5) M y original hypothesis that inter-class variation in relative proportion of area in geomorphic units would be strongly linked to channel bed material, slope, entrenchment, and width requires refinement. In the group of channels investigated, inter-class variation in relative proportion of area in riffles, rapids, and boulder cascades is associated with bed material and slope. (6) Because between and among stream variability is high in small, forested channels, increased site replication may still elucidate more systematic differences between groups of channels. 5.1.2 Aquatic Ecology (1) Total abundance of invertebrates was significantly different among habitat types. Riffles were the most desnely populated habitat followed by rapids, pools, boulder cascades, chutes, and bedrock cascades. (2) Organisms belonging to families Heptageniidae, Nemouridae, Chironomidae, Leptophlebidae, Enchytraeidae, Chloroperlidae, and Lepidostomatidae and order Tricladida were most abundant in either riffles or rapids. Representatives of the families Baetidae and Simuliidae preferred bedrock cascades and chutes, respectively. (3) The community structure of riffles, rapids, pools, and boulder cascades was not distinct. The community structure of bedrock habitats (i.e. cascades and chutes), however, was fairly distinct owing to the influence of the Baetidae and Simuliidae. 140 (4) Significant inter-stream variation was apparent in the abundance of all taxonomic groups investigated. In general, animals were more abundant in streams with perennial flow regimes compared to those with seasonal or ephemeral flow. (5) The community structure of channels with ephemeral flow regimes was very distinct as a result of the preponderance of the Enchytraeidae. In addition, the Coleoptera, Hemiptera, and various non-insect groups were better represented in ephemeral channels compared to those with more persistent flow. (6) Most of the insect families investigated exhibited seasonal variation in abundance which, presumably, was associated with life cycle stages. The abundance of non-insect groups investigated did not vary significantly among the sampling periods. (7) M y original hypothesis that the spatial heterogeneity in abundance and community composition of benthic macroinvertebrates would be strongly linked to habitat variability requires refinement. The spatial heterogeneity in abundance of benthic macroinvertebrates is associated with habitat variability and flow regime. The spatial heterogeneity in community composition of benthic macroinvertebrates is linked to bedrock substratum and flow regime. 5.2 Study Limitations 5.2.1 Microhabitat Habitat often is the focus of ecological inquiry because researchers see habitat as the unit with the highest level of resolution within an ecosystem. However, unit boundaries are perceived differently depending upon the organism in question. Because different variables 141 become dominant at different levels of resolution, different levels of generalization are appropriate (Minshall, 1988). For example, the distinction between habitat types provides boundaries relevant to some questions in aquatic ecology, such as fish dynamics (Pringle et al, 1988); therefore, characterisation of channel integrity based on habitat features is accepted widely (Bisson and Montgomery, 1996). Measurements of channel units, however, integrate information at a scale that is much too large to detect temporal and spatial variation in microhabitat which is the level of resolution appropriate for organisms as small as invertebrates (Pringle et al, 1988). Most habitat types in the study streams are mosaics of different substratum types (i.e. bedrock, boulders, cobbles, gravel, sand, silt, and organic accumulations) that may differentially influence both abiotic and biotic stream processes. Microhabitat may cause similar individuals to aggregate when one patch is favored over another. On the other hand, non-aggregated patterns of distribution have been reported in relatively uniform or homogeneous substrate (Resh, 1979). The number of aggregations occurring in any one habitat may depend, in part, upon the number of potential microhabitats within that habitat (Mackay and Kalff, 1969). To illustrate, consider bedrock cascades and chutes: bedrock substratum and swift currents prevail in both habitat types; however, the average bed slope of chutes is lower. Consequently, stable alluvium is common on chutes which provides interstitial spaces and microhabitat variety. This study was an attempt to differentiate between habitat types on the basis of their benthic macroinvertebrate communities. The occupants respond to, therefore, are indicative of conditions created by the various habitat types. Microhabitat was not considered formally. Patch diversity within a habitat type may have decreased inter-habitat variability and may have confounded trends that the study aimed to expose. Patch-to-patch variability in lotic systems, however, should be viewed as ancillary information, rather than statistical noise to be overcome with adequate sample size (Pringle et al, 1988). But without utilizing proper sampling methods, patch information is irretrievable. To overcome this, variability among the benthic communities of various microhabitat types may be investigated in a manner similar to habitat type investigations. In addition, habitat structure may be characterised on the basis of the patches of which it is comprised. It may be that specific microhabitats are innately dominant in particular habitat types. 