Comparing Hypotheses Proposed by Two Conceptual Models for Stream
Comparing hypotheses proposed by two conceptual
by
Sean Elliott Collins
B.S. Marshall University
M.S. Marshall University
A dissertation submitted to the
Graduate School
of the University of Cincinnati
in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy
in the Department of Biological Sciences
of the College of Arts and Sciences
March 2014
Committee Chair: S. F. Matter, Ph.D. Abstract
The broad goal of stream ecology is to understand and predict complex interactions between environmental factors and processes that occur within streams and rivers, such as biological community composition and their interactions, system metabolism (productivity and respiration), and nutrient sources and concentration. Multiple factors are thought to play important roles in these processes including regional environmental conditions (i.e., hydrology, geology, stream form/morphology) and longitudinal position within a stream network (defined by Strahler stream order). In the past, several theoretical concepts have been proposed to attempt to describe and explain how streams behave, and each concept uses various factors weighted differently to characterize streams and gain a better understanding of ecological processes and overall system functions. Here, two differing theories of stream ecology are compared – the River Continuum Concept (RCC) and the
Riverine Ecosystem Synthesis (RES). Each of these theories has unique predictions based on either Strahler stream order (SSO; used by the RCC) or Functional Process Zone (FPZ; used by the RES) defined by hydrogeomorphic characteristics. Predictions from both theories were tested across sites representing multiple SSOs and FPZs within the Kanawha
River Basin. Measures of environmental heterogeneity (an important concept for differentiating between FPZs) were also assessed.
This project has shown that some predictions from both the RCC and the RES are valid. The physical character of the basin is variable; sampling of riverbed substratum at each site revealed that similarities within each FPZ in riverbed composition exist and that each FPZ is distinct. Hydrogeomorphic factors including underlying geology and valley floor width strongly influence the character of the riverbed substratum. The ratio of
ii primary productivity to ecosystem respiration (i.e., a measure of metabolism) aligned with predictions from the RES where potential light availability and environmental heterogeneity were major drivers of this ratio. On the other hand, a “sliding scale” (i.e., one that changes with environmental variables) may be more appropriate for predictions from the RCC as the apparent trend toward higher rates of primary productivity was shifted from mid-order to larger streams. Neither the RCC nor the RES provide clear hypotheses for fluctuations in nutrient concentrations, and changes in these concentrations were not accurately explained by either SSO or FPZ. Finally, results from stable isotope analysis of carbon and nitrogen revealed that consumers in the Kanawha River Basin utilize a combination of aquatic and terrestrial organic carbon. While food web metrics and the proportional dependence upon in-stream and terrestrially derived carbon follow predictions of the RCC more closely, the mechanisms underlying these predictions are not always met. That is, there was no relationship between high measured rates of aquatic primary productivity and low dependence on terrestrial carbon sources. The RES also explains well the shift in organic carbon use from various sources by consumers.
Perhaps the best model for understanding these processes is a combination of both concepts. Understanding the reasons for changes in processes and functions within streams and rivers is critical for developing a useful and successful model. It is through this process of model conceptualization, testing, and refining that true advancements in knowledge can take place.
iii [This page intentionally left blank.]
iv Acknowledgements
This project would not be what it is without the help of many. I have had the privilege of working with some of the best mentors, fellow students, and colleagues that I could imagine. All the while, I have been supported by my friends and family. It is not without my deepest appreciation when I say thank you to you all.
My gratitude goes out to various sources of funding including the University of
Cincinnati, the University of New England, and the United States Environmental Protection
Agency. I have been supported by the Graduate Assistant Scholarship, the Wieman Wendel
Benedict Research Award, and the Choose Ohio First Scholarship, and I worked as a
Teaching Assistant at the University of Cincinnati. I have also worked as a Student Services
Contractor to the U.S. EPA under contract EP10D000363 and EP12D000302. Finally, I was awarded the International Graduate Student Assistance Scheme from the University of New
England. Funds for travel to conferences and regional meetings were also provided by the
Graduate School of the University of Cincinnati.
I would also like to thank my lab mates Megan Lamkin, Natasha Urban, Matt
Westbrook, Sasha Ramudit, and Hao Yuan Chen. You all have provided an excellent atmosphere for learning, sharing ideas, and developing as a scientist. I greatly appreciate the feedback you all have given on draft after draft of my grant proposals and manuscripts.
I am sure that none of those would be the same without you. Thank you all, also, for your friendship during my time at the University of Cincinnati.
During my time at UC, I have received assistance in the laboratory and the field from research scientists, other graduate students, and undergraduates. I would like to thank
Madhav Machavaram for his patience and wisdom. Without him, the stable isotope portion
v of this research would not have been completed. I would also like to thank Mark Mitchell,
Kathleen Hurley, Kate Johnson, and Jeremy Alberts for their help with nutrient analyses.
Brandon Armstrong, John Gorsuch, Brent Johnson, and Michael Moeykens also assisted me in the laboratory with organism identification and sample preparation. Brandon
Armstrong, Michael Moeykens, Sheila North, and Aaron Stahl also provided much appreciated assistance with field work. I hope that the mountains and streams of West
Virginia and Virginia were a reward for your work.
My Research Advisory Committee consisted of Stephen Matter, Ishi Buffam, Joseph
Flotemersch, Eric Maurer, Martin Thoms, and Amy Townsend-Small. Throughout my time at UC, each of you has contributed to my research project and my overall development as a scientist. You have all shared your knowledge and challenged me along the way. Steve has been an excellent advisor. He has given me enough freedom to develop my own ideas and project goals, and he has been readily available to help without hesitation. Manuscripts, grant proposals, and application documents have all passed over his desk, and each has benefited after his review. Ishi, I thank you passing on some of your knowledge of and expertise in freshwater systems. Nutrient analyses and dynamics are new concepts that I have added to my own repertoire. I would also like to thank Joe for his endless help in both the lab and field. I have learned so much from you. Thanks to Eric for challenging me and for providing valuable insight into my experimental design and implementation. Martin, I am truly grateful for the opportunity you afforded to me. Visiting the University of New
England and spending time with you and other members of the Riverine Landscape
Research Lab was incredibly rewarding. I would also like to extend my appreciation to Amy for teaching me both in and out of class. Your insight into stable isotope analysis and other
vi aspects of my project have made this dissertation better than it could have been without you.
Finally, I would like to thank my friends and family. I have had so much support from each of you, and I’ve been very fortunate to get to know so many wonderful people.
To my family in-law, Jim, Nancy, and Jeremy, thank you so much for your kind words and support throughout my time at UC. Time spent with you has been relaxing and enjoyable. A wonderful relationship with you all is something special that I am so grateful to have. I would also like to thank my own family. To my brother, Seth, his wife, Tabitha, and their family, thank you for your support, encouragement, and understanding. Although our paths may be different, we will never grow apart. To my parents, Gene and Elaine, I appreciate everything you have done for me. You instilled in me a work ethic that has allowed me to succeed in whatever pursuits I may choose. Your guidance in my life has been beneficial, and you have given me an example that I hope to follow. To my wife, Kelly, you have given me so much in our life together. Thank you for your patience, assistance, and assurance. We make a great team, and I am sure that the next steps in our life will be as wonderful as our first steps together.
vii Table of Contents Abstract ...... ii
Acknowledgements ...... v
List of Tables ...... xii
List of Figures ...... xiv
List of Abbreviations ...... xvi
Chapter 1 – An introduction into stream science ...... 1
A Brief History of Stream Science ...... 4
The River Continuum Concept ...... 6
The Riverine Ecosystem Synthesis ...... 8
Chapter Summary ...... 9
Chapter 2 – Evaluation of the sounding rod method for sampling coarse river-bed
sediments in non-wadeable streams and rivers ...... 12
Abstract ...... 13
Introduction ...... 14
Study Sites & Methods ...... 17
Results & Discussion ...... 19
Conclusion ...... 22
Chapter 3 – Hydrogeomorphic zones characterize riverbed sediment patterns
within a river network ...... 23
Abstract ...... 24
Introduction ...... 25
Study Area ...... 28
viii Methods ...... 32
Sediment Data Collection ...... 32 Statistical Analysis ...... 33 Results ...... 36
Discussion ...... 42
Conclusions ...... 46 Chapter 4 – The River Continuum Concept and Riverine Ecosystem Synthesis
explain patterns in system metabolism and macronutrients ...... 47
Abstract ...... 48
Introduction ...... 49
Methods ...... 52
Study Area ...... 52 Nutrient Data Collection ...... 54 Nutrient Analysis ...... 54 Estimation of Ecosystem Metabolism ...... 55 Statistical analyses ...... 55 Results ...... 57
Discussion ...... 70
Model Performance ...... 70 Nutrients ...... 72 Metabolism ...... 73 Conclusions ...... 75 Chapter 5 – Using stable isotope analysis of aquatic food webs to assess predictions
of the River Continuum Concept and the Riverine Ecosystem Synthesis ...... 77
Abstract ...... 78
Introduction ...... 79
Methods ...... 82
ix Study area ...... 82 Organism sampling and analysis ...... 82 Community metrics ...... 85 Isotope modeling ...... 85 Statistical analysis ...... 86 Results ...... 87
Community metrics ...... 87 Carbon sources ...... 90 Discussion ...... 94
Community metrics ...... 94 Carbon sources ...... 95 Conclusions ...... 97 Chapter 6 – Conclusions ...... 99
The Kanawha River Basin ...... 100
On models ...... 102
River Continuum Concept ...... 103
Riverine Ecosystem Synthesis ...... 105
Comparing models ...... 108
Bibliography ...... 112
x Appendix I – Habitat heterogeneity between Functional Process Zones ...... 137
Introduction ...... 137
Methods ...... 138
Results and Discussion ...... 139
Appendix II – Temporal variation in stable isotopes of carbon and nitrogen in fish
tissue ...... 142
Introduction ...... 142
Methods ...... 143
Results and Discussion ...... 144
Appendix III – Data ...... 145
xi List of Tables
Table 2.1 Description of field sites and their geographic location ……………………... 17
Table 2.2 Substrate class and size as determined by the copper pole method
and as measured using a gravelometer….………………………………………….. 19
Table 3.1 The characteristics of Functional Process Zones within the Kanawha
River Basin, USA………………………………………………………………………………...31
Table 3.2 Statistical analysis of the riverbed sediment texture for the Functional
Process Zones (FPZs) within the Kanawha River Basin.………..……………..40
Table 3.3 The proportion of sites determined via an ENTROPY analysis that are
allocated to the correct Functional Process Zone………………………………. 41
Table 4.1 Nutrient analyses using Strahler Stream Order (SSO) or Functional
Process Zone (FPZ) as factor…………………………………………………………….. 60
Table 4.2 Results from generalized linear models fitting nutrient data……………….62
Table 4.3 Metabolism predictions and results…………………………………………………... 63
Table 4.4 Results from generalized linear models fitting P/R data……..……………….69
Table 4.5 Results from generalized linear models fitting nutrients
and productivity………………………………………………………………………………..69
Table 5.1 Results from generalized linear models fitting community metrics..…….89
Table 5.2 Results from generalized linear models comparing Functional
Process Zones to other measures of environmental heterogeneity..…….89
Table 5.3 Diet contributions for each consumer listed by Functional Process
Zone (FPZ)……………………………………………………………………………………….. 92
Table 5.4 Diet contributions for each consumer listed by stream order……………... 93
xii Table 6.1 RCC and RES interpretations and results……………………………………………109
Table AIII.1 Description of thirty-five sampling locations within the Kanawha
River Basin………………………………………………………………………………………. 146
Table AIII.2 Description of substrate composition of each sampling location...………. 148
Table AIII.3 Nutrient concentrations measured at each site.………………………………… 149
Table AIII.4 Ecosystem metabolism estimates……………………………………………………… 150
Table AIII.5 Stable isotope signature for aquatic and terrestrial organic
carbon sources………………………………………………………………………………… 151
Table AIII.6 Stable isotope analysis results for all food web metrics……………………… 152
Table AIII.7 Stable isotopes of carbon and nitrogen for each consumer………………… 153
Table AIII.8 Stable isotopes of carbon and nitrogen for each producer………………….. 164
xiii List of Figures
Figure 2.1 Boxplot depicting the distribution of measured data for each size class… 20
Figure 3.1 The Kanawha River Basin……………………………………………………….………...…..29
Figure 3.2 The textural composition of riverbed sediments in each Functional
Process Zone…………………………………………………………………………….………… 37
Figure 3.3 SIMPER analysis of the riverbed sediment texture for each Functional
Process Zone within the Kanawha River…………………………………….……...... 38
Figure 3.4 Sediment character of the 35 sites within the Kanawha basin are
displayed in ordination space……………………………………………………………… 41
Figure 4.1 The Kanawha River Basin……………………………………………………….………...…..53
Figure 4.2 Concentration of each macronutrient based on SSO grouping………………. 58
Figure 4.3 Concentration of each macronutrient based on FPZ……………………………… 59
+ Figure 4.4 Ammonium concentration (NH4 ) varies significantly (F2,67 = 11.41;
P < 0.05) when SSO category was used as a factor…..……………………………. 64
Figure 4.5 Summer dissolved organic carbon (DOC) varies significantly (F3,31 = 3.66;
P < 0.05) when Functional Process Zone (FPZ) was used as a factor…….... 65
Figure 4.6 Total dissolved nitrogen (TDN) differs significantly (x2 = 9.49, d.f. = 3;
P < 0.05) when Functional Process Zone (FPZ) was used as a factor……… 66
Figure 4.7 Changes in some nutrient concentrations varied significantly
with the interaction between SSO groupings and FPZs…………………………..67
Figure 5.1 The Kanawha River Basin.……………………………………………………….………...... 83
Figure 5.2 Food web metrics respond to stream order and Functional Process
Zone (FPZ)………………………………………………………………………………………… 88
xiv Figure AI.1 Pebble size for 100 particles was measured at each site………………………. 140
Figure AI.2 Coefficient of variation was determined for daily average flow data..……. 141
xv List of Abbreviations
AIC – Akaike information criterion
ANOVA – Analysis of variance
CFS – Cubic feet per second
CV – Coefficient of variation
δ13C – Ratio of stable isotopes of carbon (13C:12C)
δ15N – Ratio of stable isotopes of nitrogen (15N:14N)
th D50 – Particle size (50 percentile)
DOC – Dissolved organic carbon
EMAP – Environmental Monitoring and Assessment Program
ER – Ecosystem respiration
FPZ – Functional Process Zone
GLM – Generalized linear model
GPP – Gross primary productivity
LA – Lowland Alluvial
LC – Lowland Constrained
+ NH4 – Ammonium
- NO3 – Nitrate
OU – Open-Valley Upland
3- PO4 – Phosphate
P/R – Ratio of productivity to respiration
RCC – River Continuum Concept
RE – Reservoir
xvi RES – Riverine Ecosystem Synthesis
SIA – Stable isotope analysis
SSO – Strahler stream order
TDN – Total dissolved nitrogen
TDP – Total dissolved phosphorus
UC – Upland Constrained
UH – Upland High-Energy
U.S. EPA – United States Environmental Protection Agency
xvii Chapter 1 – An introduction into stream science
A river seems a magic thing. A magic, moving, living part of the very earth itself.
– Laura Gilpin
1 A broad goal of ecology is to describe complex natural phenomena which includes understanding how organisms interact with their environment and how biological communities change in space and time. Many hypotheses have been proposed to attempt to explain some of the changes within and among populations and communities, and these hypotheses use a variety of principles to inform their predictions. Importantly, there are many contrasting ecological hypotheses which focus on change either along a continuum or among patches. A general example of change along a continuum is succession. In the early
1900s, Frederic Clements and Henry Gleason proposed disparate frameworks for describing succession. Broadly, succession can be described as the process of change that ecological communities undergo over time and after some disturbance. This change over time occurs over a gradient as progress is made toward a climax community. Alternatively,
G. Evelyn Hutchinson’s ideas about non-equilibrium communities are thought to be the origins of the Patch Dynamics Concept. Other ideas in patch dynamics incorporate landscape ecology perspectives on how spatial patterns influence ecological processes at multiple scales of space and time. Many habitat types including streams and rivers have been studied within the contexts of both continuum and patch frameworks, and there have been several hypotheses proposed to describe changes within river networks based on these principles.
Rivers and streams are inherently complex ecosystems, and the above ecological principles have been applied to these systems to help understand some of the functions and processes within. However, studying these systems can be difficult for several reasons including their relatively large spatial scales. Watersheds are often trans-boundary in nature (i.e., cross territory, state, or country lines) and require special consideration (e.g.,
2 permission from various entities) for effective sampling. Flowing waters also represent a challenge in that, while organisms may be free to move with or against the current, other materials are naturally forced downstream. Although point-in-time samples are useful, improper timing can result in a missed opportunity to capture a significant event with regard to the river. Fluvial systems exemplify a unique link to the terrestrial environment as they assimilate changes across the landscape. While small streams are highly influenced by their surrounding environment, larger rivers represent the accumulation of terrestrial processes as well as transformations that occur in aquatic habitats. It is for these reasons that monitoring limnological systems can show changes in both temporal and spatial scales across aquatic and terrestrial environments (Williamson et al. 2008).
Freshwater systems provide many key ecosystem services, that is, benefits people obtain from ecosystems. These include clean drinking water, nutrient cycling and transport, carbon sequestration, energy production, and cultural services (MA 2005).
These ecosystem services fall into two distinct groups: (1) marketable goods and services such as drinking water, transportation, and electricity generation; and (2) non-marketable goods and services such as biodiversity, plant and animal habitat, and cultural or societal significance (Wilson and Carpenter 1999). As stated by Costanza et al. (1998), “to say that we should not do valuation of ecosystems is to deny the reality that we already do, always have and cannot avoid doing so in the future”. Although assigning a value to these benefits is difficult, it is often essential to ensure that ecosystems are protected. Maintaining these services requires that streams and rivers are healthy and are supporting their natural processes and functions.
3 To that end, much research has been conducted to understand how freshwater systems behave within an ecological context. Whether to regulate water levels for flood control, transportation, or irrigation, to use rivers as a means for waste removal, or to provide food or biofuel, humankind has always needed to understand streams and rivers.
Natural curiosity as well as pragmatic necessity gave rise to a host of projects centered on comprehending the complexity of these important ecosystems.