5.2.2 Biotic Forces Historically, the prevailing dogma in aquatic ecology is that the distribution of benthic fauna in lotic systems primarily is a result of physical factors such as disturbance, substrate and current velocity (Frissell et al, 1986; Barmuta, 1989; Dunson and Travis, 1991; Death, 1995; and Crowl et al, 1997). More recently, ecologists have abandoned the notion that physical processes alone govern community organization and believe that in order to understand community structure, biotic forces, such as competition and predation, must be investigated in conjunction with abiotic factors (Dunson and Travis, 1991). In this study, differences in macroinvertebrate communities among habitats were attributed to abiotic factors only. In order to attribute differences to both, systems must be manipulated. Such an endeavor is beyond the scope of this project and, indeed, is rarely attempted owing to the difficulty of doing so in lotic environments (Dunson and Travis, 1991). 143 5.2.3 Invertebrate Measures What taxa are present? How are the taxa different? To what can the variation be attributed? These are fundamental questions raised when the spatial variability of biota is investigated. In this study, the aforementioned questions were explained on the basis of relative invertebrate abundance; however, numerous other measures, such as animal biomass, secondary production, and diversity, also are valuable. Results, hence conclusions, may vary depending on the measures upon which spatial variability is based (Wohl et al, 1995; Grubaugh et al, 1996). Wohl et al (1995) criticise the use of abundance measures to evaluate the associations between species distributions and environmental parameters because "abundances alone tend to under- or over estimate the roles of organisms in stream ecosystems". They suggest that measures of invertebrate biomass and production are required to draw accurate conclusions regarding inter-habitat variation. One cannot assume that interpretations of benthic community structure based on abundance data are definitive conclusions (Grubaugh et al, 1996). Other authors have used both invertebrate abundance and biomass measures to assess inter-habitat variation (Angradi, 1996; Grubaugh et al, 1996). A single measure cannot provide a complete view of the dynamics of a system because there is an inherent bias associated with each measure. Thus, a second parameter is advantageous. If multiple parameters substantiate one another, then conclusions are more likely to be accurate. However, determination of biomass through either volume estimates, using length and width measurements, or direct mass measurements requires considerable effort in the laboratory. Similarly, benthic samples required to obtain secondary production estimates are far more extensive than I was able to collect. For example, Benke et al (1984) calculated invertebrate production based on 108 separate samples from each site which were collected at two week intervals over a four month period. In another study, production calculations were based on 130 samples per site (Krueger and Martin referenced in Benke et al, 1984). In addition, secondary production estimates require consideration of biomass turnover rates of taxa for which life history information is required. Species level identifications, therefore, are required. Likewise, to be meaningful, biodiversity measures require species level information. Species can be considered discrete taxonomic units but higher taxonomic groups can not (Templeton, 1989; Davis, 1996). Therefore, assessing biodiversity with higher level groups is inappropriate owing to the huge variability in form and function of individuals belonging to even moderately sized families (Danks, 1996; Milhuc, 1997). In addition, species names allow diversity information to be associated with each taxonomic entity for future reference or for detailed comparisons. Parallel to secondary production estimates, diversity estimates require intense sampling because of temporal variability associated with development of populations and natural variation in the occurrence and abundance of organisms. Therefore, i f samples are obtained over a period of time that is too short, then they likely will provide only a small random selection of the actual diversity. Moreover, samples must be collected at frequent and regular intervals so that population trends can be detected (Danks, 1996). Measures of macroinvertebrate biomass, production, and diversity, therefore, were beyond the scope of this study. 