A Brief History of Stream Science Aquatic ecologists have long sought to understand the physical and biological processes that occur within streams and rivers. Many reasons exist for this endeavor including gaining knowledge about complex ecosystems, maintaining the healthy structure and function of fluvial systems, controlling processes to provide human-derived benefits, and preserving key ecosystem functions that streams and rivers provide. Early attempts at explanations saw streams defined by their biota – the dominant fish or macroinvertebrate became the “name” for that type of river (Gordon et al. 2004). This idea of exclusive patches inhabited by unique biota soon became widespread, and it is still sometimes used as a classification scheme. Leopold and Marchand (1968) described streams in a different way, with hydrogeomorphology (i.e., stream form, geology, hydrology) taking the lead in river patch delineation, and they called for an understanding of rivers as being made up of distinct, repeatable patches or zones based on this classification. With a background in hydrology, Leopold, Wolman, Marchand and others defined streams and rivers based on factors such as downstream slope, bed material size, and channel sinuosity. Definitive work on stream morphology was later done by Rosgen and Silvey (1996), and their work is used
4 by engineers and project managers to reconstruct or rehabilitate streams with failing form or function.
Over time, individual units of fluvial processes and functions were studied. For example, nutrient cycling is a complex issue where water velocity and temperature, substrate character, and input volume all play a role in how (and how quickly) nutrients are transformed from inorganic to organic forms. Webster (1975) coined the term nutrient spiraling to express how this cycle takes place, and Newbold et al. (1981) described a mathematical formula incorporating key components affecting nutrient spiraling. River morphology is another topic which has received individualized attention (e.g., Leopold and
Marchand 1968, Leopold and Wolman 1957, Rosgen and Silvey 1996, Strahler 1964). Here the variety of forms for rivers, as defined by the underlying geology and landscape morphology, was characterized. Concepts of hydrogeomorphology are used in stream restoration projects where “naturalization” is the goal. Finally, watershed models have been created to predict hydrological, geomorphological, and biological processes within stream networks (for a review, see Moore et al. 1991). In some cases, streamflow and hydrologic regimes are estimated based upon remotely sensed data including watershed topography and hourly, daily, or monthly precipitation (e.g., Fernandez et al. 2000, Kite
1991, Leavesley and Stannard 1995). Each of these approaches, though, lacks the interdisciplinary nature of the problem.
As a result, attempts have been made to incorporate the understanding gained from many disciplines in the form of conceptual models. Several theoretical concepts have been produced which incorporate many of these units into cohesive models for stream behavior.
The River Continuum Concept (discussed in detail below; Vannote et al. 1980) synthesized
5 physical variables including riverbed substratum, water temperature, and stream size as well as biological factors such as primary productivity, ecosystem respiration, and invertebrate and fish communities into a predictive model where longitudinal position in a stream network should define the physical and biological attributes of a stream.
Alternatively, models exist where patches, zones, or domains are thought to control stream ecosystem processes (e.g., Montgomery 1999, Thorp et al. 2006, Townsend 1989). Here, hypotheses are based upon the influence of a combination of many environmental drivers
(e.g., elevation/topography, geology, morphology, precipitation), rather than the relative position within the network. These two types of theoretical concepts are compared herein.
The River Continuum Concept The River Continuum Concept of Vannote et al. (1980) is a theoretical concept for understanding a multitude of stream ecosystem processes. Once introduced, this theoretical concept was readily adopted as a model to describe and understand riverine ecosystem functions and processes including sediment transport, temperature gradients, carbon sources, and biological communities. The River Continuum Concept (RCC) is the conceptual paradigm for stream ecology. The central tenet of the RCC is that rivers are connected, longitudinal continua. As such, physical and biological processes that occur within rivers should represent a continuous gradient from headwaters to river mouths. For example, Vannote et al. (1980) describe a shift in substrate particle size from coarse boulders and cobble in headwaters to sand and fine silt in great rivers, and between these endpoints, a monotonic gradient of particle sizes should exist. According to Vannote et al.
(1980), many other characteristics (e.g., temperature, biological communities, carbon sources) should follow a longitudinal pattern with regard to stream order, as well. Further,
6 the RCC uses Strahler stream order (SSO), a hierarchical method for categorizing streams.
SSO only increases when two streams of the same order converge. Within the context of the
RCC, these stream order groupings (e.g., headwater streams between 1st and 3rd order) are used to predict functions, processes, and other characteristics within streams.
However, the RCC does not have definitive analytical support for many of its proposed tenets. Some studies performed using RCC as an ecological basis failed to support many of its proposed hypotheses. Several models have been proposed since the debut of the RCC, most building upon it (e.g., dams’ influence defined by the serial discontinuity concept, Ward and Stanford 1983; tributaries as modifiers of the RCC, Bruns et al. 1984; the flood pulse concept states floodplains’ importance in nutrient allocation, Junk et al. 1989).
These authors adhere to the foundations and concepts of the RCC giving slight modifications to make the concept more globally applicable. Others, on the other hand, attempt to show where the RCC failed to adequately describe riverine processes (e.g.,
Winterbourne et al. 1981, Sedell et al. 1989). The RCC and its modifiers are still widely used by academia and governmental agencies to describe and predict stream ecosystem processes.
Some 30 years after its inception, the RCC remains the paradigm for understanding stream science. In many cases, discrepancies are explained by a variety of causal factors.
For example, Minshall et al. (1983) note that the “postulated gradual change” expected from RCC was met, however “regional and local deviations occur” resulting from the influence of climate, geology, tributaries, lithology and geomorphology. They go on to state that RCC should be viewed as a sliding scale that should be moved to meet local environmental conditions. Minshall et al. (1985) recognized that the RCC is “undergoing
7 evolution, testing and refinement.” Some major refinements (in part, listed above) are separate concepts that use the RCC as a basis. Other minor refinements to the RCC include the simple recognition of work that pre-dated the RCC (e.g., riparian inputs, Hynes 1963; autochthonous production, Minshall 1978) or that was done after the RCC (e.g., nutrient spiraling, Newbold et al. 1981, 1982a, 1982b, Elwood et al. 1983; floodplain inputs, Sedell and Froggart 1984). Ultimately, the exclusion of climate and geology, as well as geomorphology, is recognized as a major limitation of the RCC (Minshall et al. 1985).
The Riverine Ecosystem Synthesis In hopes of providing an updated conceptual model, Thorp et al. (2006, 2008) proposed the Riverine Ecosystem Synthesis (RES). This model provides a holistic approach to river classification, noting that longitudinal position alone (as described by the RCC) may not be the most important factor in classification or prediction of system processes.
Instead, the RES proposed that several environmental hydrogeomorphic factors including channel width, channel sinuosity, valley width, channel-to-valley width ratio, precipitation, elevation, and geology may play an important role in river classification and predictions of ecosystem processes and products. The RES views streams and rivers as arrays of repeatable patches characterized by hydrogeomorphology; however, it ignores longitudinal effects. The stated goal of the RES is to blend current knowledge of riverine processes with an existing landscape model based on hierarchical patch dynamics defined by Wu and
Loucks (1995).
Fisher (1997) noted, and Thorp et al. (2006) reinforced, that to view a river as a linear ideogram is simply incorrect. Townsend (1989) states that these patterns are “not
8 usually realized and cannot provide a world-wide generalization”. Thorp et al. (2006) go on to argue that, in many cases, longitudinal patterns are weak. Instead, where a longitudinal patterns in a river break (e.g., tributaries, Benda et al. 2004; discontinuities, Statzner and
Higler 1986) biological and physiochemical patterns tend also to break. Moving away from a longitudinal continuum to describe patterns in ecosystem processes, Thorp et al. (2006,
2008) propose that hydrogeomorphic patches (defined at one level as Functional Process
Zones, FPZs) capture more of the environmental variability and thus may better predict patterns in those processes.
While potentially transformative for stream ecology, the assumptions and predictions of the RES are largely untested. The RES provides several hypothetical tenets which revolve around patch dynamics and hierarchy. Although the RCC implies a distinct, longitudinal pattern in ecosystem processes, Thorp et al. (2006) argue many of the modifications made to the RCC include discontinuities (e.g., Ward and Stanford 1983, Bruns et al. 1984, Junk et al. 1989). Thorp et al. (2008) write that their model accounts for this hydrologic variability and patchiness; however, they recognize that the RES remains a heuristic model in need of empirical support. The work presented below is an attempt to provide empirical support to either model as well as overarching ecological principles.
Chapter Summary In Chapter 2, I present a method for the rapid quantification and qualification of the composition of riverbed substratum. Physical habitat assessment is performed for a variety of purposes including biological monitoring. Based on Wolman’s (1954) protocol, this rapid technique broadly places individual particles into one of six categories (i.e., fine sediment, sand, gravel, cobble, boulder, bedrock) based on a common size scale. I demonstrate a
9 success rate > 80% at correctly determining particle size. This technique was used to characterize some of the spatial heterogeneity within the Kanawha River Basin, a basis for comparison of the RES.
Chapter 3 describes a study to show how well riverbed substratum distributions within the Kanawha River Basin align with Functional Process Zones. As FPZs are partially determined by underlying geology as well as topography, it was expected that patterns in riverbed sediments should be fairly well-predicted by a prior FPZ determination. Based on a variety of multivariate analyses, FPZs show distinct substratum distributions, and FPZ classification and sediment data match about 70% of the time.
In Chapter 4, I present the results of a study of dissolved nutrient concentrations and system metabolism with the Kanawha basin within the context of comparing hypotheses presented by the RCC and RES. Nutrients are often monitored to show the trophic state where elevated concentrations of nutrients can play havoc with “normal” functioning of streams. System metabolism, a combination of rates of primary production and ecosystem respiration, is a valuable method for describing overall ecosystem function.
Predictions of the RES are supported by metabolism estimates (while predictions of the
RCC are not); however, nutrient concentrations were not well-predicted by either model individually, but a combination of the models was the best predictor for some nutrients.
Chapter 5 shows the results from a study of the food web dynamics within the
Kanawha River Basin. Stable isotope analysis of carbon and nitrogen was used to determine a variety of metrics associated with food webs, and carbon sources were also identified. Although many metrics were assessed, few patterns were revealed within food webs among and between various stream orders or Functional Process Zones.
10 Finally, Chapter 6 provides a summary of the research contained in the preceding chapters. I also give an overview of what evidence this project has given in support of both models, and I explain how, although “all models are wrong, […] some are useful” (Box and
Draper 1987). In this case, a combination of explanatory variables from each model may be more useful than either model independently.
11 Chapter 2 – Evaluation of the sounding rod method for sampling coarse river-bed sediments in non-wadeable streams and rivers
This work appears as:
Collins, S. E. and J. E. Flotemersch. 2013. Evaluation of the sounding rod method for
sampling coarse river-bed sediments in non-wadeable streams and rivers. River
Research and Applications DOI: 10.1002/rra.2697
12 Abstract The substrate of fluvial systems is regularly characterized as part of a larger physical habitat assessment. These measures are instrumental in meeting the regulatory responsibilities of bioassessment and monitoring programs, and essential to monitoring restoration and rehabilitation success. We describe and validate a commonly-used technique for broadly categorizing, and thus characterizing, the substrate in non-wadeable streams and rivers called the sounding rod method. In brief, a rod, often hollow, is used to probe the substrate of non-wadeable systems to characterize the substrate. We tested the viability of this method on three different systems by comparing estimated particle class and direct particle measurements. Our results indicate that substrates can adequately be defined into six broad classes (fine-particle sediment, sand, gravel, cobble, boulder, and bedrock) based on size using the sounding rod. Estimated classes were significantly positively correlated to measured classes (τ = 0.83, p < 0.001), and estimates of size class and direct measurements of size were not from significantly different distributions (χ2
0.05,9 = 569.51, p < 0.001). Further, there were significant differences between each category (H = 243.5, 3 d.f., p < 0.001). While our results affirm that actual substrate class size can be directly inferred from estimated data, it should be noted that a layer of soft sediments < 20 cm was not always detected. This finding should be carefully considered with individual study objectives. Overall, the sounding rod method can be learned quickly, and it is a low-cost and time-efficient method for substrate classification.
13 Introduction The physical habitat of fluvial systems consists of structural features of the riverine environment that influence the composition and community structure of biota. As a result, habitat diversity and biological diversity are closely linked (Williams 1980, Raven et al.
1998). Higher habitat heterogeneity is generally associated with greater biotic diversity
(Vannote et al. 1980, O’Connor 1991, Vinson and Hawkins 1998). However, as a consequence of the close relationship mankind has with these systems, the character of streams and rivers is often challenged (Flotemersch et al. 2006). One of the principle stressors to biota in streams and rivers is the loss or damage of habitat (Karr and Dudley
1981, Karr et al. 1986). In a USEPA (2007) report, excessive streambed sediments were listed as one of the most widespread stressors in wadeable streams of the United States.
Likewise, most non-wadeable streams and rivers have been altered by a variety of human activities (Welcomme 1985, Dynesius and Nilsson 1994, Galat and Frazier 1996).
Humans have altered the physical templates of rivers, the hydraulic dynamics of their channels and tributary networks, and the land-use characteristics of their basins to an extent that has had a large and complex impact on the biota (Bayley 1995, Brookes 1988,
Poff et al. 2007. Due to these activities impacting fluvial habitats, the characterization of physical habitat is recognized as a critical component to most if not all bioassessment and monitoring approaches and programs in the United States (e.g., Plafkin et al. 1989, USEPA
1990, USEPA 1992, Davis and Simon 1995, Barbour et al. 1999, USEPA 2005, Flotemersch et al. 2006, USEPA 2007) as well as globally (Maddock 1999, Oliveira and Cortes 2005,
European Commission 2007).
14 There are a variety of habitat assessment approaches which range from highly quantitative methods designed to describe the geomorphic condition of streams and riparian zones as well as the habitat condition for biota (e.g., Kauffman and Robison 1997), to more qualitative methods using visually scored elements, principally designed to grade the in-stream and adjacent riparian habitat alone (Barbour et al. 1999). When combined with land use/land cover data for adjacent and catchment areas, it is possible to draw an accurate picture of physical factors acting on a site, thus helping with initial identification of potential stressors in impaired fluvial systems (Flotemersch et al. 2006).
Regardless of the programmatic approach taken, a generally universal physical habitat parameter measured is system substrate. A commonly used method to accomplish this task in wadeable systems is the Wolman pebble count (Wolman 1954) or some ensuing derivative thereof. The method is a relatively simple, yet statistically reliable, way of assessing median bed particle size. In brief, a person walks through the stream according to some predetermined scheme (e.g., grids, transects, random step-toe) and collects substrate samples that are then measured along their intermediate axis. Once collected, the data are often mapped to aid interpretation and determination of relationships of the data to other parameters (e.g., benthic macroinvertebrates; Zweig and Rabeni 2001).
However, as Wolman (1954) pointed out, a limitation of the method is the difficulty of sampling by hand in deep water. To overcome this limitation, Fitzpatrick et al. (1998) recommended the use of a sediment coring device, Ponar sampler, or an Ekman dredge.
However, these approaches can be very time-consuming and exhausting when the number of data points typical of a Wolman peeble count are collected (generally 100). Additionally, the feasibility of these approaches decreases with depth and increasing flow rates. Given
15 these difficulties, alternative methods have been developed that use side-scan sonar
(Kaeser et al. 2012 and cited references) and Doppler current profilers (Gaeuman and
Jacobson 2005). Limitations of these methods may include the initial equipment investment, the need for experienced equipment operators, the need for staff trained in interpretation of the data, and limitations resulting from water conditions (i.e., turbid water). Spanning the gap between these methods is a non-technical, low-cost method for characterizing channel substrate in non-wadeable streams and rivers that uses a long pole to contact and characterize channel substrates that are unreachable by hand.
This approach was first developed and documented for use in characterizing the non-wadeable substrates in lakes as part of the U.S. EPA Environmental Monitoring and
Assessment Program (EMAP; Kaufmann and Robison 1997, Kaufmann and Whittier 1997), and later modified for use in the U.S. EPA EMAP’s non-wadeable streams and rivers program (Lazorchak et al. 2000). Variations or use of the method are discussed in Wilhelm
2002, 2005; Angradi 2006; Blocksom et al. 2006, 2009; Flotemersch et al. 2006; and USEPA
2007. Poles used are generally a fiberglass surveying rod, or one constructed out of PVC or copper that has been capped on both ends. Of recent, a preference seems to have emerged for copper poles given their apparent ability to better allow the sampler to hear and feel the nature of the substrate. Common names for the method include the sounding rod, pole drag, thalweg pole, and copper-pole method. Nonetheless, given that “sounding rod” is the name first used to describe this method in writing (Kaufmanm and Whittier 1997), we will hereinafter refer to the method by that name.
Despite widespread use of the sounding rod method in the United States by programs designed to support or meet the reporting requirements of the Clean Water Act
16 of 1972 (USGPO, 1989), this method remains empirically unsupported. The objective of this paper is to evaluate this technique and determine its utility as a quick and effective method for assessing substrate composition in non-wadeable streams and rivers.
Study Sites & Methods We collected data from the Licking River (KY), the North Fork of the Cherry River
(WV) and Walker Creek (VA), all of which are located within the mid-Atlantic region of the eastern United States (Table 2.1). Sampling reaches varied in width from 10 m to 30 m and in median discharge from approximately 250 cfs to 4500 cfs. Sampling points in each stream were generally wadeable (< 1 m) at the time of sampling to allow for particle recovery. At each study site, a 3/4 inch (19 mm) hollow copper or aluminum pipe was used to estimate substrate class at > 100 locations within each stream site. The pipe was capped on each end to avoid filling the end with sand or fine-particle sediment. The length of the pipe used at each site was dictated by individual site conditions as use of an excessive length of pipe presents a recognizable safety hazard. We used the Wolman procedure as modified by Bevenger and King (1995) to include a zigzag pattern. Beginning at the upstream end of a survey area, > 100 observations were made. We used audible and tactile cues to determine particle class. Special care was taken to avoid viewing the particle before
Table 2.1 Description of field sites and their geographic location.
Site Latitude Longitude County State Median Discharge (cfs)
Licking River 38.789 -84.325 Pendleton KY 4540
Walker Creek 37.159 -80.858 Giles VA 377
N. Fork of Cherry River 38.229 -80.444 Webster WV 247
17 a class estimate was made, however we could not safely blindfold the pole user. The sounding rod was repeatedly tapped on the substrate, and a particle class estimate was recorded. In slower reaches, once the bottom was contacted, the sounding rod was rotated to help distinguish between sand (gritty) from silt and smaller particles. Six common categories were recorded based on the Wentworth (1922) scale of particle sizes. These included fine-particle sediment (clay/silt), sand (< 2 mm), gravel (2-64 mm), cobble (65-
256 mm), boulder (257-500 mm), and bedrock (large, unbroken surface). Each particle that was estimated using the sounding rod was also retrieved by hand and measured along its intermediate axis using an Al-Sci Field Sieve - Gravelometer (Albert Scientific; Bunte and
Abt 2001).
Data failed the Shapiro-Wilk test of normality. Silt and bedrock were not included in these analyses because physical measurements of intermediate axis length were not taken for these classes. For all analyses, α was set at 0.05. We performed a Wilcoxon signed-rank test on our measured intermediate axis lengths for each class against the median particle size for each accepted size class to determine if the sounding rod method correctly estimated size class. Data collected from field sites were ranked based on increasing estimated size class (fine-particle sediment = 1, sand = 2, etc.). Corresponding intermediate axis lengths were also ranked. Kendall’s tau-b was used to test the strength of the relationship between estimated and measured values for each particle. This measure is similar to Spearman’s rank correlation, but it gives a direct measure of the probability of matched data pairs (estimated and measured particles). We used a 4 x 4 contingency table to determine if estimated data and measured data were from a common distribution. We performed a Kruskal-Wallis test on ranked data using estimated class as a factor and used
18 Dunn’s multiple comparisons to determine if significant differences between groups were present.
Results & Discussion We found that there is a significant positive relationship between particle size estimated using the sounding rod and measured particle size (τ = 0.83, p < 0.001). This means that approximately 83% of our estimated data correctly placed substrate particles based on direct measurement of size. We also found that estimates of substrate size class and direct measurements of substrate size were not from significantly different
2 distributions (χ 0.05,9 = 569.51, p < 0.001). Sand, gravel, and cobble were correctly estimated (Table 2.2). Although the relationship not significant, the sounding rod estimates of boulder fell within the accepted range of that class 76.4% of the time. Each estimated class was significantly different than all other classes (H = 243.5, 3 d.f., p < 0.001; Fig. 2.1).
Table 2.2 Substrate class and size as determined by the copper pole method and as measured using a gravelometer. For the sounding rod method, the estimated size class is given. For the gravelometer, mean (+ standard error) is given. We calculated a percentage of the times the sounding rod correctly estimates class based on measured intermediate axis. A Wilcoxon signed- rank test was performed on each size class against class median to determine if the sounding rod estimate correctly placed particles.
Wentworth scale (mm) Sounding rod est. Gravelometer (mm) % correct p Fine-particle (< 0.06) Silt/clay ------Sand (< 2) Sand 3.2 (+ 0.31) 65.9 < 0.001 Gravel (2-64) Gravel 33.8 (+ 2.78) 90.0 0.003 Cobble (65-256) Cobble 168.8 (+ 6.73) 90.2 0.015 Boulder (257-500) Boulder 362.7 (+ 18.45) 76.4 0.157 Bedrock (> 500, unbroken) Bedrock ------
19
Figure 2.1 Boxplot depicting the distribution of measured data for each estimated size class. Kruskal-Wallis on ranked data (H = 243.5, 3 d.f., p < 0.001) and Dunn’s multiple comparisons post- hoc procedure was performed. All groups were significantly different than all others (p < 0.05).
Use of the sounding rod method for characterization of substrate in non-wadeable streams and rivers offers several benefits. First, data that meets the data quality objectives
(Flotemersch et al. 2006) of most non-wadeable stream and river studies can be collected with an effort comparable to that in wadeable systems. Our work showed that data collection for approximately 0.5 km of stream substrate took only 1 hour. Additionally, as a result of the simplicity of the method, there is no data processing time (beyond data entry)
20 prior to analysis. Second, the sounding rod technique is a relatively low-cost process. A copper pipe (or similar material) and simple statistical software are all that is required to analyze substrate composition.
The sounding rod method also has some shortcomings. For example, in being consistent with the design of the Wolman pebble count method (Wolman 1954) multiple substrates at a single sample point are classified as the dominant substrate type. The assumption made here is that the variability at a sample location will be captured by the approximately 100 points sampled by the method. We have not investigated a technique for combining two or more substrate types into one categorical variable. Combining values is a possibility. We must, however, take into account proportions of each substrate type to accurately portray the area.
We also found some limitations with fine-particle sediment and bedrock. In five out of nine cases where 200 mm or less of fine sediments were present over a harder underlying substrate (e.g., cobble, boulder), the overlying fines were not detected by the sampler. When more than 200 mm of fine sediment was present, the silt was detected with the underlying substrate being noted. Wolman (1954) noted a limitation of the method was the inability to accurately distinguish between finer sediments and recommended that the smallest sizes be lumped together. The limitation we note is different in that, using the sounding rod, we were unable to consistently detect the presence of fine sediments when the thickness was less than 200 mm.
We also had difficulty discerning between very large boulders and bedrock. Unless a very obvious edge of the boulder can be felt, differentiating between these two classes can be difficult. Similar problems have been reported (Phil Kaufmann; personal
21 communications) and were circumvented in that case by combining bedrock and hardpan in to a single category.
Conclusion The sounding rod technique allows for the quick, cost-effective analysis of coarse river-bed materials in non-wadeable streams and rivers up to a depth of about 4 m. As presented, the method partly overcomes the depth limitation of the original Wolman method (1954), retains the limitations of an inability to distinguish among finer sediments, and has the added limitation of an inability to consistently detect the presence of fine sediments < 200 mm thick. Even so, the sounding rod method is a low-tech cost-effective method for characterizing coarse river-bed sediments in non-wadeable streams and rivers.
22 Chapter 3 – Hydrogeomorphic zones characterize riverbed sediment patterns within a river network
This work has been accepted as:
Collins, S.E, M.C. Thoms, and J.E. Flotemersch. 2014. Hydrogeomorphic zones characterize riverbed sediment patterns within a river network. River Systems.
23 Abstract Simplified techniques for assessing the condition of aquatic ecosystems are widely used at multiple levels of investigation. Tools for the rapid characterization of the physical structure of river systems at the entire network scale are limited. Functional Process Zones
(FPZs) are remotely-derived river sectors representing different hydrogeomorphic and ecological perspectives of river networks. Here we investigate associations between FPZs and the composition of 3500 field-derived riverbed sediment samples for the Kanawha
River Basin, USA. Using various statistical analyses, we confirmed the texture of the riverbed substratum differed between FPZs of the Kanawha River. Catchment geology and valley floor width, surrogates for sediment supply and overbank hydraulic conditions, respectively, were significant factors associated with textural differences in riverbed substratum between FPZs. Self-emergent groups of riverbed sediments, derived independently via Entropy analysis, agreed with the a priori FPZ classification approximately 70 percent of the time. We demonstrate that rapid assessment of riverbed sediments can be undertaken through an initial analysis of FPZs within a stream network.
FPZs can be generated with remotely sensed data and can account for variability in riverbed sediments at a basin scale thereby contributing to management activities including bioassessment, monitoring, restoration, impact identification, and characterization for environmental flows programs.
24 Introduction Understanding the character and distribution of riverbed substratum types within river networks is important for a variety of concerns. The size of riverbed sediments is the primary factor influencing the abundance and distribution of aquatic insects (Minshall
1984); it can also influence the success of spawning of certain fishes (Kondolf 2000), and is a critical factor of the overall stability of the river substratum, both in terms of the initiation of movement and the depth of disturbance (Carling 1983). Furthermore, the relative contribution of “finer” particle sizes has been used as an indicator of the impairment of the riverbed substratum as a result of catchment disturbance (Petts 1989;
Walling 1995; Walling & Fang 2003; Owens et al. 2005). Environmental resource managers, commonly use information on the riverbed sediments for a variety of reasons, including improving reference site selection, contributing to the robustness of condition assessments, developing ecological endpoints for restoration, evaluating ecosystem services, as well as assisting in establishing asset trading (Flotemersch et al. 2010; Thorp et al. 2010). Riverbed sediments have been reported to decrease in size with distance downstream (Yatsu 1955; Church & Kellerhals 1978). This diminution, commonly expressed as an exponential power function, is governed by two primary variables: decline in stream power with distance downstream thereby limiting conditions of competence, resulting in selective transport (Scott 1967); and increased abrasion of particles with increasing distance (Bradley et al. 1972). Despite studies confirming “Sternberg's (1875)
Law” to be a reasonable approximation for the longitudinal sorting of riverbed sediments, discontinuities do occur and are often reported (Lisle et al. 2000). Improving our knowledge of the variable composition of riverbed sediments contributes to a greater
25 understanding of the structure and function of river ecosystems and is therefore critical to their successful management.
Sampling of riverbed sediments to obtain detailed granulometric information is labor intensive (Thoms 1992). Field techniques can be a 2-3 person operation requiring specialist equipment and the collection of large quantities of riverbed materials at a site
(Mosley & Tindale 1992). An array of sampling methods is available for collecting riverbed sediments (see Klingeman & Emmett 1982 for a review) reflecting various field conditions and different studies for which they are designed. To date, focus has been on developing protocols and procedures for the collection of coarse surface sediments at-a-site that are not readily applicable to the assessment of riverbed sediments across entire stream networks. Non-quantitative (e.g., Bjorkland et al. 1999) and video-based (e.g., McConkey et al. 2009) methods have been developed, but many of these are also time-consuming and have limited value at a basin-level. Other remote sensing techniques (e.g., LiDAR,
Terrestrial Laser Scanning) have been applied to determine channel boundary roughness
(Heritage & Milan 2009); however, they lack the resolution necessary for sediment characterization at a site and are only applicable to dry channels or gravel bars (Milan et al. 2007).
Simplified techniques for assessing the state or condition of aquatic ecosystems are widely used at multiple levels of investigation (e.g., via remote sensing, semi-quantitative).
Rapid bioassessment protocols based on semi-quantitative biotic sampling are routinely used in many countries to obtain measures of river health (Norris & Thoms 1999). To facilitate the rapid assessment of riverbed sediments, indirect methods can be employed
26 especially where information on their character may not be available or is deficient (Thorp et al. 2008).
Potentially, Functional Process Zones (FPZs) –segments of river channel that have similar geological histories, flow and sediment regimes, as well as channel morphologies – could be used to this end. Determined by a range of approaches including desktop methods and or detailed field investigations, they are unique hydrogeomorphic sections within a stream network and represent an important level of organization in the hierarchy of fluvial ecosystems (Thoms et al. 2004; Thorp et al. 2008). Regardless of the approaches used to delineate FPZs, knowledge of their spatial organization within a stream network represents a potential means to determine, a priori, the abiotic and biotic character of riverine ecosystems (Thorp et al. 2008). Knowledge of the spatial organization of FPZs have been successfully used in several scenarios including auditing and assessment of rivers within the Murray-Darling Basin, Australia (Thoms et al. 2007) and setting environmental water allocations in the rivers of the Kingdom of Lesotho (King & Brown 2006). Although FPZ determination represents an approach allowing for inferences to be made about river ecosystems, testing of some of these inferences is required if they are to be universally accepted as a surrogate measure for the structure of, and processes operating within, river ecosystems.
The aim of this study is to determine if a priori derived FPZs can differentiate differences in the textural character of riverbed sediments found using the sounding rod method from the Kanawha River Basin, USA. If significant differences in riverbed substratum exist between a priori determined FPZs, they could be used as a cost-effective desktop (i.e., remotely sensed) assessment method for characterization of patterns in river
27 network substrate composition and thereby increase the overall understanding of the structure and function of river ecosystems. Consequently, FPZ determination could replace inefficient methods for riverbed sediment measurement, and this may impact monitoring and management of river networks.
Study Area The Kanawha River Basin, located in the Mid-Atlantic region, USA (Figure 3.1a), has a catchment area of 31,691 km2 that is relatively undisturbed. The catchment contains mostly (>80 percent) a mixture of northern hardwoods (e.g., Acer saccharum, A. negundo,
Betula nigra, Liriodendron tulipifera, Prunus serotina, Ulmus americana), oak (e.g., Quercus alba, Q. rubra), pine (e.g., Pinus strobus, P. virginiana), and mixed mesophytic forests that extend from an altitude of 1,355 m to 164 m (Messinger & Hughes 2000). There are three main physiographic provinces within the catchment (Figure 3.1b), with each province having a unique mixture of sandstone, shale, limestone, dolostone, chert and alluvium. The
Appalachian Plateaus, located in the northern sections of the basin accounts for ~60 percent of the catchment area. Paleozoic sedimentary rocks dominate this uplifted province of the Appalachian Plateaus. Sedimentary rocks of various ages dominate the
Valley and Ridge Province located in the mid regions of the basin. This province consists of a series of parallel northeast trending ridges and intervening valleys that are strongly folded and faulted. The Blue Ridge Province occupies the southern sections of the basin, and it contains a series of metamorphic and igneous rocks that are exposed in a series of thrust sheets and, as a result, is a region characterized by relatively deep, narrow valleys.
The Kanawha River Basin has a typical continental climate with marked seasonal differences in precipitation and runoff. Long-term mean annual precipitation (1961-2000)
28
Figure 3.1 The Kanawha River Basin. a) The river basin and its approximate location within the continental USA; b) the stream network of the basin where the Kanawha River is in boldface and physiographic provinces are shown; and, c) the location of Functional Process Zones and the riverbed sediment sampling sites.
29 for the basin ranges from 910-1,500 mm and overall there is 50 percent more precipitation in the winter months compared to the summer months. This seasonal variation is consistent between the three provinces. Precipitation is also variable across the basin.
Although precipitation is significantly positively correlated with elevation (Messinger &
Hughes 2000) there is a strong orographic influence with marked reductions on the leeward side of all mountains and ridges. Annual precipitation is generally greatest in the
Blue Ridge Province (1,270 mm) compared to Appalachian Plateaus (1,060 mm) and Valley and Ridge Province (990 mm). The long-term (1961-2000) mean daily discharge is 432 m3 s-1 with a range of 31 - 6,116 m3 s-1, at Charleston, WV, approximately 60 km from the confluence of the Kanawha and Ohio Rivers. Runoff across the basin is highly correlated with rainfall thus the most runoff is generated in the Blue Ridge Province.
The Kanawha River is a 6th order stream with a typical trellised network pattern
(Figure 3.1b) suggesting a relatively strong geological influence on the overall spatial organization of stream network draining the Kanawha Basin. Williams et al. (2013) characterized six distinct FPZs within the Kanawha River Basin (Table 3.1). The Upland
Constrained FPZ dominates the stream network of the Kanawha, accounting for 37.3 percent of the total network length. Contributions of the other FPZs are Lowland
Constrained (18.8 percent), Upland High-energy (16.8 percent), Reservoir (11.7 percent),
Lowland Alluvial (11.6 percent) and Open-valley Upland (3.8 percent). The spatial organization of FPZs resembles a mosaic of patches throughout the stream network, however there are repetitive patterns to the arrangement of FPZs within the basin (Figure
3.1c). Most FPZs are found in each physiographic province however, the Lowland
Constrained FPZ is only located in the Appalachian Plateaus Province. In general, the
30 Table 3.1 The characteristics of Functional Process Zones within the Kanawha River Basin, USA.
Functional Description Proportion of the Process Zone total stream length (abbreviation) (%) Lowland Alluvial geology, Low elevation, Wide valley floors and 11.6 Alluvial lateral floodplains, Low ratio of valley width to valley floor (LA) width, Relatively steep valley side slopes, Low down valley slope Lowland Bedrock geology, Low elevation, Relatively narrow valleys, 18.8 Constrained Narrow valley floor, High ratio of valley width to valley (LC) floor width, Steep valley side slopes, Moderate down valley slope Upland Bedrock geology, High elevation, Relatively narrow valleys, 37.3 Constrained Narrow valley floor, High ratio of valley width to valley (UC) floor width, Relatively steep valley side slopes, Moderate down valley slope Open-Valley Bedrock Geology, High Elevation, Very wide valleys, 3.8 Upland Relatively wide valley floor, Relatively lower ratio valley (OU) width to valley floor width, Low valley side slopes, Steep down valley slope Upland High- Bedrock geology, High elevation, High mean annual 16.8 Energy precipitation, Relatively narrow valleys, Very narrow valley (UH) floors, High ratio of valley width to valley floor width, Relatively steep valley side slopes, Steep down valley slope Reservoir (RE) Lentic waterbody 11.7
31 sequence of FPZs along the stream network reflects the varying influence of catchment and valley conditions. Upland High Energy FPZs are located in the upper regions of the network, while Upland Constrained and Open-valley Upland FPZs tend to be more prominent in the mid sections of the stream network. In the lower-most sections of the
Kanawha River basin the Lowland Constrained and Lowland Alluvial FPZs dominate. A feature of the Kanawha River Basin is that FPZs repeat themselves along the stream network rather than follow a clinal pattern.
Methods
Sediment Data Collection Thirty-five sites were selected for sampling within four FPZs of the Kanawha River basin during summer 2012 (Figure 3.1c). Nine sites were sampled in the Lowland Alluvial,
Lowland Constrained and Upland Constrained FPZs while eight sites were sampled within the Upland High-energy FPZ due to limited access. The Open-valley Upland FPZ was not sampled because of its low representation within Kanawha River stream network; Open- valley Upland represents 3.8 percent of the total stream network length. The Reservoir FPZ was also not sampled because it represented the impounded sections of the Kanawha network. Sites were randomly located within each FPZ and sampled at the nearest available access point.
At each of the 35 sites, 100 samples were recorded using a rapid sediment textural method described below. This method records sediment texture allocated to a series of the textural groups using a sounding rod technique, and it has been demonstrated to be a sampling strategy that requires a low time commitment (approximately 30 min/100 samples) and minimal equipment. It is an accurate and precise method, with > 80 percent
32 of samples being allocated into the correct textural group (Collins & Flotemersch 2013).
This procedure follows the Wolman (1954) method, with some modifications as recommended by Bevenger & King (1995) to include a zigzag pattern at each site. In brief, sediment size data were collected along a haphazard zigzag from bank to bank until 100 samples were measured. Each sediment sample recorded was allocated to one of six textural groups, these being silt/clay, sand, gravel, cobble, boulder particles, and bedrock.
These textural groups are based on the Wentworth (1922) scale, which is commonly used for sediment textural analysis. Mean particle size (D50) was estimated as the upper limit of each textural group.
Statistical Analysis The method of sediment data collection used in this study precludes standard textural analyses, which commonly derive a series of sediment moments (i.e., summary statistics for sediment distributions) (Folk 1966). A least squares regression between site
D50 and distance downstream was undertaken to determine if the textural character of the riverbed sediment could be explained solely by position along the river network. Sediment size data for each FPZ were then initially pooled and sediment textural group distributions constructed for each FPZ. Each distribution was tested for normality via the Shapiro-Wilk test, and then differences of D50 between FPZs were analyzed using the non-parametric
Kruskal-Wallis test since the assumption of normality was not met. We used R (version
2.15.1; R Core Team, 2013) to perform univariate statistical analysis.