5.3 Final Comments 5.3.1 Integration of Ecological and Geomorphological Results Information regarding inter-habitat variation in stream benthic communities and important factors influencing community structure and function may be gathered at the habitat scale and used to extrapolate variation in benthic communities among stream reaches (Minshall, 1988; Angradi, 1996). Insight into the dynamics of a specific habitat type, together with knowledge of the relative proportion of that habitat in a specific channel class may provide insight into the dynamics of that class. For example, A2 channels had high proportions of riffles and rapids; total invertebrate abundance results indicate these are the most productive habitat types. Therefore, A2 channels plausibly are productive also. Similarly, class B3b channels have high proportions of cascades which are relatively unproductive habitats; therefore, B3b channels undoubtedly are relatively unproductive. However, how far can we reasonably extrapolate information gathered at specific sites (Frissell et al, 1986)? The occurrence and detailed distribution of invertebrates in lotic systems may be a result of a number of interrelated abiotic factors (Hynes, 1970; Minshall, 1984; Ormerod, 1988; Ward, 1992), the form and structure of the physical habitat being just one. For instance, the results in chapter four suggest that water permanence is a major determinant of community structure and animal abundance. For lotic habitats subject to prolonged dry periods, other physical variables, such as habitat type, appear to be of secondary importance. To illustrate, NMT4 and NMT5 are A2 channels with 82% and 79%, respectively, of stream area in riffles and rapids. The N M T channels also are ephemeral; therefore, regardless of the habitat of which they are comprised, their biotas are depressed 146 severely. As well, T Q l is a B3b channel with 59% of its area in boulder cascades; however, it was the second most productive stream in the study, which was assumed to be a consequence of substantial perennial base flow (figure 5-1). 5.3.2 Ecological Significance of the CSP Hydroriparian Classification Because the basis of the CSP hydroriparian classification is physical, Chan-MacLeod (1996) claimed that the classification does not account for attendant biological properties and ecological dynamics of hydroriparian ecosystems. However, since the form, structure, and dynamics of a channel determine, to a large extent, the structure, operation, and other aspects of the organization and development of aquatic communities, the physical environment may provide insight into the ecological dynamics of a stream. For example, the results of this study showed that flow regime was a classification criterion of primary ecological significance, for the group of channels investigated. The biota of channels with ephemeral flow regimes was damped severely compared to the biota of streams with seasonal and perennial flow regimes. Flow regime currently is a criterion included in the CSP hydroriparian classification, albeit, a minor one (figure 2-2). Perhaps this ecologically significant criterion deserves greater emphasis. Channel-bed slope, which approximates current velocity, and channel-material grain size are physical measures used to characterise habitat units and which, like flow regime, provide insight into the ecological dynamics of aquatic environments. Within any small segment of stream, current velocity and substratum characteristics are likely to exert substantial influence on animal numbers (Minshall and Minshall, 1977). These factors exhibit considerable 147 250 200 " 150 -100 -50 -X I I | perennial I seasonal | | ephemeral I A2 B2a B3a B3b Figure 5-1: Mean invertebrate abundance (+ 1 SE) expressed as animals per sample for each study stream pooled across habitat types and sampling dates. 148 heterogeneity at the habitat scale (Hawkins et al, 1993) which is a feasible level of spatial resolution upon which to focus management initiatives. However, in order to gather information at specific sites, time consuming field visits almost certainly would be required. To avoid onerous field work, physical measures such as substratum characteristics intentionally were excluded from the current classification system. The reach scale is a more pragmatic level of spatial resolution upon which to focus management activities because information may be accessed from terrain maps or aerial photographs (CSP, 1995). However, measures at the reach scale do not account for spatial and temporal variability at higher levels of spatial resolution. For instance, measures at the reach scale integrate information apparent at the habitat level of resolution. Similarly, measures at the habitat scale integrate microhabitat information, which is the level of resolution most appropriate for the examination of minute organisms (Pringle, 1988). Patterns in macroinvertebrate organization that may be evident at the microhabitat scale may become less distinct at the habitat scale and completely ambiguous at the reach scale. Therefore, systematic variation in macroinvertebrate abundance and community organization, at the reach scale, is likely to be inconspicuous. In light of the ubiquity of macroinvertebrates among reaches with seasonal and perennial flow regimes, the ecological differences among such reaches are of little consequence. 149 References Abell, D.L. 1984. Benthic invertebrates of some California intermittent stream Pages 46-60 in S. Jain and P. Moyle (eds.), Vernal Pools and Intermittent Streams. University of California, Davis, Institute of Ecology Publication No. 28. Abrahams, A.D. , G. L i , and J.F. Atkinson. 1995. Step pool streams: Adjustment to maximum flow resistance. Water Resources Research. 31: 2593-2602. Andrews, E.D. and G. Parker. 1987. Formation of a coarse layer as the response to gravel mobility. Pages 269-325 in C.R. Thorne, J.C. Bathurst, and R.D. Hey (eds.), Sediment Transport in Gravel-Bed Rivers, Toronto, John Wiley & Sons. 995 pp. Angradi, T.R. 1996. Inter-habitat variation in benthic community structure, function, and organic matter storage in 3 Appalation headwater streams. J. N. Am. Benthol. Soc. 15(1): 42-63. Baker, V.R. 1977. Stream-channel response to floods, with examples from central Texas. Geological Society of America Bulletin, 88: 1057-1071. Baker, V.R. and D.F. Ritter. 