Initially, the representativeness of the sediment samples within each FPZ was determined using two measures of similarity. First, the Bray-Curtis (1957) dissimilarity measure among sites within each FPZ was determined. This metric has a range between 0-
33 1 where low values represent more homogenous sites. Second, the Sorensen (1948) index for each site within each FPZ was calculated. This metric also has a range between 0-1, and low values indicate homogeneity. The Bray-Curtis dissimilarity provides a quantitative measure of the dissimilarity among sites within each FPZ, while the Sorensen index uses a binary index for the same calculation.
The textural character of different FPZ’s was then further examined via a range of multivariate statistical analyses performed on sediment size class estimates at each site.
Initially, a one-way ANalysis Of SIMilarity (ANOSIM) tested for differences between FPZs based on the derived Gower matrix using 1000 permutations. Note that Bonferroni correction for multiple pairwise tests is inappropriate for ANOSIM (Clarke 1993), thus significance for multiple pairwise tests for ANOSIM was determined with α = 0.05. If significance was determined using ANOSIM, a SIMilarity PERcentage (SIMPER) analysis was used to determine which individual sediment textural group contributed significantly to within-FPZ similarity and to assess whether this varied between FPZs. The Gower environmental distance measure (Belbin 1993), which incorporates an implicit range- standardization of variables, was used to derive a matrix of environmental distances between the four FPZs based on the contribution of the different textural groups. In addition, a Semi-Strong-Hybrid Multi Dimensional Scaling (MDS) (Belbin 1993) ordination allowed the similarity matrix to be presented graphically. A stress level of less than 0.2 indicated that the MDS ordination solution was not random and could be represented in two dimensions. Distances between the centroid of samples representing each FPZ were then determined to assess relative differences, in multidimensional space, between FPZs.
34 Relationships between a series of 15 catchment variables (as described in Thoms et al. 2004) and the position of the sediment character of each FPZ in multi-dimensional space were determined using Principal Axis Correlation (PCC) (Belbin 1993). The catchment variables used in the PCC were elevation, geology, mean annual rainfall, valley width, valley trough width, ratio of valley width to valley trough width, valley slide slope left and right, down valley gradient, wave length of channel belt, sinuosity of channel belt, width of channel belt, sinuosity of river channel, number of river channels, river channel planform and only those variables with an R2 greater than 0.5 were considered. The variability of the riverbed sediment texture within each FPZ was also determined according to the rank dissimilarity used to compute the comparative Index of Multivariate Dispersion (IMD) as described by Warwick & Clarke (1993). This measure determines if substrate composition
(i.e., the assemblage of sediment textural groups) differs in terms of its variability between
FPZs (Warwick & Clarke 1993). Thus, this multivariate measure of rank dissimilarity (i.e.,
IMD) was adopted for the present study to determine if patterns or relationships existed between FPZs and the variability of riverbed sediment texture within each FPZ.
Finally, the riverbed sediment data were analyzed using a multivariate statistical procedure - namely Entropy - as outlined in (Forrest & Clarke 1989), to independently identify groups of samples with similar sediment textural character. These groupings were then compared to the a priori FPZs to assess which sites were correctly allocated to the different FPZs. All multivariate analyses were performed in PATN v3 (Belbin 1993) and
PRIMER 6 (Warwick & Clark 1993) software packages.
35 Results The Bray-Curtis dissimilarity and Sorensen index calculated for each FPZ shows that sites within each FPZ were highly homogenous in their distributions. Bray-Curtis dissimilarity values for the four FPZs in the Kanawha River were all less than 0.5 indicating the quantitative distribution of sites within each FPZ was more similar to sites within each
FPZ than to random distributions. In addition, the Sorensen index calculated for each FPZ was less than 0.2 confirming the sediment character of sites within each FPZ was highly similar. Furthermore, textural character at a site was poorly explained by distance along the river network. Although the regression was significant, a very low least squares regression fit was achieved (F1,33 = 14.02; P = 0.00068; R2 = 0.29).
Each FPZ of the Kanawha River Basin has a unique textural character (Figure 3.2), in terms of the relative composition of the six textural groups recorded. The textural composition for each FPZ is not normally distributed (W=0.907, P=0.006) and significant differences exist between FPZs, in terms of the average sediment composition (x2=13.334, d.f.=3, P=0.004 – Kruskal-Wallis). The Lowland Alluvial and Lowland Constrained FPZs did not contain all six textural groups. The bedrock textural group was not present in these zones. Cobbles were the dominant textural group in all FPZs except the Lowland Alluvial
FPZ. This textural group represented 29.2, 37.8 and 31.5 percent of the substratum composition, on average, in the Lowland Constrained, and Upland High-energy FPZs respectively. The silt/clay textural group was dominant in the Lowland Alluvial FPZ, contributing 39.2 percent to the substratum textural composition.
Differences in the relative composition of the riverbed substratum between FPZs are confirmed by the SIMPER analysis (Figure 3.3). The relative importance of the six different
36
Figure 3.2 The textural composition of riverbed sediments in each Functional Process Zone. a) Lowland Alluvial; b) Lowland Constrained; c) Upland Constrained; and, d) Upland High-Energy. Error bars represent standard error. Significant differences exist between the average sediment composition of each FPZ based on a Kruskal-Wallis test (x2=13.3335, d.f.=3, P=0.003968).
37
Figure 3.3 SIMPER analysis of the riverbed sediment texture for each Functional Process Zone within the Kanawha River. SIMPER analysis shows the relative contribution (%) each category has to the observed similarity within each FPZ.
38 textural groups varies between FPZs. The silt/clay textural group was more prominent in the Lowland Alluvial and Lowland Constrained FPZs, contributing 53.3 and 44.6 percent respectively of the within FPZ sediment composition similarity. While gravel, cobble, and bedrock textural groups contributed 25.5, 21.8 and 27.1 percent to the within zone similarity of Upland Constrained FPZ and gravel, cobble, and boulders contributed 21.3,
31.5, and 25.4 percent respectively to the Upland High-energy FPZ.
The ANOSIM confirmed significant differences in the sediment composition between
FPZs of the Kanawha River basin (Global R =0.58). Multiple pair-wise comparisons show significant differences between most of the FPZs (i.e., four of the six pair wise relationships) (Table 3.2). The Lowland Alluvial FPZ was significantly different to all FPZs and the Lowland Constrained FPZ was different to Upland High-energy (Table 3.2). Only two pair-wise comparisons had values less than 0.3 (i.e., Upland Constrained vs. Lowland
Constrained and Upland Constrained vs. Upland High-Energy) suggesting the overlap between zones in terms of sediment composition was too great. This ordination confirms the wide distribution of sites in multivariate space and the clear separation of sites based upon FPZs (Figure 3.4). There are greater distances between the centroids of the Lowland
Alluvial and Upland Constrained FPZs (0.24) and Lowland Alluvial and the Upland High- energy FPZs (0.21) compared to the other FPZs in ordination space (Table 3.2). A feature of this ordination is differences in the spread of sites associated with each FPZ, and this is confirmed by the multivariate dispersion index (IMD) for each zone (Table 3.2). The
Lowland Alluvial and Lowland Constrained FPZs have the higher IMD values of 1.07 and
1.06, respectively compared to 0.82 for the Upland Constrained FPZ and 0.83 for Upland
High-energy FPZ. Higher IMD values indicate a greater variance between sites within a FPZ.
39
Table 3.2 Statistical analysis of the riverbed sediment texture for the Functional Process Zones (FPZ) within the Kanawha River Basin. Results of an Analysis of Similarity (ANOSIM) shown as R-values, centroid distances between each FPZ, and the Index of Multivariate Dispersion (IMD) are given. All significant p-values for ANOSIM were <0.001. ANOSIM Centroid distance IMD Lowland Lowland Upland Upland High- Lowland Lowland Upland Upland High- FPZ Alluvial Constrained Constrained Energy Alluvial Constrained Constrained Energy Lowland ------1.07 Alluvial Lowland 0.318 * - - - 0.132 - - - 1.06 Constrained Upland 0.468 * 0.011 ns - - 0.240 0.127 - - 0.92 Constrained Upland 0.349 * 0.307 * 0.032 ns - 0.209 0.108 0.064 - 0.93 High-energy ns – no significant difference * - overlap between groups but significant difference exist ** - no overlap between groups and a significant difference
40
Figure 3.4 Sediment character of the 35 sites within the Kanawha basin are displayed in ordination space, and the main independent catchment variables valley floor width (VFW) and geology are shown as PCC vectors.
Table 3.3 The proportion of sites determined via an ENTROPY analysis that are allocated to the correct Functional Process Zone. FPZ % Correct Lowland Alluvial 55.6 Lowland Constrained 44.4 Upland Constrained 66.7 Upland High-energy 22.2 Overall 70.0
41 Two catchment variables from the PCC are highly associated with the location of sites in this ordination space. Catchment geology and the width of the valley floor trough have relatively high R2 values of 0.658 and 0.559, respectively. The catchment geology vector is orientated towards the bottom right quadrant of the ordination, while the valley floor width vector is towards the top left quadrant. These two variables are surrogates for sediment supply and high flow energy conditions. The remaining variables all had R2 values less than 0.23 which were weak.
Five distinct riverbed sediment classes self-emerged from the Entropy analysis and these classes accounted for 76.89 percent of the total variance of the 3,500 riverbed sediment samples collected in the Kanawha River. However, 70 percent of the samples grouped via the Entropy analysis matched the a priori FPZ classification (Table 3.3). In terms of the individual FPZs, 66.7 percent of the samples from the Upland Constrained FPZ are correctly aligned with a corresponding riverbed sediment class emerging from the
Entropy analysis. However, only 22.2 percent of sites from the Upland High-energy FPZ are aligned with that emerging from the Entropy analysis.
Discussion The textural character of riverbed sediments does not possess an entirely random pattern within river systems (Church et al., 1987). Variations in sediment supply and sorting mechanisms act either individually or together, to provide complex spatial textural patterns at a range of scales. Although a decrease in particle size with distance downstream has been demonstrated for riverbed sediment (Yatsu 1955; Church & Kellerhals 1978),
“Sternberg's (1875) Law” may not be a universal approximation for the longitudinal sorting of riverbed sediments. In the Kanawha River Basin, position along the river
42 network was not a major influence on the textural character of the riverbed sediment.
Discontinuities in the downstream continua have been reported to result from tributary inputs (Knighton 1980), differential abrasion of differing geologies, variations in river base- levels (Shaw & Kellerhals 1982), discontinuities of slope, the lack of particles within a particular clast size (Yatsu 1955), and the interaction with adjacent floodplain deposits
(Pickup 1983). Progressive downstream sorting and abrasion of riverbed sediment resulting in a systematic and regular decrease in sediment size downstream are a gross over-simplification, even accounting for different geologies and the influence of local source areas. Alternate models depict riverbed sediments to be organized into a series of zones. For example, Pickup & Warner (1984) identified a series of five well-defined riverbed sediment zones, within the Fly and Purari River systems of Papua New Guinea that changed in a systematic manner along each river.
The spatial organization of FPZs within the Kanawha River Basin does not display a systematic progression downstream (Williams et al. 2013). FPZs are repeatable along the stream network. Given the distinct textural character of riverbed sediment between the different FPZs, the spatial organization of the riverbed sediment can be portrayed as a complex array of larger scaled patches or riverbed sediment zones within the Kanawha
River Basin. The complex spatial organization of these riverbed sediment zones, in association with the physical character of the riverine landscape, is a result, in part, of the larger scale catchment conditions.
The stream network of the Kanawha River Basin is composed of series of FPZs. This study has shown each FPZ to have a unique riverbed sediment texture. Most of the pair- wise comparisons between FPZs were significantly different from one another (i.e., four out
43 of the possible six pairwise comparisons). Although significant differences in the textural character of riverbed substratum are evident between the four FPZs of the Kanawha River
Basin, the textural character does vary between individual FPZs. The textural character of riverbed sediment in the Lowland Alluvial FPZ is statistically different from all others, with finer sediment classes being more prevalent in this FPZ. In contrast, the riverbed sediments of the Upland Constrained FPZ are different in textural character to those of the
Lowland Constrained or the Upland High-energy. Coarser sediment classes dominate these three FPZs, albeit with slightly different compositions. Differences between the four FPZs are associated with two-catchment scale variables: catchment geology and valley floor width. Catchment geology is a surrogate for the quantity and quality of sediment supply to a river network (Shaw & Kellerhals 1982). Relatively resistant geologies, located in the southern regions of the Kanawha River Basin are associated with increases in the presence of bedrock of the Upland High-energy FPZ, while the dominance of sandstone, shale, and limestone geologies, which are less resistant geologies, are associated with increases in the abundance of finer sediment classes in the Lowland Alluvial FPZ.
In addition, the geomorphic effectiveness of overbank flows and their ability to influence the character of riverbed sediment is conditioned by the nature of the valley floor setting. In a study of the Potomac River basin, Miller (1995) found the spatial pattern of flood hydraulic conditions, and the ability to influence sediment competence and the character of the riverbed sediment, to be directly associated with the longitudinal variations in valley-floor width. In the Kanawha River Basin, the riverbed sediment textures of both constrained FPZs (i.e., Lowland Constrained and Upland Constrained) are dominated by larger sediment classes irrespective of where they are located in the stream
44 network. By comparison, those FPZs with wider valley floor widths tended to have finer riverbed sediments. The potential of multiple catchment scale variables to influence the character of riverbed sediments illustrates the complex nature of top-down controls on the spatial organization within river ecosystems.
A majority of the field sampling sites was correctly allocated to the a priori FPZ classification of the Kanawha River Basin. An independent classification of the field data resulted in the emergence of five riverbed sediment texture groups and of these, 70 percent were correctly associated with a priori FPZs, the latter being determined via the method outlined by Thoms et al. (2004) and Williams et al. (2013). The remaining allocation mismatch is likely to have multiple origins. The study design employed was initially to determine differences in the riverbed sediment character between FPZs and not to define sediment groups de novo within the Kanawha River Basin. The study design used may have influenced the manner in which FPZs were aligned with the self-emergent riverbed sediment groups. Replacing the nested design of this study with a completely random design for sediment collection within the river network of the Kanawha River Basin and then comparing to the FPZs would improve the independence of the comparison between the two. Second, some field sites were located near transition zones between FPZs. In these transition zones, physical habitat is likely to reflect a combination of types from both upstream and downstream FPZs. Finally, the scale at which data were collected may play a role in producing differences between FPZs and self-emergent groups. Samples were collected along a reach until 100 samples were recorded. As we followed the sampling strategy outlined by Collins & Flotemersch (2013), we did not take into account stream width to determine reach length sampled (e.g., 20 times the wetted width; Rosgen, 1996)
45 or collect data until the largest particle comprised only 5 percent of the total sampled material (e.g., Mosley & Tindale, 1985). These sampling strategies are common in biological monitoring and geomorphological investigations, respectively. Changing the sampling strategy to include some of the above would with all probability improve the fidelity between a priori FPZs and self-emergent groups.
Conclusions A major implication of this work is the rapid assessment potential of river ecosystems that results from the relationship between FPZs and riverbed sediments. Here, we demonstrated that FPZs within the Kanawha River Basin contain unique riverbed sediment textural character. As such, FPZs can be used via desk-top analysis to account for some of the variability in riverbed sediments at a basin scale and to predict what the riverbed sediment texture should be as a reference for future field work efforts. A reliable, rapid assessment technique for determining differences in, and characteristics of, riverbed sediments can lead to a better understanding of the physical habitat (e.g., Newson &
Newson 2000) and biota (e.g., Minshall & Minshall 1977; Gurnell et al. 2007) that could be supported within different FPZs. This ability to account for an increased amount of natural variability in systems at larger scales will contribute to an increased understanding of structure and function of river systems. It will also contribute to the development of improved basin-scale management strategies (e.g., bioassessment and monitoring, environmental flows) and increase the ability of regulatory organizations to identify areas where undesirable changes may occur or have occurred in a system. We see FPZ determination as a highly valuable first step in physical and biological monitoring of streams and rivers.
46 Chapter 4 – The River Continuum Concept and Riverine Ecosystem
Synthesis explain patterns in system metabolism and macronutrients
This work is in progress and will be submitted for peer review as:
Collins, S.E, I. Buffam, J.E. Flotemersch, and S.F. Matter
47 Abstract Ecosystems are constrained by a balance of primary productivity and ecosystem respiration, that is, ecosystem metabolism. Freshwater systems can encompass a range of conditions resulting in a dependence on terrestrial primary productivity or total reliance on in-stream production. Nutrient availability, among other variables, regulates in-stream primary productivity in many lotic systems. Conceptual models (the River Continuum
Concept and Riverine Ecosystem Synthesis) provide predictions of ecosystem metabolism, and we used these predictions to infer nutrient concentrations. We tested these models using estimates of metabolism from Rivermet © software and data from the Kanawha
River Basin; we also determined how nutrient concentrations changed within the basin to determine if either conceptual model was a better predictor. We found that the Riverine
Ecosystem Synthesis was a better model for predicting ecosystem metabolism within this system. In this case, catchment scale and local environmental factors used as a basis for predictions of the Riverine Ecosystem Synthesis may be more important than hierarchical groupings based on linear position within a stream network to predict ecosystem metabolism. While neither model provided a robust framework for understanding the dynamics of nutrient concentrations, a combination of both factors produced a model that explained changes in some nutrients. We propose that the fast-flowing nature of the
Kanawha River Basin, as well as the overall oligotrophic condition, resulted in little difference in nutrient concentrations among sites arranged by either Functional Process
Zone or stream order. It is also possible that nutrient concentrations do not follow either model as they may be more closely linked with land use or other factors. Further, the relatively short time-frame of this study may not be indicative of longer-term patterns.
48 Introduction Metabolism in stream ecosystems can be defined based on rates of primary productivity and overall respiration. The ratio of the two (i.e., P/R) is an important ecosystem parameter describing the relative dependence on in-stream and terrestrially derived chemical energy. In brief, observed rates of primary productivity reflect the amount of carbon dioxide converted to organic carbon while respiration includes transformations back to inorganic forms of carbon. In recent years there has been a surge in interest in measuring metabolism because of its increased recognition as a highly useful metric for understanding the flux of energy in an array of ecosystems (e.g., Saleska et al.