1975. Competence of rivers to transport coarse bedload material. Geological Society of America Bulletin, 86: 975-978. Barkham, J.P. and J .M. Norris. 1970. Multivariate procedures in an investigation of vegetation and soil relations of two beach woodlands, Cotswold Hills, England. Ecology, 51: 630-639. Barmuta, L .A . 1989. Habitat patchiness and macrobenthic community structure in an upland stream in temperate Victoria, Australia. Freshwater Biology, 21: 223-236. Baumann, R.W. 1987. Order Plecoptera. Pages 186-195 in F.W. Stehr (ed.), Immature Insects. Dubuque, Iowa, Kendall/Hunt Publishing Company. 754 pp. Benda, L . 1990. The influence of debris flows on channels and valley floors in the Oregon Coast range. Earth Surface Processes and Landforms. 15: 457-466. Benke, A .C . , T.C. Van Ardsall, Jr., D . M . Gillespie, and F.K. Parrish. 1984. Invertebrate Productivity is a subtropical blackwater river: the importance of habitat and life history. Ecological Monographs, 54(1): 25-63. Beschta, R.L. and W.S. Platts. 1986. Morphological features of small streams: significance and function. Water Resources Bulletin, 22(3): 369-379. 150 Bisson, P.A. and D.R. Montgomery. 1996. Valley segments, stream reaches, and channel units. Pages 23-52 in F.R. Hauer and G.A. Lamberti (eds.), Methods in Stream Ecology, Toronto, Academic Press. 674 pp. Bisson, P.A., J.L. Nielsen, R.A. Palmason, and L.E. Grove. 1981. A System of naming habitat types in small streams, with examples of habitat utilization by salmonids during low stream flow, in Armentrout, N.B. , ed., Acquisition and utilization of aquatic habitat inventory information: Portland, Oregon. American Fisheries Society, Western Division, pp. 291-298. Boulton, A J . , and P.J. Suter. 1986. Ecology of temporary streams - an Australian perspective. Pages 313-327 in P. De Deckker, and W.D. Williams (eds.), Limnology in Australia, Melbourne and Dordrecht, CSORO/DR. W. Junk. Boulton, A.J . , and P.S. Lake. 1988. Australian temporary streams-Some ecological characteristics. Verh. Internal Verein. Limnol. 23: 1380-1383. Bournaud, M . , B . Cellot, P. Richoux, and A . Berrahou. 1996. Macroinvertebrate community structure and environmental characteristics along a large river: congruity of patterns for identification to species or families. J. N. Am. Benthol. Soc. 15(2): 232-253. Brinkhurst, R.O. 1986. Guide to the Freshwater Aquatic Microdile Oligochaetes of North America. Can. Spec. Publ. Fish. Aquat. Sci. 84: 259 pp. Brown, A . V . , and P.P. Brussock. 1991. Comparisons of benthic invertebrates between riffles and pools. Hydrobiologia. 220: 99-108. Brussock, P.P., and A . V . Brown. 1991. Riffle-pool geomorphology disrupts longitudinal patterns of stream benthos. Hydrobiologia. 220: 109-117. Bunnell, P., S. Rautio, C. Fletcher, and A. Van Woudenberg. 1995. Problem analysis of integrated resource management of riparian areas in British Columbia. B.C. Min. For. And B.C. Min. Environ., Lands and Parks, Victoria, B.C. Work. Pap. 11/1995. Carlson, J.Y., C.W. Andrus, and H.A. Froehlich. 1990. Woody debris, channel features, and macroinvertebrates of streams with logged and undisturbed riparian timber in northeastern Oregon, U.S.A. Can. J. Fish. Aquat. Sci. 47: 1103-1111. Chan-McLeod, A . 1996 (unpublished report). Plants and amphibians in the Clayoquot Sound area, progress report prepared for Long Beach Model Forest Society, April, 1996, University of British Columbia, Vancouver, B.C. Chen, J., J.F. Franklin, and T.A. Spies. 1995. Growing season and microclimate gradients from clearcut edges into old-growth Douglas-fir forests. Ecological Applications, 5(1): 74-86. 151 Chin, A. 1989. Step pools in stream channels. Progress in Physical Geography. 13: 391-407. Church, M . 1984. On experimental method in geomorphology. Pages 562 - 579 in T.P. Burt and D.E. Walling (eds.), Catchment Experiments in Fluvial Geomorphology, Norwich, Geo Books. Church, M . 1992. Channel morphology and typology. Pages 126-143 in P. Calow and G.E. Petts (eds.), The Rivers Handbook Volume I, Oxford, Blackwell. Church, M . 1996 (unpublished report).The hydrological recommendations of the Clayoquot Scientific Panel: background and basis, progress report prepared for Long Beach Model Forest Society, April, 1996, University of British Columbia, Department of Geography, Vancouver, B.C. Clifford, H.F. 1991. Aquatic Invertebrates of Alberta. Edmonton, Alberta, The University of Alberta Press. 538 pp. Cochran, W.G. and G . M . Cox. 1950. Experimental Designs. New York, John Wiley & Sons, Inc. 454 pp. Connell, J.H. 1978. Diversity in the tropical raiforests and coral reefs. Science, 199: 1302-1310. Corkum, L.D. 1989. Patterns of benthic invertebrate assemblages in rivers of northwestern North America. Freshwater Biology. 21: 191-205. Crowl, T.A., C.R. Townsend, N . Bouwas, and H . Thomas. 1997. Scales and causes of patchiness in stream invertebrate assemblages: top-down predator effects? J. N. Am. Benthol. Soc, 16(1): 277-285. Cummins, K.W. , and G.H. Lauff. 1969. The influence of substrate particle size on the microdistribution of stream macrobenthos. Hydrobiologia. 