2009, Duarte et al. 2010, Elser et al. 2010, Beaulieu et al. 2013). River system metabolism measurements, in particular, are useful for several reasons: (1) they provide an integrated measure of function over large sections of river (as a result of the natural movement and mixing of water); (2) they assess the balance between energy supply and consumption; and
(3) they can be easily and intuitively described to the general public (Young et al. 2008).
Several factors have been shown to influence rates of gross primary productivity
(GPP) and ecosystem respiration (ER) in fluvial systems. Nutrient availability, for example, can influence both GPP and ER where high nutrient levels may lead to an increase in productivity which, in turn, increases respiration (Mulholland et al., 2001). In particular,
+ - 3- ammonium (NH4 ), nitrate (NO3 ), and orthophosphate (PO4 ) are commonly monitored inorganic nutrients because, along with silicate (Si(OH)4) and carbon, these compounds provide the major elements required by primary producers (Ringuet et al. 2010). Light availability also influences GPP and can be seasonally variable but can also vary as a result of topography, vegetation, channel geometry, and hydrologic regime (Julian et al. 2008). ER
49 can be affected by temperature where higher temperatures increase respiration as well as the amount of organic carbon within the system (Sinsabaugh 1997). Increased spatial environmental heterogeneity has also been shown to increase rates of both GPP and ER
(Cardinale et al. 2002).
Predicting ecosystem functions and processes is valuable, and to that end many conceptual models have been put forth to help explain the behavior of river ecosystems
(e.g., river continuum concept, Vannote et al. 1980; serial discontinuity concept, Ward and
Stanford 1983; flood pulse concept, Junk et al. 1989; riverine ecosystem synthesis, Thorp et al. 2008). While all contribute to a conceptual understanding of the functions and processes at play (e.g., ecosystem metabolism and nutrient concentration dynamics), no model provides a comprehensive explanation of riverine ecosystem processes. We consider the ability of two of these concepts to explain observed P/R (i.e., one measure of overall system metabolism) and dissolved nutrients throughout a watershed.
The first is the River Continuum Concept (RCC; Vannote et al. 1980) which has as its central tenet that rivers are connected, longitudinal continua. As such, physical and biological processes that occur within rivers should represent a continuous gradient from headwaters to river mouths. For example, Vannote et al. (1980) describe a shift in substrate particle size from coarse boulders and cobble in headwaters to sand and fine silt in great rivers, and between these endpoints, a monotonic gradient of particle sizes is expected. Many other characteristics (e.g., temperature, biological communities, carbon sources) are predicted to follow a longitudinal pattern with regard to Strahler stream order
(SSO; Vannote et al. 1980). Several other studies have supported and built upon these ideas
(e.g., Bruns et al. 1982, Cushing et al. 1983, Minshall et al. 1983, Bruns et al. 1984). In each
50 case, the basic ideas of the RCC are supported, but ad hoc modifications such as local environmental influences or a “sliding scale” of SSOs have been proposed. The second concept is the Riverine Ecosystem Synthesis (RES; Thorp et al. 2006, 2008) which postulates that key hydrogeomorphic factors (i.e., channel width, channel sinuosity, valley width, channel-to-valley width ratio, precipitation, elevation, geology) structure the river in an array of repeatable zones or patches. These Functional Process Zones (FPZs) explain differences in ecosystem processes observed within and across systems.
Here, we compare the RCC and RES using empirical measurements of macronutrient concentrations and ecosystem metabolism within the Kanawha River Basin (USA). The RCC hypothesizes that P/R changes from much less than one in headwater streams (1-3rd order) to approximately one in mid-order reaches (4-6th order) back to less than one in large rivers (> 6th order), and this change is largely attributed to changes in light availability. The RES, on the other hand, proposes that P/R will be affected by light availability and environmental heterogeneity with those zones having the highest light availability and most homogeneous character (e.g., stable flow velocities) supporting P/R of approximately one while other zones would have P/R less than one. Neither the RCC nor the RES make clear testable predictions for variation in nutrients within stream networks.
However, based on a conceptual understanding of the factors limiting primary and secondary production in streams, we propose two relationships between productivity and nutrients. First, if light or factors other than nutrients are limiting primary productivity, then we would predict that sites with high GPP (and relatively high P/R) would give rise to low dissolved inorganic nutrient concentrations. For example, in streams where P/R is less than one, dissolved inorganic nutrients should accumulate as a result of light-limited
51 autochthonous production. However, in streams where P/R approaches one, dissolved inorganic nutrients should be relatively depleted because they would be readily consumed by photosynthesis. Alternatively, if nutrient availability is limiting to primary production, then we would predict that sites with high inorganic nutrient availability would give rise to high GPP (and relatively high P/R) and predictions of the RCC and RES would fail.
Methods
Study Area The Kanawha River Basin, located in the Mid-Atlantic region of the USA (Figure
4.1a), has a catchment area of 31,691 km2. It is a relatively pristine watershed containing mostly (>80%) a mixture of northern hardwoods, oak-pine, and mixed mesophytic forests that extend from an altitude of 1,355 m to 164 m (Messinger & Hughes, 2000). The
Kanawha River is a 6th order stream with a typical dendritic network pattern (Figure 4.1b).
It has a long-term (1961-2000) mean daily discharge of 432 m3s-1 with a range of 31 -
6,116 m3s-1, at Charleston, WV, approximately 50 km from the confluence of the Kanawha and Ohio Rivers. In this system, we are defining the 6th order sections of this drainage as
“large” (in the context of the RCC) due to their low velocity, wide channel, and great average depth which is maintained for navigation. Fourth and fifth order streams were assigned as mid-order, and second and third order streams represent headwaters based on the description of each category within the RCC (Vannote et al. 1980). Hydrogeomorphic classification of the system identified six distinct FPZs within the Kanawha River Basin
(Williams et al., 2013; Figure 4.1c). The Upland Constrained (UC) FPZ dominates the stream network of the Kanawha, accounting for 37.3% of the total network length. Other FPZs are
52
Figure 4.1 The Kanawha River Basin. a) The river basin and its location within the continental USA; b) the stream network of the basin where the Kanawha mainstem is highlighted; and, c) the location of Functional Process Zones and the riverbed sediment sampling sites labeled with Strahler stream order.
53 Lowland Constrained (LC: 18.8%), Upland High-Energy (UH: 16.8%), Reservoir (RE:
11.7%), Lowland Alluvial (LA: 11.6%) and Open-Valley Upland (OU: 3.8%). A detailed description of the study area can be found in Collins et al. (accepted).
Nutrient Data Collection We sampled 35 sites representing five SSOs (2nd to 6th order) spanning headwaters to large rivers and four FPZs (Figure 4.1c). Macronutrients were sampled in two seasons
(spring 2012 or 2013 and summer 2011 or 2012). First order streams were not sampled because of limited accessibility and unreliable water levels. The OU FPZ was not sampled due to its low representation within the watershed, and the RE FPZ was not sampled as it represents a lentic system. All summer samples were collected at flow less than the minimum monthly average; all spring samples were collected at or greater than median annual flow. Water samples were filtered (0.45 μm) to remove any particulate matter and stored in acid-washed plastic bottles. All water samples were frozen in the field. We
+ - 3- assayed samples for dissolved organic carbon (DOC), NH4 , NO3 , PO4 , total dissolved nitrogen (TDN), and total dissolved phosphorous (TDP).
Nutrient Analysis + Analysis of NH4 was based on a protocol by Weatherburn (1967) adapted for use on 96-well microplate. This colorimetric procedure uses the phenol-hypochlorite reaction
+ - for determining NH4 concentration. Analysis of NO3 was based on a single-reagent protocol adapted from Doane and Horwath (2003) to be performed on a 96-well
3- microplate. PO4 analysis was based on a 96-well microplate method adapted from Eaton et al. (2005) using H2SO4, potassium antimonyl tartrate solution, ammonium molybdate solution, and ascorbic acid. Dissolved organic carbon and total dissolved nitrogen
54 concentrations were determined using a Shimadzu TOC-VCPH analyzer. Total phosphorus was determined using a Thermo Scientific iCAP 6000 Series ICP spectrometer. All nutrient concentrations are expressed as mg l-1 of N, P, or C, respectively.
Estimation of Ecosystem Metabolism We visited a sub-sample of 21 sites in June 2010. Oxygen concentration (mg L-1) and water temperature (°C) were measured every 15 min for > 72 h using TROLL® 9500 multiparameter water quality instruments (In-Situ Inc., Ft. Collins, CO, USA). Discharge was determined from the nearest U.S. Geological Survey stream gage (range: ~0-40 km; average: ~15 km). Light intensity (lux) at a depth of approximately 1 m was monitored concurrently using HOBO Pendant® Temperature/Light data loggers (Onset Computer
Corp., Bourne, MA, USA). Productivity and respiration estimates were determined using
Rivermet©, an Excel-based tool (Izagirre et al. 2007). Rivermet© requires site locations, sunrise and sunset times, and measurements of water temperature, dissolved oxygen, and discharge to determine GPP and ER for each site.
Statistical analyses We first tested the hypothesis that macronutrient concentrations in this river network were expected to follow the changes in P/R. Where P/R is expected to be close to one (i.e., mid-order streams and the LA FPZ), inorganic nutrient concentrations should be low as high rates of primary productivity quickly removes these nutrients from the system.
To achieve normality, data from each macronutrient were log10 or square-root transformed, if needed (see Table 4.1). A paired t-test was performed to assess for seasonal differences in each macronutrient. When significant differences between seasons did not exist at a site, data from both spring and summer were used together for analysis. ANOVA
55 was used to determine if there were significant differences in nutrient concentrations between FPZs or stream order groupings and to determine the amount of variation explained by either model. Where normality could not be achieved, a non-parametric
Kruskal-Wallis one-way ANOVA was performed. In the case of statistical significance, a
Tukey HSD post hoc test for ANOVA was used, and multiple pairwise comparisons using
Mann-Whitney-Wilcoxon tests with a Bonferroni correction for Kruskal-Wallis were used.
We used a generalized linear model (GLM) to determine if a combination of both models (RCC and RES), or the interaction between models, explained more variation in the concentration of macronutrients. We tested each model against a null model to determine significance. Akaike Information Criterion (AIC) was used to determine if one model (RCC or RES) performed better in describing patterns of macronutrients; a lower AIC indicates a higher quality model, but models with ΔAIC < 2 are indistinguishable. AIC could only be calculated for data fit by maximum likelihood, in our case, the normally distributed data.
For non-normal data, we simply compared the test statistic. Non-parametric equivalents for model selection could not be performed because models were not fully crossed (i.e., not every SSO is represented in each FPZ). We looked at each macronutrient independently rather than using a multivariate approach because each macronutrient may have different sources (e.g., phosphorus may be most related to geology whereas nitrate may be linked to land use).
We grouped the sub-sample of 21 sites sampled for metabolism by FPZ or stream order categories (i.e., large rivers, mid-order streams, headwaters) according to the hypothesis of each model. To determine if the predictions of the RCC or RES better matched observed ecosystem metabolism, P/R estimates were tested against each model’s
56 predictions using a one-sample t-test (where µ = 1). Only mid-order streams and the
Lowland Alluvial zone are predicted to show P/R of approximately one. We determined if there was a significant difference between one and the estimated P/R in each group to determine where model predictions matched empirical observations.
We also used analysis of variance (ANOVA) to determine if there were significant differences in light concentrations or water temperatures between FPZs or stream order groupings. Further, we used ANOVA to determine if there were significant differences in
P/R between FPZs or stream order groupings and to see how much variance was explained by each model. We used a GLM to determine if a combination of both models (RCC and RES) explained more variation in the P/R data than either model alone. A model that explains more variation in the data, even when significant differences in P/R between groups do not exist, is inherently the better model. Finally, we used a GLM to determine if there was a relationship between productivity and nutrients at sites where both were measured. Prior to analyses, we square-root transformed P/R to achieve normality and homogeneity of variance.
We used R (version 2.15.1; R Core Team, 2013) to perform all statistical analyses.
Significance for all tests was set at α < 0.05.
Results - There were seasonal differences in DOC and NO3 (paired t-tests results, P < 0.05),
+ 3- but not NH4 , PO4 , or TDN; therefore, we used combined spring and summer concentrations for the latter nutrients. Most nutrient concentrations were relatively low and did not differ significantly when either stream order groupings (Figure 4.2) or FPZs
+ (Figure 4.3) were used as factors (Table 4.1). NH4 differed among SSO groupings (F2,67 =
57
1.2 12
1 10
) 1 - 0.8 8
0.6 6
0.4 4
Concentration (mg l (mg Concentration DOC Concentration (mg (mg Concentration DOC
0.2 2
0 0 PO4 NH4 TDN NO3Su NO3Sp DOCSu DOCSp 2nd 3rd 4th 5th 6th
Figure 4.2 Concentration of each macronutrient based on SSO. DOC concentration is on right axis. All other nutrients are on left axis. Nitrate and DOC show both spring (Sp) and summer (Su) time periods. Bars represent standard error.
58
1 10
0.9 9
0.8 8
)
1
-
) 0.7 7
1 -
0.6 6
0.5 5
0.4 4
Concentration (mg l (mg Concentration 0.3 3 DOC Concentration (mg l (mg Concentration DOC 0.2 2
0.1 1
0 0 PO4 NH4 TDN NO3Su NO3Sp DOCSu DOCSp Upland High-Energy Upland Constrained Lowland Constrained Lowland Alluvial
Figure 4.3 Concentration of each macronutrient based on FPZ. DOC concentration is on right axis. All other nutrients are on left axis. Nitrate and DOC show both spring (Sp) and summer (Su) time periods. Bars represent standard error.
59
Table 4.1 Nutrient analyses using Strahler Stream Order (SSO) grouping or Functional Process Zone (FPZ) as factor. K-W represents a Kruskal Wallis test. P-values in boldface represent statistically significant tests. AIC values in boldface represent the better (more parsimonious) model.
SSO FPZ Concentration Test Nutrient range (mg/L) Transformation Test Statistic d.f. P AIC Test Statistic d.f. P AIC 3- PO4 0.039-0.111 log(10) ANOVA F=1.94 2,67 0.15 -1.95 F=0.44 3,66 0.72 2.60 + NH4 d.l.*-0.158 square root ANOVA F=11.35 2,67 <0.01 -169.79 F=2.23 3,66 0.09 -154.13 TDN 0.029-2.674 log(10) K-W x2=3.62 2 0.16 - x2=9.49 3 0.02 - - NO3 (spring) 0.188-1.021 log(10) ANOVA F=1.36 2,32 0.26 42.63 F=2.09 3,31 0.12 41.04 - NO3 (summer) 0.032-2.153 log(10) ANOVA F=0.32 2,32 0.72 88.82 F=0.62 3,31 0.60 89.47 DOC (spring) 0.759-4.609 log(10) K-W x2=2.20 2 0.33 - x2=5.62 3 0.13 - DOC (summer) 0.895-6.580 log(10) K-W x2=1.05 2 0.59 - x2=8.82 3 0.03 -
-1 + * denotes values below detection limit of 0.02 mg l for NH4 .
- denotes that AIC could not be performed; non-normal data.
60 11.41; P = <0.001); it was highest in large rivers and lowest in headwater streams (Figure
+ 4.4). However, this change in NH4 does not reflect the expectation that mid-order streams would have the lowest concentration of nutrients resulting from the highest productivity.
TDN and summer DOC varied among FPZs. Summer DOC was lowest in the Upland High-
Energy zone (Figure 4.5, χ2 = 8.82, d.f. = 3, P = 0.03). TDN was lowest in the Upland High-
Energy zone (Figure 4.6, χ2 = 9.49, d.f. = 3, P = 0.02). Again these results do not follow the expected trend where the LA FPZ should have the lowest nutrient concentrations. Overall,
AIC analysis showed that the RCC was a better model for differentiating nutrient
3- - concentrations than the RES in two cases (i.e., PO4 and NH4 ), and the RES produced a significant model in two cases where the RCC did not (i.e., TDN and summer DOC). TDP concentrations were below the detection limit; therefore, they are not presented here.
A model using the interaction of both SSO groupings and FPZs, however, was useful in describing changes in some nutrients. The interaction between these factors captures
3- + variability in PO4 and NH4 (Table 4.2, Figure 4.7). Phosphate levels in large Lowland
Alluvial streams were significantly higher than those in mid-order streams within the
3- Lowland Alluvial zone. The concentration of PO4 in mid-order streams within the Lowland
Constrained zone was significantly higher than in both mid-order streams within the
Lowland Alluvial zone and headwater streams within the Upland Constrained zone.
Ammonium concentrations for large rivers within the Lowland Alluvial zone were significantly higher than those within several other zones and groupings. Although SSO
+ groupings produced a significant model for NH4 , the interaction between factors produced
3- a better model. The combined model was the only significant model for PO4 .
61 Table 4.2 Results from generalized linear models fitting Functional Process Zone (FPZ), Strahler stream order (SSO) group, and the interaction between FPZ and SSO group. AIC is Akaike Information Criterion which is used to determine the most parsimonious model. Null deviance shows the amount of variation in the data. Residual deviance shows how much variation in the data was not explained by the given model, and values closer to the null deviance indicate poor model fit. Test statistics were computed using a GLM comparing each model versus the null model. The best model is in boldface.
Deviance Statistics Nutrient Variable AIC ΔAIC Null Residual F d.f. P SSO group -1.99 13.67 3.76 3.55 1.94 2,67 0.15
3- FPZ 2.62 18.28 3.76 3.69 0.44 3,66 0.72 PO4 SSO group+FPZ -2.62 13.04 3.76 3.23 2.09 5,63 0.07 SSO group*FPZ -15.66 0.00 3.76 2.46 6.35 3,61 <0.01 SSO group -170.36 5.99 0.43 0.32 11.35 2,67 <0.01
+ FPZ -154.73 21.62 0.43 0.39 2.23 3,66 0.09 NH4 SSO group+FPZ -168.32 8.03 0.43 0.30 5.37 5,63 <0.01 SSO group*FPZ -176.35 0.00 0.43 0.25 4.51 3,61 <0.01 SSO group 42.64 1.60 5.98 5.51 1.36 2,32 0.26
- FPZ 41.04 0.00 5.98 4.97 2.09 3,31 0.12 NO3 (spring) SSO group+FPZ 44.84 3.80 5.98 4.95 1.22 5,28 0.33 SSO group*FPZ 42.69 1.65 5.98 3.92 2.27 3,26 0.10 SSO group 97.23 0.00 26.49 26.23 0.32 2,32 0.72
- FPZ 97.27 0.04 26.49 24.79 0.62 3,31 0.60 NO3 (summer) SSO group+FPZ 101.09 3.86 26.49 24.68 0.43 5,28 0.82 SSO group*FPZ 104.20 6.97 26.49 22.72 0.74 3,26 0.53
62
Table 4.3 Metabolism predictions and results. Values for primary productivity and respiration were estimated using RIVERMET © and are presented as oxygen concentration in g m-2 d-1. Estimates are compared to proposed values from either the Riverine Ecosystem Synthesis based on Functional Process Zones (FPZs) or the River Continuum Concept based on Strahler Stream Order (SSO) groupings.