34: 145-181. Danks, H.V. 1996. How to assess insect biodiversity without wasting your time. Biological Survey of Canada (Terrestrial Arthropods) Document Series No. 5, Ottawa, Biological Survey of Canada. 20 pp. Davis, J.I. 1996. Phylogenetics, molecular variation, and species concepts. Bioscience, 46(7): 502-511. Death, R.G. 1995. Spatial patterns in benthic invertebrate community structure: products of habitat stability or are they habitat specific? Freshwater Biology. 33: 455-467. 152 Delucchi, C M . 1988. Comparison of community structure among streams with different temporal flow regimes. Can. J. Zool. 66: 579-586. Dieterich, M . 1992. Insect community composition and physiochemical processes in summer dry streams of western Oregon. Ph.D. thesis. Oregon State University. 191 pp. Doledec, S. and D. Chessel. 1994. Co-inertia analysis: an alternative method for studying species-environment relationships. Freshwater Biology, 31: 277-294. Dunne, T. and L .B . Leopold. 1978. Water in Environmental Planning. New York, N Y , W.H. Freeman and Company. 818 p. Dunson W.A., and J. Travis. 1991. The role of abiotic factors in community organization. The American Naturalist. 138(5): 1067-1091. Edmunds, G.F. and R.K. Allen. 1987. Order Ephemeroptera. Pages 75-94 in F.W. Stehr (ed.), Immature Insects. Dubuque, Iowa, Kendall/Hunt Publishing Company. 754 pp. Edmunds, G.F., S.L. Jensen, and L. Berner. 1976. The Mayflies of North and Central America. Minneapolis, University of Minnesota Press. 330 pp. Flueck, J. A . 1978. The role of statistics in weather modification experiments. Atmosphere-Ocean, 16: 60-80. Frissell, C.A., W.J. Liss, C.E. Warren, and M.D. Hurley. 1986. A hierarchical framework for stream habitat classification: Viewing streams in a watershed context. Environmental Management. 10(2): 199-214. Garcia, J .C 1985. A method for assessing the value of stream corridors to fish and wildlife resources. Pages 335-338 in R.R. Johnson, C D . Zeibell, D.R. Patton, P.F. Pfolliott, and R.H. Hamre (eds.), Riparian Ecosystems and Their Management: Reconciling Conflicting Uses. United States Forest Service, General Technical Report M-120, Rocky Mountain Forest and Range Experimental Station, Fort Collins, Colorado. Gore, J.A. 1996. Discharge measurements and streamflow analysis. Pages 53-74 in F.R. Hauer and G.A. Lamberti (eds.), Methods in Stream Ecology, Toronto, Academic Press. 674 pp. Goudie, A. , B.W. Atkinson, K.J . Gregory, I.G. Simmons, D.R. Stoddart, and D. Sugden (eds.). 1994. The Encyclopedic Dictionary of Physical Geography. Oxford, U K , Basil Blackwell Inc. 611 pp. 153 Grant, Gordon E., F J . Swanson, and M . G . Wolman. 1990. Pattern and origin of stepped-bed morphology in high gradient streams, Western Cascades, Oregon. Geological Society of America Bulletin. 102: 340-352. Gregory, S.V., F.J. Swanson, W.A. McKee, and K.W. Cummins. 1991. A n ecosystem perspective of riparian zones: Focus on links between land and water. Bioscience. 41 (8): 540-551. Grubaugh, J.W., J.B. Wallace, and E.S. Houston. 1996. Longitudinal changes of macroinvertebrate communities along an Appalachian stream continuum. Can. J. Fish. Aquat. Sci. 53: 896-909. Harcombe, A.P. and E.T. Oswald. 1990. Vegetation resources of Vancouver Island Volume 1: Forest Zonation. M O E Technical Report 27. Ministry of Environment, Victoria, B.C. Harmon, M.E . and twelve other authors. 1986. Ecology of coarse woody debris in temperate ecosystems. Pages 133-302 in A . Macfadyen and E.D. Ford (eds.) Advances in Ecological Research. Harper, P.P. 1981. Ecology of streams at high latitudes. Pages 313-338 in M . A . Lock and D.D. Williams (eds.), Perspectives in Running Water Ecology. New York, Plenum Press. 430 pp. Harrel, R.C. and T.C. Dorris. 1968. Stream order, morphometry, physio-chemical conditions, and community structure of benthic macroinvertebrates in an intermittent stream system. Am. Midi. Nat., 80: 220-251. Harris, Richard R. 1988. Associations between stream valley geomorphology and riparian vegetation as a basis for landscape analysis in the Eastern Sierra Nevada, California, USA. Environmental Management. 12(2): 219-228. Hauer, F.R. and V . H . Resh. 1996. Benthic Macroinvertebrates. Pages 339-369 in F.R. Hauer and G.A. Lamberti (eds.), Methods in Stream Ecology, Toronto, Academic Press. 674 pp. Hawkins, C P . , J.L Kershner, P.A. Bisson, M.D. Bryant, L . M . Decker, S.V. Gregory, D.A McCullough, C.K. Overton, G.H. Reeves, R.J. Steedman, and M . K . Young. 1993. A hierarchical approach to classifying stream habitat features. Fisheries. 18 (6): 3-12. Heede, B .H. 1981. Dynamics of selected mountain streams in the western United States of America. Zeitschrift fur Geomorphologie, NF25: 17-32. 154 Hetherington, E. 1987. Hydrology and logging in the Carnation Creek watershed - what have we learned? Pages 11- 15 in T.W. Chamberlin (ed.), Proceedings of the workshop: Applying 15 years of Carnation Creek results. Nanaimo, British Columbia, Carnation Creek Steering Committee. 239 pp. Hicks, C.R. 1993. Fundamental Concepts in the Design of Experiments. Toronto, Saunders College Publishing. 509 pp. Hogan, D.L. and M . Church. 1989. Hydraulic geometry in small coastal streams: progress toward quantification of salmonid habitat. Canadian Journal of Fisheries and Aquatic Sciences. 46: 844-852. Horton, R.E. 1945. Erosional development of streams and their drainage basins: hydrophysical approach to quantitative morphology. Bulletin of the Geological Society of America, 56: 275-370. Howard, A .D . 