Riverine Ecosystem Productivity Respiration Predicted Actual Value Statistics Matches
Synthesis (mean ± SE) (mean ± SE) P/R Value (mean ± SE) (t, df, P) Prediction
Lowland Alluvial 10.6 ± 9.2 52.2 ± 46.6 ≈ 1 0.57 ± 0.40 -1.06, 2, 0.39 Yes
Lowland Constrained 5.0 ± 1.9 16.1 ± 10.1 < 1 0.52 ± 0.11 -4.34, 5, <0.01 Yes
Upland Constrained 8.3 ± 2.6 15.6 ± 4.0 < 1 0.51 ± 0.07 -7.09, 6, <0.01 Yes
Upland High-Energy 3.2 ± 0.3 12.0 ± 1.2 < 1 0.28 ± 0.04 -17.09, 4, <0.01 Yes
River Continuum Productivity Respiration Predicted Actual Value Statistics Matches
Concept (SSO) (mean ± SE) (mean ± SE) P/R Value (mean ± SE) (t, df, P) Prediction
Large (> 6) 8.4 ± 6.9 39.8 ± 35.2 < 1 0.62 ± 0.29 -1.33, 3, 0.28 No
Mid-order (4-5) 5.5 ± 1.0 14.4 ± 4.6 ≈ 1 0.47 ± 0.07 -9.38, 12, <0.01 No
Headwater (1-3) 7.7 ± 4.6 19.1 ± 5.3 < 1 0.32 ± 0.10 -6.66, 3, <0.01 Yes
63
+ Figure 4.4 Ammonium concentration (NH4 ) does not vary significantly when FPZ was used (F3,66 = 2.28; P > 0.05; b) but does when SSO grouping was used as a factor (F2,67 = 11.41; P < 0.05; a). Tukey’s HSD shows that each category is significantly different from the others when Strahler stream order (SSO) grouping was used as a factor. Bars represent standard error of the mean.
64
Figure 4.5 Summer dissolved organic carbon (DOC) varies significantly (χ2 = 8.82, d.f. = 3, P = 0.03) when Functional Process Zone (FPZ) was used as a factor (a) but not when SSO was used (χ2 = 1.05, d.f. = 2, P > 0.05; b); however post-hoc Mann-Whitney-Wilcoxon tests with a Bonferroni correction for multiple pairwise comparisons showed that many FPZs have similar DOC concentrations. FPZs that are not significantly different share a letter. Bars represent standard error of the mean.
65
Figure 4.6 Total dissolved nitrogen (TDN) differs significantly (x2 = 9.49, d.f. = 3, P < 0.05) when Functional Process Zone (FPZ) was used as a factor (a), but not when SSO grouping was used (χ2 = 3.62, d.f. = 2, P = 0.16; b). Post-hoc Mann-Whitney-Wilcoxon tests with a Bonferroni correction for multiple pairwise comparisons showed that significant differences exist, and FPZs that are not significantly different share a letter. Bars represent standard error of the mean.
66
0.09 ‡* †˘ 0.08 ‡*
† Headwater-LC
)
1 0.07 - Headwater-UC 0.06 ‡ Headwater-UH 0.05 †* Mid-order-LA 0.04 † Mid-order-LC Mid-order-UC 0.03 * ˘ Mid-order-UH
Concentration (mg l (mg Concentration 0.02 ‡ Large-LA 0.01 Large-LC 0 PO4 NH4
Figure 4.7 Concentrations of phosphate and ammonium varied significantly when the interaction between both Strahler stream order groupings and Functional Process Zones were used (Table 4.2). LA = Lowland Alluvial; LC = Lowland Constrained; UC = Upland Constrained; UH = Upland High- Energy. Tukey’s HSD post-hoc test was used to compute differences. Columns for each nutrient that share a symbol differ significantly (P < 0.05).
67 The ratio of productivity to respiration matched the predictions of the RES in all
FPZs (Table 4.3). Specifically, in three FPZs (i.e., Lowland Constrained, Upland Constrained, and Upland High-Energy), P/R was expected to be significantly less than one due to high environmental heterogeneity (e.g., substrate, Figure 3.2), and each of these FPZs is predicted to have less light availability as a result of relatively narrow, steep valleys. In the
Lowland Alluvial FPZ, P/R was not expected to differ significantly from one because of relatively low water velocity, low variation in flow rates, wide valleys, and primary productivity in both the benthos and water column. Interestingly, though, neither total light levels nor average lighted time per day (as determined by first and last light recordings at a site) were significantly different among FPZs (F3,43 = 1.57, P = 0.21; F3,43 = 1.5, P = 0.23, respectively). However, water temperatures were significantly lower in the Upland High-
Energy than in all other zones, and temperatures in the Upland Constrained zone were lower than those in the Lowland Alluvial zone (F3,17 = 18.0, P = <0.01).
Alternatively, predictions of the RCC were not always met within SSO groupings
(Table 4.3). P/R of approximately one was predicted to occur only in mid-order streams; however, we found this was the case in large rivers. In both mid-order and headwater streams, P/R was significantly less than one. Again, measured light levels did not differ between SSOs (F3,43 = 0.62, P = 0.6), and the lowest water temperatures were found in headwater streams (F2,18 = 5.5, P = 0.01).
Although significant differences in P/R do not exist among FPZs (F3,17=0.81, P=0.51) or SSOs (F2,18=0.71, P=0.51), predictions from the RES were correct in each case. Further, a model using both SSO groupings and FPZs did not explain significantly more variation in the data than either model alone (Table 4.4). The mechanisms proposed under
68
Table 4.4 Results from generalized linear models fitting P/R data to Functional Process Zones (FPZ), Strahler stream order (SSO) group, and the interaction between FPZ and SSO group. AIC is Akaike Information Criterion which is used to determine the most parsimonious model. Null deviance shows the amount of variation in the data. Residual deviance shows how much variation in the data was not explained by the given model, and values closer to the null deviance indicate poor model fit. Test statistics were computed using a GLM comparing each model versus the null model. Note that no models are significant, but the “best” models are in boldface.
Deviance Statistics Variable AIC ΔAIC Null Residual F d.f. P FPZ -0.41 0.78 0.86 0.75 0.81 3,17 0.51 SSO group -1.19 0.00 0.86 0.79 0.71 2,18 0.51 SSO group+FPZ 2.48 3.67 0.86 0.71 0.61 5,14 0.69 SSO group*FPZ 3.58 4.77 0.86 0.68 0.61 3,14 0.44
Table 4.5 Results from GLMs fitting productivity data to summer nutrient concentrations at sites where both data were collected during the same year. There were no significant relationships between nutrients and productivity. The parameter B (±SE) is the slope of the relationship.
Statistics Nutrient B F d.f. P
3- PO4 113.12 (142.27) 0.60 1,19 0.45 - NO3 -2.22 (3.72) 0.29 1,19 0.59 + NH4 80.17 (51.82) 2.39 1,19 0.14 DOC 1.69 (1.04) 2.63 1,19 0.12 TDN -0.37 (3.03) 0.02 1,19 0.90
69 the RES (i.e., that P/R should be a function of light levels and environmental heterogeneity) were not always met. For instance, the Lowland Constrained zone would be to a have similar P/R to the Lowland Alluvial zone as they do not differ significantly in variation in flow rates (see Appendix I). The hypothesis that nutrient concentrations would change with P/R was also not supported. No significant relationships were found between nutrient concentrations and productivity (Table 4.5). The RCC and the RES model, nor their underlying mechanisms, were not conclusively supported.
Discussion
Model Performance Overall, neither assumptions based on the RCC nor the RES accurately described changes in nutrient concentrations within the Kanawha River Basin (Table 4.1). Although there are cases in which one model out-performs the other, neither provides a comprehensive explanation for the measured nutrient concentrations. A combination of both models performed better than either model independently for ammonium and phosphate (Table 4.2, Figure 4.7) indicating that these are a function of both the position in the stream network and hydrogeomorphology. It is possible that factors other than those used by either model are governing nutrient dynamics within this system, but it is more likely that relatively low nutrient concentrations throughout the Kanawha River Basin left us unable to detect significant differences among groups (SSOs or FPZs) in many cases.
This system may not be well-suited for testing these models on the basis of nutrient concentrations, or a long-term study may be needed.
Predictions for metabolism by the RCC were not well-supported by this system
(Table 4.3). Additionally, we could not show that light (the main tenet for changes in P/R
70 from the RCC) differed among SSOs. Our ability to quantify “light” was limited. We used point measurements of light intensity rather than photosynthetically active radiation over an entire reach. Thus, our comparisons may not accurately reflect variation in light available for primary productivity among sites.
The predictions for metabolism by the RES, on the other hand, were supported in this system. The Lowland Alluvial zone had matching rates of production and respiration, and this was expected as a result of increased primary productivity from both benthic and pelagic sources in the deep and relatively stable water column. A decrease in the proportional contact area between the water column and biochemically active substrates may lead to increased residence time for nutrients in the Lowland Alluvial zone which could support higher rates of primary productivity. However, the Lowland Constrained zone had similar levels in some measures of environmental heterogeneity (e.g., flow, Figure
AI.2), and this should result in similar P/R estimates in this zone. These factors do not appear to be controlling P/R, thus the perceived mechanism of the RES was not well- supported. Factors of environmental heterogeneity other than those measured may be influencing P/R in this system.
In many cases spatial changes in nutrients do not seem to relate to the metabolism measurements within this network (Table 4.5). The expectation that nutrient concentrations would change commensurate with changes in metabolism was not met.
Changes in nutrients at the site scale may be related to land use or other anthropogenic drivers, although the majority (> 80%) of the watershed is forested.
71 Nutrients Nutrient concentrations within the Kanawha River Basin are quite low throughout the system. Dodds et al. (1998) suggest an oligotrophic-mesotrophic boundary for total nitrogen to be 0.7 mg l-1 and for total phosphorus to be 0.025 mg l-1, while the mesotrophic- eutrophic boundary is suggested to be 1.5 mg l-1 and 0.075 mg l-1 for total nitrogen and total phosphorus, respectively. By these standards, over 90% of sites visited within the
Kanawha River Basin are oligotrophic with regard to nitrogen and roughly 85% of sites are
3- mesotrophic with regard to phosphorus (using PO4 as a surrogate as TDP was below detection limit). Although trophic enrichment often leads to an increase in algal biomass
(Dodds et al. 1997, Lohman et al. 1992, Welch et al. 1992), nitrogen and phosphorus concentrations within Kanawha River Basin rarely cross the thresholds of eutrophication.
TDN was only measured above 1.5 mg l-1 at two of 35 total sites visited, and potential
3- eutrophic status related to phosphorus was only seen at 10 sites (based on PO4 ). While
+ concentrations of NH4 and TDN vary significantly when either stream order groupings or
FPZs are used as factors, this variability does not appear to be related to variation in
+ primary productivity in this river network. NH4 concentration is highest in large order
+ streams; however, trends in primary productivity do not seem to cause this change in NH4
+ (Table 4.3, Figure 4.4). Alternatively, this increase in NH4 may be facilitating high rates of primary productivity in large streams. Similarly, productivity does not reflect commensurate changes in TDN in various FPZs (Table 4.3, Figure 4.6). Overall, primary productivity does not appear to be nutrient limited in the Kanawha River Basin; further, the lack of significant differences in measured light levels do not help explain changes in productivity.
72 Nutrient dynamics, though, are difficult to understand with a point-in-time sample.
Short term sampling strategies and experiments are valuable because they provide the ability to assess ecological processes and functions that operate on truncated time scales.
However, these brief surveys do not provide adequate information to assess long-term ecological processes that may be representative of ecosystem-wide environmental changes
(Shindler 1998, Lindenmayer and Likens 2010, Dodds et al. 2012). Further, the variation in some nutrient concentrations explained by the interaction between SSO grouping and FPZ confirms that a combination of factors (e.g., longitudinal position and hydrogeomorphic variables) is likely responsible for changes in nutrient dynamics within a system.
Metabolism The pattern of variation in GPP throughout the basin is unexpected. Whether stream order categories or FPZs are used, GPP does not change predictably based on nutrient concentration or light availability. Neither total light (lux) per day nor average lighted day length explain changes in GPP when FPZ or stream order groupings are used. Channel width is greater in large rivers and Lowland Alluvial zones which could lead to greater light exposure due to proportionally less shading by riparian vegetation; however, this alone does not account for all the changes in GPP. Although GPP is high in both these categories
(i.e., large rivers, Lowland Alluvial zones), it is nearly equally high in headwater streams
(compared to large rivers) and the Upland Constrained zone (compared to the Lowland
Alluvial zone). Further, seasonally derived heterogeneity in light conditions may also play a role in this system. Specifically, streams that are predominately shaded in the summer and fall as a result of leaf cover may have high light levels in the spring and winter. However, these seasonal effects were not addressed as part of this study.
73 Changes in ER, too, are substantial throughout the basin. Low rates of respiration were found in Upland High-Energy where the lowest amounts of DOC were also found
(Figure 4.5). Significantly higher amounts of DOC and higher rates of respiration were found in other zones, and this supports the idea that higher energy and resource availability may yield an increase in food chain length (Kaunzinger and Morin 1998, Post et al. 2000, Slobokin 1961). The Upland High-Energy zone has lower food chain length than all other zones (Figure 5.2b). As food chain length increases, so too should biological oxygen demand (thereby reducing dissolved oxygen concentrations) which may influence estimated rates of ER. The Upland High-Energy zone also had significantly lower average temperatures during the study period, and lower temperatures are known to suppress metabolic rates (Gillooly et al. 2001). Additionally, variability in ER was highest in large rivers. Perhaps this increase in variability is a realized accumulation from multiple sources within the watershed.
The ratio of GPP to ER, though, changes predictably within the framework of the
RES. In each FPZ, predictions of P/R are met within the Kanawha River Basin. The Lowland
Alluvial zone has the highest P/R value, while P/R was significantly lower than one in each other zone. The Lowland Alluvial zone was predicted to have GPP and ER in balance due to an increase in primary productivity resulting from high amounts of light and low environmental heterogeneity. Further, this zone had the most stable flow velocities which have been shown to increase primary productivity (Townsend and Padovan 2005). Other zones had P/R less than one, and this could be the result of lower temperatures, unmeasured differences in shading, nutrient limitations (which could reduce GPP), or
74 higher rates of respiration in relation to GPP. Increased temperature, as well as quantity and quality of organic resources, can elevate rates of ER.
Predictions from the RCC, on the other hand, did not match well with measurements from the Kanawha River Basin. Only predictions for headwater streams (i.e., P/R << 1) matched the data. It is likely that light and nutrient limitations in headwater streams reduce potential in-stream productivity, while respiration rates are driven by terrestrial subsidies (see also Chapter 5). It is also likely that the overall trend of the RCC (i.e., higher
P/R with an increase in stream order) is also supported by this study. Perhaps the “sliding scale” (Minshall et al. 1983) is applicable in the Kanawha River Basin, and higher rates of primary productivity (relative to respiration) may be achieved in streams larger than sixth order.
Conclusions The Kanawha River Basin was selected for this research project because it is a rather intact river network and is spatially compact. That is, the basin has been minimally affected by anthropogenic influences, and it contains a variety of FPZs and SSOs in a relatively small area. Although the condensed nature of this system allowed for a robust sampling design for testing predictions from both the RCC and RES, there were also some limitations. For example, the variety of FPZs and SSOs within the Kanawha River Basin was likely due to the overall high-gradient nature of the network. However, this high-gradient nature also reduced residence time of water at sites which may have reduced the ability to detect differences in metabolism and nutrient concentrations. Further the mesotrophic
(often bordering on oligotrophic) nature of the system possibly reduced the ability to detect any changes in nutrient levels and those associated differences in rates giving rise to
75 P/R. To alleviate these problems, we suggest performing similar research in a system with longer residence times (or at least an increase in variability thereof) and with higher nutrient loads.
76 Chapter 5 – Using stable isotope analysis of aquatic food webs to assess predictions of the River Continuum Concept and the Riverine Ecosystem
Synthesis
This work is in progress and will be submitted for peer review as:
Collins, S.E, M.D. Delong, J.E. Flotemersch, and S.F. Matter
77 Abstract Stream communities in various environments have diverse trophic structure and utilize a combination of organic carbon from in-stream and terrestrial sources. Food web structure consists of food chain length, variety of basal carbon sources, and degree of niche overlap between community members. Food chain length may respond to rates of primary productivity or ecosystem size. Organic carbon availability, including dependence on terrestrial sources, varies as a result of limited in-stream primary productivity due to inorganic nutrient deficits or limited light availability. Niche overlap occurs when the diets of several members of the community contain similar resources. Here, we compare hypotheses proposed by two prominent conceptual models, the River Continuum Concept
(RCC) and the Riverine Ecosystem Synthesis (RES), using stable isotope analysis of primary producers and consumers from thirty-five sites within the Kanawha River Basin, USA.
Stable isotopes of carbon and nitrogen were analyzed, and several community metrics were determined for each site. The proportional dependence upon terrestrial and aquatic carbon sources was also assessed. Our results show that community metrics and organic carbon source can be partially explained by both the RCC and RES. While the RCC provides a better framework than the RES for understanding some changes in community metrics and organic carbon sources, neither model is complete. Although the RCC was a better predictor, the mechanisms underlying those predictions (e.g., high rates of in-stream productivity should result in low use of terrestrial carbon subsidies) were often not correct. In some cases, a combination of both models produces a significant and more parsimonious model than either individually.
78 Introduction Understanding food web structure and organic carbon sources in river systems is difficult, though important. First, identification of this structure can lead to a better understanding of trophic interactions between producers and consumers, utilization of carbon sources, and nutrient dynamics (de Ruiter et al. 2005, Pingram et al. 2012,
Woodward and Hildrew 2002). Second, streams of various sizes and across ecosystem types rely upon different sources of organic carbon. Some streams primarily utilize autochthonous production, while others depend heavily on terrestrial subsidies.
Communities use proportionally more allochthonous materials due to limited in-stream production because of limited inorganic nutrient concentrations (Smith et al. 2003) and proportionally high shading as a result of over-reaching forest canopies. Although challenging, gaining a better understanding of these processes is necessary to increase the understanding aquatic systems for protection, management, and mitigation of these habitats.
Food web structure is a multifaceted concept in streams and rivers, and there are many measureable components within this structure. Layman et al. (2007) described a suite of useful metrics for understanding community-wide trophic structure. Food chain length can be used to infer to total number of trohpic linkages within a community which may increase with increased primary productivity (Pimm 1982, Kaunzinger and Morin
1998), increased ecosystem size (Cohen and Newman 1991), or both (Slobodkin 1961,
Schoener 1989). Carbon source diversity can also be examined to help understand the use of multiple sources of energy. The total area of the food web (a measure of the overall niche space) is an estimate of the trophic diversity of the community. Finally, nearest neighbor
79 distance (NND) between community members is used to determine the extent of trophic overlap within a food web. Small NNDs are the results of food webs with many species occupying the same or similar trophic ecologies or niche requirements.
Additionally, both terrestrial and aquatic carbon (energy) sources play a role in the structure of aquatic food webs. Overall system metabolism (i.e., the ratio of primary productivity to respiration) can be assessed by determining how much terrestrial carbon is utilized in an aquatic food web. Due to differences in metabolic pathways and environmental conditions, carbon varies isotopically between terrestrial and aquatic sources (Rau 1980, Rounick et al. 1982). Understanding the sources of carbon and the food web’s proportional dependence upon terrestrial and aquatic carbon sources is key to understanding how aquatic food webs function as well as the overall biogeochemistry of these environments (Doucett et al. 2007).
Stable isotope analysis (SIA) is a commonly used method for trophic investigations
(Angradi 1994, Boecklen et al. 2011, Doucett et al. 1996, Fry 1991, Herwig et al. 2007) as well as determining the relative importance of various carbon sources in aquatic systems
(Babler et al. 2011, Cole et al. 2011, Doucett et al. 2007). Information gained from SIA of isotopes of nitrogen (δ15N) and carbon (δ13C) can be used to calculate a variety of metrics
(e.g., Layman et al. 2007, Jackson et al. 2011) to describe food chain length, basal resources, and the occurrence of niche overlap within the food web.
Here, two widely cited models are tested for their applicability to food web structure and carbon source dependence. Although many conceptual models detailing stream functions and processes have been developed, in most cases, empirical support is lacking. The River Continuum Concept (RCC; Vannote et al. 1980) describes a longitudinal
80
transition from headwaters to large rivers with gradual and predictable changes in carbon sources and food web dynamics. In headwaters (1st-3rd order), stream communities are thought to depend heavily on terrestrial subsidies as a result of limited primary productivity within the stream due to shading from riparian trees. Mid-order streams (4th-
5th order) should depend less on terrestrial carbon sources, due greater light availability and large rivers (> 6th order) again have limited primary productivity due to light attenuation with depth and so depend more on terrestrial carbon. The Riverine Ecosystem
Synthesis (RES; Thorp et al. 2006, 2008), on the other hand, uses various environmental factors (e.g., precipitation, topography, geomorphology) to make predictions about stream functions and processes. Here, streams within functional process zones (FPZs) with high environmental heterogeneity (i.e., variation in flow and physical habitat) should depend upon terrestrial carbon sources much more than those with low environmental heterogeneity because of limited primary productivity as a result of high stochasticity and disturbance. In the case of both models, food web structure should respond to both productivity and ecosystem size (Takimoto and Post 2013). Large streams and those with high amounts of primary productivity should yield longer food chain length and total food web area. These streams should also support more diverse communities, and thus, smaller
NNDs. We used SIA to develop food webs for 35 sites within the Kanawha River Basin. In each case, food web structure and organic carbon sources were tested against model predictions.
81 Methods
Study area The Kanawha River Basin, located in the Mid-Atlantic region, USA, (Figure 5.1a) represents an excellent study area for testing tenets from the RCC and RES for several reasons. Both these models propose a set of hypotheses for unimpeded watersheds within forested systems. The Kanawha watershed is very close to natural flow conditions (i.e., only
9 dams within the watershed), it is mainly forested (> 80%), and minimally impacted by agriculture or industry. The catchment area is 31,691 km2, and the long-term (1961-2000) mean daily discharge is 432 m3 s-1 at Charleston, WV (Messinger & Hughes 2000). Thirty- five sites representing 2nd through 6th order streams (i.e., Strahler stream order, SSO) and four FPZs (i.e., Lowland Alluvial, Lowland Constrained, Upland Constrained, Upland High-
Energy) were chosen (Figure 5.1b). Although six FPZs exist within the basin Open-Valley
Upland and Reservoir zones were not sampled. Open-Valley Upland only make up less than
4% of the streams within this basin, and Reservoir zones do not represent lotic systems.
Organism sampling and analysis Aquatic and terrestrial primary producers were collected once at each site during the summers (June-August) of 2010, 2011, or 2012. Four sites were sampled concurrently for two years to determine if stable isotope signatures were similar over this time scale
(they were; paired t-tests, P > 0.05; Appendix II). Dead leaves from two or three of the predominant tree species found within the riparian zone at each site were collected from the ground. Organic detritus was gathered from within the bankfull width of each stream.
Benthic algae were collected by scrubbing then rinsing rocks or woody debris.
82
Figure 5.1 (a) The approximate location of the Kanawha River Basin within the Mid-Atlantic region, USA; and (b) the location of Functional Process Zones and sampling sites.
83 Aquatic macrophytes were collected by hand. Benthic macroinvertebrates were collected by hand from four targeted primary consumer groups: mayflies (Ephemeroptera), caddisflies (Trichoptera), clams (Corbiculidae) and snails (Caenogastropoda and
Heterobranchia). Approximately 15 individuals in each group were collected per site to allow for composited samples as individuals were often too small for analysis. Fishes from as many representative feeding guilds (e.g., herbivores, insectivores, piscivores) as possible at each site were collected using a 5.0 GPP Smith-Root boat-mounted electrofishing unit or a Smith-Root LR-24 Electrofisher backpack (Smith-Root, Vancouver,
Washington, USA), depending on site conditions. Up to five replicates for each species were collected where possible. Fishes > 35 cm were identified on site, and a dermal punch was used to remove approximately 2 g of tissue from above the lateral line and below the dorsal fin. Smaller individuals were ethically euthanized under University of Cincinnati IACUC protocol 11-11-15-01. All samples were placed in portable freezers for transfer to the laboratory and stored at -20° C until processing for stable isotope analysis.
All benthic macroinvertebrates and fish not identified in the field were identified to the lowest possible taxon in the laboratory. For fish where a tissue sample was not collected in the field, a dorsal muscle tissue sample was collected. All samples were dried at
60° C for 24-48 h and ground to a fine powder in a mortar and pestle. Tin capsules were packed with approximately 0.3 µg animal tissue or 0.6 µg plant tissue and combusted to gas in a ThermoQuest NC2500 elemental analyzer coupled to a Finnigan Delta Plus for isotope analysis of carbon (C) and nitrogen (N). In-house standards including ammonium sulfate, glutamic acid, sulfanilamide, spiders, and leaves were used for quality assurance/quality control and to encompass the expected range of isotopes in animal and plant samples. Lipid
84 extraction was not performed as it has been shown to yield only a 0.64‰ difference in C
(Boecklen et al 2011).
Community metrics Layman et al. (2007) described several metrics for understanding community-wide trophic structure, and Jackson et al. (2011) provided an updated method for calculating these metrics in a Bayesian framework. This method takes into account the inherent uncertainty of sampled data and also removes sample size bias allowing for more widespread comparison of results (Jackson et al. 2011). A large δ15N (i.e., the ratio of 15N to
14N) range for consumers implies a longer food chain and potentially more trophic diversity. Likewise, a broad δ13C (i.e., the ratio of 13C to 12C) range suggests a wide variety of basal resources. The total area of the food web (a measure of the overall niche space) represents the overall trophic diversity of the food web. NND is used to determine the extent of trophic overlap within a food web. Small NNDs are the result of food webs with many species occupying the same or similar trophic ecologies or niche requirements.
Isotope modeling Food web and community metrics for each site were calculated using the SIAR
(Stable Isotope Analysis in R) package (version 2.15.1; R Core Team, 2013) based on
Layman et al. (2007) as updated by Jackson et al. (2011). Calculated metrics include nitrogen (δ15N) range (i.e., food chain length), carbon (δ13C) range (i.e., variety in basal resources of organic C), hull area (i.e., total niche space), and NND (i.e., degree of niche overlap within food webs).
Proportional dependence on sampled aquatic and terrestrial carbon sources was determined for each consumer group at each site also using the SIAR model. The mean
85 value, along with the 1st and 99th percentile, for each source was calculated for each primary consumer (Phillips and Gregg 2003). A calculated value greater than zero for the
1st percentile can be used to show a potential diet source for each consumer. We chose to use SIAR because the Bayesian framework can circumvent some of the limitations of other models (Parnell et al. 2010).
Statistical analysis Calculated community metric data were normally distributed and had homogeneity of variance. Differences in community metrics among stream orders or FPZs were assessed using ANOVA. If ANOVA was found to be significant, a Tukey’s HSD post-hoc test was used to determine differences between groups. To determine if SSO or FPZ better explained patterns in food web structure data, we compared models using generalized linear models
(GLMs). We developed a GLM to compare the performance of each model independently as well as a combined model with both SSO and FPZ as factors, as well as their interaction. We tested each model against a null model to determine significance. Model selection was based on Akaike Information Criterion (AIC). ΔAIC > 2 indicates a less parsimonious model.
We also used a GLM to determine if measures of heterogeneity (i.e., coefficient of variation for flow, substrate; Appendix I) were redundant when using FPZ as a factor. If measures of heterogeneity explain overlapping variation in these data when compared to FPZ, the mechanism proposed by the RES would be supported in their effect on food web structure.
86 Results
Community metrics In the context of the RCC, the lowest nitrogen range (NR) and carbon range (CR) were within 2nd order streams (Figure 5.2a), supporting the hypothesis that small streams with limited productivity (Table 4.3) would have the shortest food chain length and most limited basal resources. The Upland High-Energy zone had the lowest NR (Figure 5.2b).
Again, the hypothesis that zones with high heterogeneity (see Appendix I) would have the lowest food chain length was supported. NR varied with regard to both stream order and
FPZ (F4,30 = 4.65, P = <0.01; F3,31 = 5.44, P = <0.01, respectively). CR varied only with regard to stream order (F4,30 = 6.53, P = <0.01). Second order streams had the lowest total hull area, and fifth order streams had the highest area (Figure 5.2a). Similarly, the Upland High-
Energy zone had a lower total hull area compared to the Lowland Constrained zone (Figure
5.2b). Hull area also varied with both stream order and FPZ (F4,30 = 8.80, P = <0.01; F3,31 =
3.66, P = 0.02, respectively). Finally, second order streams had higher NND than any other stream order (Figure 5.2a). This confirms that small streams with limited productivity do not support large communities with a high degree of trophic overlap between community members. NND only varies with regard to stream order (F4,30 = 6.12, P = <0.01).
Results from model-fitting exercises show that SSO and FPZ differentiate between some community metrics (Table 5.1). When SSO was used, a significant model was produced in all four instances (i.e., NR, CR, area, NND), and when FPZ was used, a significant model was produced with two metrics (i.e., NR and area). Also, the RCC explained more variation (as determined by residual deviance) in every case (Table 5.1).
87
18 180 B B a 16 B 160 AB B B 14 AB C 140 A A 12 120 B 2nd B 100 10 A AB 3rd 80 8 4th A 60 MetricValues 6 5th 4 A 40 6th B B B B 2 20 0 0 NR CR NND Area Community Metrics
18 180 A b 16 160 AB A 14 A 140 12 B 120 AB UH 10 AB 100 B UC 8 80 LC 60
Metric Values Metric 6 LA 4 40 2 20 0 0 NR CR NND Area
Community Metrics
Figure 5.2 Food web metrics respond to stream order and Functional Process Zone (FPZ). NR is δ15N range; CR is δ13C range; NND is nearest neighbor distance which provides an estimate of trophic diversity. Units on this axis represent a ‰ change in isotopes of carbon or nitrogen. Area is represented on the right axis. Columns of the same measure (e.g., nitrogen range) that share a letter do not differ significantly. Error bars represent standard error of the mean. (a) Nitrogen range (F4,30 = 4.65, P = 0.0048), carbon range (F4,30 = 6.53, P = 0.00066), hull area (F4,30 = 8.80, P = 0.000008), and NND (F4,30 = 6.12, P = 0.001) differ significantly among stream orders. (b) Nitrogen range (F3,31 = 5.44, P = 0.0040) and hull area (F3,31 = 3.66, P = 0.023) differ significantly among FPZs (where LA is Lowland Alluvial, LC is Lowland Constrained, UC is Upland Constrained, and UH is Upland High- Energy).
88 Table 5.1 Results from generalized linear models fitting Functional Process Zone (FPZ), Strahler stream order (SSO) group, and the interaction between FPZ and SSO. Null deviance shows the amount of variation in the data. Residual deviance shows how much variation in the data was not explained by the given model, and values closer to the null deviance indicate poor model fit. Test statistics were computed using ANOVA. NR is δ15N range; CR is δ13C range; NND is nearest neighbor distance which provides an estimate of trophic diversity. SSO explained more variation in the data for each community metric and produced a significant model in all cases. The best model is in boldface. Test statistics were computed using a GLM comparing each model versus the null model Deviance Statistics Metric Variable AIC ΔAIC Null Residual F d.f. P SSO 164.37 10.77 258.16 159.32 4.65 4,30 <0.01 FPZ 164.47 10.87 258.16 169.17 5.44 3,31 <0.01 NR SSO+FPZ 155.41 1.81 258.16 103.90 5.73 7,26 <0.01 SSO*FPZ 153.60 0.00 258.16 74.27 4.54 5,22 <0.01 SSO 170.25 0.00 352.75 188.46 6.54 4,30 <0.01 FPZ 186.69 16.44 352.75 319.20 1.09 3,31 0.36 CR SSO+FPZ 171.83 1.58 352.75 166.09 4.34 7,26 <0.01 SSO*FPZ 179.01 8.76 352.75 153.26 2.39 5,22 0.03 SSO 331.64 2.66 41217 18960 8.80 4,30 <0.01 FPZ 346.20 17.22 41217 30430 3.66 3,31 0.02 Hull area SSO+FPZ 328.98 0.00 41217 14803 6.88 7,26 <0.01 SSO*FPZ 334.62 5.64 41217 13071 3.95 5,22 <0.01 SSO 42.21 0.00 9.91 4.86 7.80 4,30 <0.01 FPZ 61.76 19.55 9.91 8.99 1.05 3,31 0.38 NND SSO+FPZ 42.66 0.45 9.91 4.15 5.36 7,26 <0.01 SSO*FPZ 45.26 3.05 9.91 3.36 3.58 5,22 <0.01
Table 5.2 Results from generalized linear models fitting Functional Process Zone (FPZ), and measures of environmental heterogeneity (i.e., coefficient of variation in flow, substrate) Null deviance shows the amount of variation in the data. Residual deviance shows how much variation in the data was not explained by the given model, and values closer to the null deviance indicate poor model fit. Test statistics were computed using ANOVA. NR is δ15N range. Measures of environmental heterogeneity did not explain significantly more variation in these data than FPZ alone which supports the hypothesis that FPZs may use these measures along with others to show differences in community metrics. Deviance Statistics Metric Variable Null Residual F d.f. P FPZ 258.16 169.17 5.44 3,31 <0.01 NR Flow 258.16 257.52 0.08 1,30 0.78 Substrate 258.16 249.81 <0.01 1,29 0.99 FPZ 41217 30430 3.66 3,31 0.02 Hull area Flow 41217 41217 <0.01 1,33 0.98 Substrate 41217 37121 3.64 1,29 0.07
89
This shows that the RCC (i.e., SSO) provides a more robust framework for determining changes in food web community metrics within this system. Importantly, though, when both models produce significant differences independently (e.g., NR, area), the interaction between models is not significant. This result shows that each model explained different variation within the dataset, and the interaction between models was not important.
Finally, measures of heterogeneity did not explain significantly more variation in the data than FPZ (Table 5.2), nor were models built with FPZ and both measures of heterogeneity significantly different than models using FPZ alone for NR (F2,29 = 2.90, P = 0.07) or hull area (F2,29 = 2.81, P = 0.07).
Carbon sources Terrestrial and aquatic carbon sources were examined at each site to determine if patterns emerged across sites as to the extent that each source contributed to food webs.
Using the perspective of the RES, for the Lowland Alluvial FPZ, aquatic vascular plants were the most important resource showing up in the diet of 80% of the sampled macroinvertebrates (Table 5.3). This supports the hypothesis that zones with low environmental heterogeneity should have high amounts of in-stream primary productivity.
Benthic algae and tree leaves were also represented, although minimally. Terrestrial detritus was not shown to be a part of the diet of primary consumers within the Lowland
Alluvial zone, and this may be a result of a relatively limited spatiotemporal connection with terrestrial areas within this zone. In the Lowland Constrained FPZ, aquatic vascular plants were again the most important organic carbon source. Both tree leaves and benthic algae were also consumed; however, terrestrial detritus was not an important organic
90 carbon source. Benthic algae were important resources in both the Upland Constrained and
Upland High-Energy zones; however, terrestrial detritus was the most important organic carbon resource in the Upland High-Energy zone. Terrestrial detritus was found to be in the diet of greater than 60% of the sampled macroinvertebrates in the Upland High-Energy zone (Table 5.3) supporting the RES. The Lowland Constrained, Upland Constrained, and
Upland High-Energy zones all have relatively high amounts of environmental heterogeneity, and were expected to use a combination of aquatic and terrestrial energy sources.
Organic carbon sources were also analyzed using the RCC as a basis. Primary consumers in sixth order streams used a combination of tree leaves and aquatic vascular plants with an emphasis on aquatic vascular plants (Table 5.4). This result supports the hypothesis that very large rivers rely on both aquatic and terrestrial primary productivity.
In fourth and fifth order streams, in-stream derived organic carbon sources were used almost exclusively by primary consumers. Again, the hypothesis of the RCC was supported.