1987. Modeling fluvial systems: Rock-, gravel-, and sand-bed channels. Pages 69-94 in K . Richards (ed.), River Channels: Environment and Process. Basil Blackwell Inc. 391 p. Huryn, A.D. , and J.B. Wallace. 1987. Local geomorphology as a determinant of macro faunal production in a mountain stream. Ecology. 68(6): 1932-1942. Hynes, H.B.N. 1970. The Ecology of Running Waters. Toronto, University of Toronto Press. 555 pp. Jarrett, R.D. 1984. Hydraulics of high-gradient streams. Journal of Hydraulic Engineering, 110: 1519-1539. Jongman, R.H.G., C.J.F. ter Braak, and O.F.R. Van Tongeren (eds.). 1995. Data Analysis in Community and Landscape Ecology. New York, Press Syndicate of the University of Cambridge. 299 pp. Jungen, J.R. and T. Lewis. 1978. The Coast Mountains and Islands. Pages 101-120 in K.W.G. Valentine, P.N. Sprout, T.E. Baker, and L . M . Lavkulich (eds.), The Soil landscapes of British Columbia. Resource Analysis Branch, Ministry of Environment, Victoria, B.C. Keller, E.A. and W.N. Melhorn. 1978. Rhythmic spacing and origin of pools and riffles. Geological Society of America Bulletin, 89: 723-730. Kitano, Y . (ed.) 1975. Geochemistry of Water. Stroudsburg, Pennsylvania, Dowden, Hutchinson & Ross, Inc. 455 pp. 155 Knighton, A .D . 1987. River channel adjustment - The downstream dimension. Pages 95-128 in K . Richards (ed.), River Channels: Environment and Process. New York, N Y , Basil Blackwell Inc. 391 p. Krajina, V J . 1965. Biogeclimatic zones and biogeocoenoses of British Columbia. Ecol. West. N. Amer. 1: 1-17. Larson, J.L. 1994. Interpreting Nature: The Science of Living Form From Linnaeus to Kant. Baltimore, M D , The Johns Hopkins University Press. 9 pp. Leopold, L .B . , M . G . Wolman, and J.P. Miller. 1964. Fluvial Processes in Geomorphology. San Francisco, California, W.H. Freeman and Co., 522 p. Logan, P., and M.P. Brooker. 1983. The macroinvertebrate faunas of riffles and pools. Water Res. 17(3): 269-270. Lotspeich, F.B. and W.S. Platts. 1982. An integrated land-aquatic classification system. N. Am. J. Fish. Manag., 2: 138-149. Mackay, R.J. and J. Kalff. 1969. Seasonal variation in standing crop and species diversity of insect communities in a small Quebec stream. Ecology, 50: 101-109. MacKenzie, W. and A . Banner. 1995 (unpublished report). A wetland riparian ecosystem classification (WREC) for British Columbia, working draft. B.C. Ministry of Forests Meidinger, D. and J. Pojar (eds.). 1991. Ecosystems of British Columbia. B.C. Ministry of Forests, Victoria, B.C. Special Report Series No. 6. Merritt, R.W. and K.W. Cummins. 1978. An Introduction to the Aquatic Insects of North America. Dubuque, Iowa, Kendall/Hunt Publishing Company. 441 pp. 1996. An Introduction to the Aquatic Insects of North America. Dubuque, Iowa, Kendall/Hunt Publishing Company (third edition) 862 pp. Milhuc, T.B. 1997. The functional trophic role of lotic primary consumers: generalist versus specialist strategies. Freshwater Biology, 37: 455-462. Minshall, G.W. 1968. Community dynamics of the benthic fauna in a woodland springbrook. Hydrobiologia, 32: 305-339. Minshall, G.W. 1984. Aquatic insect-substratum relationships. Pages 358-399 in V . H . Resh, and D . M . Rosenburg (eds.) Ecology of Aquatic Insects. Praegor, New York. Minshall, G.W. 1988. Stream ecosystem theory: a global perspective. J. N. Am. Benthol. Soc.,l(4): 263-288. 156 Minshall, G.W. and J.N. Minshall. 1977. Microdistribution of benthic invertebrates in a Rocky Mountain (U.S.A.) stream. Hydrobiologia, 55(3): 231-249. Montgomery, D.L. and J.M. Buffington. 1997. Channel-reach morphology in mountain drainage basins. Geological Society of America Bulletin, 109(5): 596-611. Montgomery, D.R., T.B. Abbe, J .M. Buffington, N.P. Peerson, K . M . Schmidt, and J.D. Stock. 1996. Distribution of bedrock and alluvial channels in forested mountain drainage basins. Nature. 381: 587-589. Muller, R.A. and T .M. Oberlander. 1984. Physical Geography Today A Portrait of a Planet, Third Edition. New York, N Y , Random House, Inc. 591 p. Naiman, R.J. 1991. Influence of forests on streams. Pages 151-155 in Forest ecology. McGraw Hi l l Yearbook. Naiman, R.J., D.G. Lonzarich, T.J. Beechie, and S.C. Ralph. 1992. General principles of classification and the assessment of conservation potential in rivers. Pages 93-124 in P.J. Boon, P. Calow, and G.E. Petts (eds.), River Conservation and Management, John Wiley & Sons Ltd. O'Neill, M.P. and A.D Abrahams. 1984. Objective identification of pools and riffles. Water Resources Research, 20(7): 921-926. Oikos Ecological Services Ltd. And T. Johnson & Associates. 1996. Riparian Assessment and Prescription Procedures Field Guide (Draft #3). Watershed Restoration Program, B.C. Ministry of Environment, Lands and Parks. Ormerod, S.J. 1988. The micro-distribution of aquatic macroinvertebrates in the Wye river system: the result of abiotic or biotic factors? Freshwater Biology, 20: 241-247. Pennak, R.W. 1978. Fresh-water Invertebrates of the United States (second edition). Toronto, A Wiley Interscience Publication. 803 pp. Pennak, R.W. 1989. Fresh-water Invertebrates of the United States, 3 r d Edition: Protozoa to Mollusca. Toronto, John Wiley & Sons, Inc. 628 pp. Pennak, R.W. and E.D. Van Gerpen. 1947. Bottom fauna production and physical nature of the substrate in a northern Colorado trout stream. Ecology, 28: 42-48. Platts, W.S. 1980. A plea for fishery habitat classification. Fisheries, 5(1): 1-6. Pojar, J. and A . MacKinnon (eds.). 1994 Plants of Coastal British Columbia Including Washington, Oregon & Alaska. Vancouver, B.C. Ministry of Forests and Lone Pine Publishing. 527 pp. 157 Pringle, C M . , R.J. Naiman, G. Bretschko, J.R. Karr, M.W. Oswood, J.R. Webster, R.L. Welcomme, and M.J. Winterbourn. 1988. Patch dynamics in lotic systems: the stream as a mosaic. J. N. Am. Benthol. Soc, 7(4): 503-524. Province of British Columbia. Ministry of Forests and Ministry of Environ., Lands, and Parks Riparian Management Area Guidebook. Victoria: 1995. Ralph, S .C, G . C Pool, L . L . Conquest, and R.J. Naiman. 1994. Stream channel morphology and woody debris in logged and unlogged basins of western Washington. Can. J. Fish. Aquat. Sci., 51: 37-51. Resh, V . H . 1979. Sampling variability and life history features: basic considerations in the design of aquatic insect studies. J. Fish. Res. Board Can., 36: 290-311. Resh, V . H . , A . V . Brown, A.P. Covich, M.E . Gurtz, H.W. L i , G.W. Minshall, S.R. Reice, A . L . Sheldon, J.B. Wallace, and R . C Wissmar. 1988. The role of disturbance in stream ecology. J. N. Am. Benthol. Soc. 7 (4): 433-455. Richards, K . 1976a. Channel width and the riffle-pool sequence. Geological Society of , America Bulletin, 87: 883-890. 1976b. The morphology of riffle-pool sequences. Earth Surface Processes, 1: 71-88. Rood, K. , and M . Church. 1994. Modified freeze-core techniques for sampling the permanently wetted streambed. North American Journal of Fisheries Management. 14: 852-861. Roper, Brett B. , and D.L. Scarnecchia. 1995. Observer variability in classifying habitat types in stream surveys. North American Journal of Fisheries Management. 15: 49-53. Rosgen, D.L. 1996. A classification of natural rivers: reply to comments by J.R. Miller and J.B. Ritter. Catena, 27: 301-307. Ryder, J .M. 1978. Geology, Landforms and Surficial Materials. Pages 11-33 in K .W.G. Valentine, P.N. Sprout, T.E. Baker, and L . M . Lavkulich (eds.), The Soil landscapes of British Columbia. Resource Analysis Branch, Ministry of Environment, Victoria, B. C. SAS Institute Inc. 1990. SAS/STAT user's guide, 4 t h ed. A. Allen (ed.), SAS Institute Inc., North Carolina. Scarsbrook, M.R. and C R . Townsend. 1993. Stream community structure in relation to spatial and temporal variation: a habitat templet study of two contrasting New Zealand streams. Freshwater Biology, 29: 395-410. 158 Schaefer, D.G. 1978. Climate. Pages 3-10 in K .W.G. Valentine, P.N. Sprout, T.E. Baker, and L . M . Lavkulich (eds.), The Soil landscapes of British Columbia. Resource Analysis Branch, Ministry of Environment, Victoria, B.C. Schumm, S.A. 1977. The Fluvial System. Toronto, John Wiley & Sons, Inc. 338 p. Snedecor, G.W. and W.G. Cochran. 1967. Statistical Methods. Ames, Iowa, The Iowa State University Press, 593 p. Stanford, J. A . and J.V. Ward. 1983. Insect species diversity as a function of environment variability and disturbance in stream systems. Pages 265-278 in J.R. Barnes and G.W. Minshall (eds.), Stream Ecology - Application and Testing of General Ecological Theory, New York, Plenum Press. 399 pp. Statsoft, Inc. 1994. Statistica Volume I: General Conventions and Statistics I. Statsoft, Inc., Tulsa, Oklahoma. 1718 pp. Statsoft, Inc. 1994. Statistica Volume III: Statistics II. Statsoft, Inc., Tulsa, Oklahoma. 3958 pp. Stehr, W.C. and J.W. Branson. 1938. An ecological study of an intermittent stream. Ecology, 19: 294-310. Sullivan, K . 1986. Hydraulics and fish habitat in relation to channel morphology. Ph.D. dissertation, The Johns Hopkins University, Baltimore. Sullivan, K. , T.E. Lisle, C.A. Dolloff, G.E. Grant, and L . M . Reid. 1987. Stream channels: The linking between forests and fishes. Pages 39-97 in E.O Salo and T.W. Cundy (eds.), Streamside Management: Forestry and Fishery Interactions. Seattle, Washington, The Institute of Forest Resources, University of Washington. 471 p. Sweeney, B.W. 1984. Factors influencing life-history patterns of aquatic insects. Pages 56-100 in V . H . Resh, and D . M . Rosenburg (eds.) Ecology of Aquatic Insects. Praegor, New York. 625 pp. Templeton, A.R. 1989. The meaning of species and speciation: a genetic perspective. Pages 3-27 in D. Otte and J.A. Endler (eds.), Speciation and its Consequences, Massachusetts, Sinauer. The Scientific Panel for Sustainable Forest Practices in Clayoquot Sound. 1995. Sustainable Ecosystem Management in Clayoquot Sound: Planning and Practices. Report 5. 296 pp. 159 Thome, C.R. 1990. Effects of vegetation on riverbank erosion and stability. Pages 125-144 in J.B. Thorne (ed.), Vegetation and Erosion. West Sussex, England, John Wiley and Sons. Thorp, J.A. and A.P. Covich (eds.). 1991. Ecology and Classification of North American Freshwater Invertebrates. Toronto, Academic Press, Inc. 911 pp. Trewartha, G.T. and L . H . Horn. 1980. An Introduction to Climate (Fifth Edition). Toronto, McGraw-Hill Book Company. 416 pp. Triska, F.J. and K . Cromack, Jr. 1980. The role of wood debris in forest streams. Pages 171-190 in R .H. Waring (ed.), Forests: Fresh Perspectives From Ecosystem Analysis, Proceedings of the Fortieth Annual Biology Colloquium, Oregon State University, Corvallis. Triska, F.J., V . C . Kennedy, R.J. Avanzino, and B.N. Reilly. 1983. Effect of simulated canopy cover on regulation of nitrate uptake and primary production by natural periphyton assemblages. Pages 129-159 in T.D. Fontaine and S.M. Bartell (eds.), Dynamics of Lotic Ecosystems. Ann Arbor, Michigan, Ann Arbor Science. 494 pp. Vannote, Robin L. , W.G. Minshall, K.W. Cummins, J.R. Sedell, and C.E. Cushing. 1980. The river continuum concept. Can. J. Fish. Aquat. Sci. 37: 130-137. Ward, J.V. 1975. Bottom fauna-substrate relationships in a northern Colorado trout stream: 1945 and 1971. Ecology. 56: 1429-1434. Ward, J.V. 1992. Aquatic Insect Ecology 1. Biology and Habitat. Toronto, John Wiley & Sons, Inc. 438 pp. Ward, J.V. and J.A. Stanford, 1983. The intermediate-disturbance hypothesis: an explanation for biotic diversity patterns in lotic ecosystems. Pages 347-356 in T.D. Fontaine III and S.M. Bartell (eds.), The Dynamics of Lotic Ecosystems, Anne Arbor, Michigan, Anne Arbor Science Publishers. 494 pp. Waters, T.F. 1979. Benthic life histories: summary and future needs. J. Fish. Res. Board. Can., 36: 342-345. Webster, J.R. and T.P. Ehrman. 1996. Solute Dynamics. Pages 145-160 in F.R. Hauer and G.A. Lamberti (eds.), Methods in Stream Ecology. Toronto, Academic Press. 674 pp. Weinstock, J. (ed.). 1985. Contemporary Perspectives on Linnaeus. Lanham, M D , University Press of America. PP v. 160 Whiting, P.J., and J.B. Bradley. 1993. A process-based classification system for headwater streams. Earth Surface Processes and Landforms. 18: 603-612. Whittaker, J.G. and M.N.R. Jaeggi. 1982. Origin of step-pool systems in mountain streams. Proceeding s of the American Society of Civi l Engineers, Journal of Hydraulics Division, 108: 774-796. Wiggins, G.B. 1987. Order Trichoptera. Pages 253-287 in F.W. Stehr (ed.), Immature Insects. Dubuque, Iowa, Kendall/Hunt Publishing Company. 754 pp. Wiggins, G.B. 1996. Larvae of the North American Caddisfly Genera (Trichoptera) (second edition). Toronto, University of Toronto Press. 457 pp. Williams, D.D. 1978. Substrate size selection by stream invertebrates and the influence of sand. Limnol. Oceanogr., 23(5): 1030-1033. Williams, D.D. 1980. Some relationships between stream benthos and substrate heterogeneity. Limnol. Oceanogr., 25(1): 166-172. Williams, D.D. 1987. The Ecology of Temporary Waters. Portland, Oregon, Timber Press. 205 pp. Williams, D.D., and H.B.N. Hynes. 1977. The ecology of temporary streams II. General remarks on temporary streams. Int. Revue ges. Hydrobiol., 62: 53-61. Wohl, D.L., J.B. Wallace, and J.L. Meyer. 1995. Benthic macroinvertebrate community structure, function and production with respect to habitat type, reach and drainage basin in the southern Appalations (U.S .A. ) . Freshwater Biology. 34: 447-464. Wood-Smith, R.D., and J.M. Buffington. 1996. Multivariate geomorphic analysis of forest streams: Implications for assessment of land use impacts on channel condition. Earth Surface Processes and Landforms. 21: 377-393. Wright, J.F., P.D. Hiley, D.A. Cooling, A .C . Cameron, M.E. Wingham, and A.D. Berrie. 1984. The invertebrate fauna of a small chalk stream in Berkshire, England, and the effect of intermittent flow. Arch. Hydrobiol., 99: 176-199. Zumora-Munoz, C. and J. Alba-Tercedor. 1996. Bioassessment of organically polluted Spanish rivers, using a biotic index and multivariate methods. J. N. Am. Benthol. Soc, 15: 332-352. 161 Appendix A : Families, orders, and classes represented by macroinvertebrates collected from the study reaches. Family Order Class Ameletidae Ephemeroptera Insecta Apataniidae Trichoptera Insecta Astacidae Decapoda Crustacea Baetidae Ephemeroptera Insecta Blephariceridae Diptera Insecta Brachycentridae Trichoptera Insecta Capniidae Plecoptera Insecta Carabidae Coleoptera Insecta Cecidomyiidae Diptera Insecta Ceratopogonidae Diptera Insecta Chaoboridae Diptera Insecta Chironomidae Diptera Insecta Chloroperlidae Plecoptera Insecta Corydalidae Megaloptera Insecta Curculionidae Coleoptera Insecta Dixidae Diptera Insecta Dolichopodidae Diptera Insecta Dytiscidae Coleoptera Insecta Elmidae Coleoptera Insecta Empididae Diptera Insecta Enchytraeidae Oligochaeta Annelida Ephemerellidae Ephemeroptera Insecta Ephydridae Diptera Insecta Georyssidae Coleoptera Insecta Gerridae Hemiptera Insecta Glossosomatidae Trichoptera Insecta Goeridae Trichoptera Insecta Haliplidae Coleoptera Insecta Heptageniidae Ephemeroptera Insecta Heteroceridae Coleoptera Insecta Hydrophilidae Coleoptera Insecta Hydropsychidae Trichoptera Insecta Hydroptilidae Trichoptera Insecta Lepidostomatidae Trichoptera Insecta Leptophlebiidae Ephemeroptera Insecta Limnephilidae Trichoptera Insecta Luectridae Plecoptera Insecta Lumbriculidae Oligochaeta Annelida Mycetophilidae Diptera Insecta Naididae Oligochaeta Annelida 162 Appendix A (continued): Families, orders, and classes represented by macroinvertebrates collected from the study reaches. Family Order Class Nemouridae Plecoptera Insecta Perlodidae Plecoptera Insecta Philopotamidae Trichoptera Insecta Polycentropodidae Trichoptera Insecta Rhagionidae Diptera Insecta Rhyacophilidae Trichoptera Insecta Sciomyzidae Diptera Insecta Simulidae Diptera Insecta Spongillidae Haplosclerina Demospongea Staphylinidae Coleoptera Insecta Stratiomyidae Diptera Insecta Syrphidae Diptera Insecta Tipulidae Diptera Insecta Tubificidae Oligochaeta Annelida Uenoidae Trichoptera Insecta — Cladocera Crustacea — Collembola Insecta — Hydracarina Arachnoidea — Lepidoptera Insecta — Ostracoda Crustacea — Tricladida Turbellaria — — Gastropoda — ~ Nemotoda 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items