Tree leaves have a very small contribution to the overall food web showing up only in the diet of the caddisfly, Limnephilidae, from fourth order streams (Table 5.4). Primary consumers in second and third order streams used a combination of terrestrially and aquatically derived sources; however, benthic algae were shown to be the most abundantly used resource in streams of this size (Table 5.4). This, again, supports the prediction of the
RCC that small streams should rely upon both aquatic and terrestrial carbon sources, although the majority of energy comes from in-stream sources rather than terrestrial ones.
91 Table 5.3 Diet contributions for each consumer listed by Functional Process Zone (FPZ). Items in boldface represent potential source contributions.
Benthic Algae Aquatic Vascular Tree Leaves Terrestrial Detritus FPZ Consumer Mean 1% 99% Mean 1% 99% Mean 1% 99% Mean 1% 99% LA Corbicula 0.04 0.00 0.14 0.62 0.51 0.74 0.20 0.01 0.39 0.14 0.00 0.33 LA Heptageniidae 0.22 0.07 0.35 0.58 0.48 0.67 0.09 0.00 0.24 0.11 0.00 0.27 LA Leptoxis 0.15 0.00 0.46 0.41 0.02 0.72 0.23 0.00 0.57 0.22 0.00 0.54 LA Limnephilidae 0.18 0.00 0.45 0.20 0.00 0.46 0.33 0.00 0.69 0.29 0.00 0.62 LA Pleurocera 0.15 0.00 0.47 0.79 0.38 0.98 0.03 0.00 0.14 0.03 0.00 0.15 LC Baetidae 0.27 0.00 0.59 0.26 0.00 0.57 0.23 0.00 0.56 0.24 0.00 0.56 LC Brachycentridae 0.16 0.00 0.50 0.34 0.00 0.80 0.26 0.00 0.65 0.24 0.00 0.60 LC Campeloma 0.28 0.00 0.62 0.29 0.00 0.71 0.21 0.00 0.55 0.22 0.00 0.56 LC Corbicula 0.10 0.00 0.24 0.57 0.45 0.68 0.16 0.00 0.36 0.17 0.00 0.39 LC Heptageniidae 0.21 0.00 0.42 0.48 0.34 0.61 0.15 0.00 0.41 0.16 0.00 0.42 LC Hydropsychidae 0.18 0.00 0.37 0.64 0.40 0.83 0.09 0.00 0.26 0.09 0.00 0.27 LC Isonychiidae 0.31 0.00 0.63 0.27 0.00 0.57 0.20 0.00 0.52 0.21 0.00 0.54 LC Limnephilidae 0.25 0.03 0.45 0.16 0.00 0.35 0.30 0.01 0.59 0.28 0.00 0.58 LC Philopotamidae 0.21 0.00 0.51 0.35 0.01 0.64 0.22 0.00 0.53 0.23 0.00 0.55 LC Physella 0.25 0.00 0.57 0.28 0.00 0.67 0.24 0.00 0.58 0.24 0.00 0.58 UC Birgella 0.19 0.00 0.53 0.32 0.00 0.65 0.25 0.00 0.63 0.24 0.00 0.58 UC Brachycentridae 0.27 0.00 0.54 0.09 0.00 0.25 0.31 0.00 0.61 0.34 0.01 0.66 UC Campeloma 0.26 0.00 0.62 0.31 0.00 0.59 0.22 0.00 0.53 0.22 0.00 0.54 UC Corbicula 0.40 0.21 0.59 0.29 0.07 0.49 0.14 0.00 0.48 0.17 0.00 0.41 UC Elimia 0.22 0.00 0.46 0.47 0.13 0.76 0.16 0.00 0.31 0.16 0.00 0.43 UC Helicopsychidae 0.25 0.00 0.56 0.19 0.00 0.52 0.28 0.00 0.63 0.27 0.00 0.61 UC Heptageniidae 0.54 0.39 0.68 0.13 0.01 0.27 0.14 0.00 0.37 0.19 0.00 0.41 UC Hydropsychidae 0.25 0.03 0.46 0.24 0.07 0.38 0.24 0.00 0.46 0.28 0.01 0.53 UC Isonychiidae 0.25 0.00 0.61 0.24 0.00 0.49 0.25 0.00 0.58 0.26 0.00 0.59 UC Leptoxis 0.25 0.00 0.52 0.27 0.00 0.51 0.23 0.00 0.52 0.24 0.00 0.54 UC Limnephilidae 0.37 0.15 0.58 0.06 0.00 0.23 0.29 0.00 0.59 0.29 0.00 0.63 UC Odontoceridae 0.22 0.00 0.52 0.24 0.00 0.53 0.27 0.00 0.62 0.26 0.00 0.59 UC Physella 0.27 0.00 0.65 0.36 0.00 0.70 0.18 0.00 0.50 0.19 0.00 0.51 UC Planorbella 0.26 0.00 0.64 0.30 0.00 0.66 0.22 0.00 0.56 0.22 0.00 0.56 UC Uenoidae 0.26 0.00 0.62 0.21 0.00 0.56 0.26 0.00 0.64 0.27 0.00 0.65 UH Brachycentridae 0.29 0.00 0.62 0.14 0.00 0.34 0.23 0.00 0.56 0.34 0.01 0.66 UH Glossosomatidae 0.15 0.00 0.49 0.06 0.00 0.35 0.24 0.00 0.62 0.54 0.06 0.95 UH Heptageniidae 0.41 0.15 0.66 0.07 0.00 0.24 0.19 0.00 0.50 0.33 0.01 0.60 UH Hydropsychidae 0.23 0.01 0.44 0.07 0.00 0.18 0.22 0.00 0.51 0.47 0.18 0.75 UH Isonychiidae 0.27 0.00 0.64 0.20 0.00 0.55 0.25 0.00 0.62 0.28 0.00 0.67 UH Limnephilidae 0.22 0.02 0.40 0.05 0.00 0.22 0.40 0.07 0.70 0.33 0.02 0.64 UH Philopotamidae 0.26 0.00 0.61 0.20 0.00 0.51 0.25 0.00 0.60 0.28 0.00 0.63 UH Polycentropodidae 0.26 0.00 0.62 0.21 0.00 0.51 0.25 0.00 0.60 0.28 0.00 0.63
92 Table 5.4 Diet contributions for each consumer listed by stream order. Items in boldface represent potential source contributions.
Benthic Algae Aquatic Vascular Tree Leaves Terrestrial Detritus SSO Consumer Mean 1% 99% Mean 1% 99% Mean 1% 99% Mean 1% 99% 2 Glossosomatidae 0.19 0.00 0.52 0.19 0.00 0.54 0.31 0.00 0.73 0.30 0.00 0.68 2 Heptageniidae 0.21 0.00 0.44 0.36 0.05 0.66 0.22 0.00 0.50 0.21 0.00 0.51 2 Limnephilidae 0.24 0.01 0.46 0.16 0.00 0.42 0.33 0.01 0.67 0.28 0.00 0.62 3 Campeloma 0.24 0.00 0.61 0.27 0.00 0.56 0.24 0.00 0.58 0.25 0.00 0.59 3 Glossosomatidae 0.26 0.00 0.63 0.21 0.00 0.56 0.26 0.00 0.65 0.26 0.00 0.66 3 Heptageniidae 0.42 0.24 0.59 0.05 0.00 0.19 0.26 0.00 0.54 0.27 0.00 0.58 3 Hydropsychidae 0.33 0.08 0.55 0.05 0.00 0.18 0.29 0.00 0.61 0.33 0.00 0.68 3 Limnephilidae 0.25 0.05 0.44 0.06 0.00 0.26 0.40 0.06 0.74 0.29 0.00 0.62 3 Philopotamidae 0.28 0.00 0.60 0.21 0.00 0.52 0.25 0.00 0.61 0.26 0.00 0.62 3 Physella 0.28 0.00 0.64 0.30 0.00 0.67 0.21 0.00 0.54 0.21 0.00 0.56 4 Brachycentridae 0.35 0.01 0.65 0.09 0.00 0.30 0.26 0.00 0.60 0.29 0.00 0.64 4 Corbicula 0.36 0.10 0.59 0.26 0.01 0.49 0.18 0.00 0.45 0.20 0.00 0.47 4 Heptageniidae 0.44 0.15 0.73 0.18 0.00 0.37 0.18 0.00 0.45 0.20 0.00 0.48 4 Hydropsychidae 0.40 0.15 0.64 0.24 0.01 0.44 0.16 0.00 0.43 0.19 0.00 0.46 4 Isonychiidae 0.29 0.00 0.65 0.24 0.00 0.59 0.24 0.00 0.59 0.24 0.00 0.60 4 Leptoxis 0.17 0.00 0.41 0.39 0.13 0.63 0.21 0.00 0.48 0.22 0.00 0.51 4 Limnephilidae 0.29 0.06 0.51 0.07 0.00 0.30 0.34 0.01 0.66 0.30 0.00 0.65 5 Birgella 0.19 0.00 0.52 0.32 0.00 0.65 0.25 0.00 0.61 0.24 0.00 0.58 5 Brachycentridae 0.34 0.01 0.65 0.24 0.00 0.49 0.20 0.00 0.51 0.22 0.00 0.52 5 Campeloma 0.30 0.00 0.62 0.36 0.00 0.75 0.17 0.00 0.48 0.18 0.00 0.50 5 Corbicula 0.16 0.00 0.31 0.54 0.41 0.66 0.14 0.00 0.35 0.16 0.00 0.38 5 Elimia 0.22 0.00 0.47 0.46 0.12 0.76 0.16 0.00 0.42 0.16 0.00 0.43 5 Helicopsychidae 0.25 0.00 0.56 0.20 0.00 0.53 0.28 0.00 0.62 0.27 0.00 0.64 5 Heptageniidae 0.24 0.00 0.48 0.33 0.18 0.48 0.22 0.00 0.50 0.22 0.00 0.51 5 Hydropsychidae 0.25 0.05 0.45 0.36 0.65 0.60 0.20 0.00 0.44 0.18 0.00 0.44 5 Isonychiidae 0.29 0.00 0.64 0.23 0.00 0.44 0.23 0.00 0.54 0.25 0.00 0.56 5 Limnephilidae 0.31 0.06 0.53 0.14 0.00 0.33 0.27 0.00 0.54 0.31 0.00 0.59 5 Odontoceridae 0.22 0.00 0.53 0.24 0.00 0.53 0.27 0.00 0.62 0.26 0.00 0.60 5 Philopotamidae 0.23 0.00 0.57 0.30 0.00 0.63 0.23 0.00 0.58 0.24 0.00 0.58 5 Physella 0.27 0.00 0.65 0.28 0.00 0.68 0.23 0.00 0.58 0.23 0.00 0.59 5 Planorbella 0.27 0.00 0.65 0.28 0.00 0.67 0.22 0.00 0.58 0.23 0.00 0.58 6 Baetidae 0.27 0.00 0.59 0.27 0.00 0.57 0.23 0.00 0.56 0.24 0.00 0.57 6 Corbicula 0.05 0.00 0.15 0.61 0.51 0.72 0.19 0.01 0.37 0.14 0.00 0.33 6 Heptageniidae 0.35 0.25 0.45 0.50 0.40 0.60 0.06 0.02 0.17 0.08 0.00 0.20 6 Pleurocera 0.14 0.00 0.46 0.80 0.38 0.99 0.03 0.01 0.15 0.03 0.00 0.15
93 Discussion
Community metrics The community metrics revealed that while both the River Continuum Concept and the Riverine Ecosystem Synthesis provided reasonable estimates in some cases, neither adequately explained all observed values. Stream order was useful in describing changes in
NR, a surrogate for food chain length. In this case, the longest food chain length was found in 5th order streams where it was significantly longer than the food chain length in 2nd and
3rd order streams (Figure 5.2a). Long food chains are thought to arise as a result of high rates of primary productivity (Pimm 1982, Kaunzinger and Morin 1998), very large ecosystem sizes (Cohen and Newman 1991), or a combination of both (Slobodkin 1961,
Schoener 1989). CR also varied across stream order with the lowest range in 2nd order streams (Figure 5.2a). This result indicated that the basal resources were more limited in
2nd order streams than in streams of other magnitudes. The depressed food chain length in small streams may have resulted from a limitation in carbon sources in quality, productivity, or overall distribution (Power and Dietrich 2002). The food web area, or overall trophic niche space, was highest in 5th order and lowest in 2nd order streams
(Figure 5.2a). This followed the pattern of NR and CR stated above. Finally, a significant difference was found in NND between 2nd order streams and streams of all other orders
(Figure 5.2a). Large NNDs indicate that there is not a high degree of trophic overlap between community members (Layman et al. 2007). In 2nd order streams, it is likely that each niche was filled with relatively fewer species than in larger environments. High environmental variability, however, has not been shown to alter the degree of niche
94 overlap (May and MacArthur 1972). Overall, though, the RCC provided a framework for understanding some of the variation seen in these community metrics.
The RES was also useful in describing changes in some community metrics. NR varied with regard to FPZs; the lowest NR was found in the Upland High-Energy zone
(Figure 5.2b). This zone was characterized by high environmental heterogeneity and limited primary production; these attributes likely act to depress the overall food chain length. Changes in hull area were also seen between FPZs. The Lowland Constrained zone had the highest hull area whereas the Upland High-Energy zone had the lowest (Figure
5.2b). The variation in NR was echoed by the changes in hull area between FPZs. A combination of factors including in-stream productivity and ecosystem size may be responsible for the changes in both NR and hull area (Post 2002).
While both models provided a useful framework for understanding changes in some community metrics, the RCC was a more robust model explaining more variation in these data (Table 5.1). The RCC produced a significant model for each metric, and only NR and area were well-explained by the RES. The mechanisms of the RES for determining differences between FPZs, however, were supported as measures of environmental heterogeneity did not explain more variation within the data for food web metrics than FPZ alone (Table 5.2), and models using these measures were not significantly different from models using only FPZ as a factor.
Carbon sources Both the RCC and RES provide a set of hypotheses to help understand how carbon sources may vary within river systems. We used only those sources that have contributions greater than zero at the first percentile as potential diet items (Phillips and Gregg 2003).
95 The RCC states that headwater streams (1st-3rd order) should rely upon both in-stream and terrestrially derived primary productivity (Vannote et al. 1980). Here, we find that diets in
2nd order streams do rely heavily upon both aquatic and terrestrial sources; however, diets of primary consumers in 3rd order streams are based almost exclusively on aquatic carbon sources (Table 5.4). This change could be the result of increased rates of aquatic primary productivity in 3rd order streams relative to predictions of the RCC. The RCC also predicts that mid-order streams (4th-5th order) should be solely dependent upon autochthonous
(i.e., in-stream) production (Vannote et al. 1980). In this system, we find that while primary consumers in 5th order streams use only aquatic carbon sources, in 4th order streams some terrestrial components were also used (Table 5.4). Finally, in large rivers (> 6th order),
Vannote et al. (1980) predict that a combination of carbon sources is important. Again, we find agreement between our data and the predictions of the RCC (Table 5.4). Overall, the
RCC provides a robust framework for understanding the importance of both aquatic and terrestrial primary productivity in the diets of primary consumers.
Predictions of the RES for the importance of various carbon sources follow those predictions for relative amounts of in-stream primary productivity. The RES states that environments with low environmental heterogeneity and relatively high light levels (i.e., the Lowland Alluvial zone) should have high rates of primary productivity. Other zones in the Kanawha River Basin (i.e., Lowland Constrained, Upland Constrained, Upland High-
Energy) should have lower amounts of primary productivity and, thus, should be more dependent upon terrestrial carbon sources (Thorp et al. 2008). Our results showed that the diets of primary consumers within the Lowland Alluvial zone were dominated by aquatic sources, primarily aquatic vascular plants (Table 5.3). Diets in all other zones contained a
96 mixture of both aquatic and terrestrial sources, and primary consumers in the Upland
High-Energy zone used predominantly terrestrial sources.
While the RCC provides some hypotheses that are met in this system, mechanistically, they may still be inaccurate. For example, the RCC predicts that mid-order streams (i.e., 4th-5th order) should rely solely on autochthonous carbon; however, we found this expectation was not met. Although most members of the food web within 4th and 5th order streams did consume mostly aquatic carbon sources, these same streams were shown to have a deficit of in-stream primary production relative to respiration (Table 4.2).
Similarly, the RCC predicts that large rivers (> 6th order) should rely heavily upon terrestrial carbon subsidies. This is supported by SIA, but again, conflicting results arose with regard to in-stream primary productivity (Table 4.2). Similarly, the RES provides hypotheses that are not well-met in this system (i.e., very low use of terrestrial organic carbon in Lowland Constrained zone), although these hypotheses are supported by previous work (Table 4.2). The overall system metabolism in each FPZ was congruent with predictions from the RES; however, in some FPZs where respiration was greater than in- stream productivity (i.e., Lowland Constrained, Upland Constrained), autochthonous materials make up the majority of the diets of consumers.
Conclusions In the Kanawha River Basin, both the RCC and RES provide insight into community metrics and organic carbon sources. The RCC uses the linear paradigm of a continuum to describe changes in carbon sources. Following these changes, community metrics are also variable between stream orders. The RES, on the other hand, uses a suite of environmental variables to describe multiple zones within a basin. Although some factors (i.e., NR, hull
97 area) change with regard to these zones, the predictions of the RCC were met more often than the predictions of the RES. Several reasons for these results may exist. First, the
Kanawha River Basin contains relatively low amounts of hydrogeomorphic variability.
While several FPZs exist, similarities in some environmental variables between FPZs also exist. Second, the scale (e.g., temporal, spatial) of this project may have been more conducive to detecting differences between stream orders rather than FPZs. For example, as a result of the relatively fast-flowing nature of streams within the Kanawha River Basin, the rate of changes in ecological processes may also be variable. For example, the distance over which changes occur could be very large as a result of high water velocities throughout the basin relative to low-gradient systems. In terms of stream orders, this alteration in process rates may not manifest in the same way as in the context of FPZs.
Finally, the relative arrangement of FPZs is variable. Third order streams always proceed from second order streams; however, streams within the Lowland Constrained zone do not always have the same type of zones upstream (i.e., Lowland Constrained may have any other zone type directly upstream). Although this is a notable feature in patch dynamics, it does not always confer congruous results. To conclude, both models offer valid hypotheses for understanding some of the components of food web structure and organic carbon resources; however, neither model provides a robust and complete framework for this understanding.
98 Chapter 6 – Conclusions
No man ever steps in the same river twice, for it's not the same river and he's not the same man.
– Heraclitus
99 The Kanawha River Basin Streams contained within the Kanawha River basin have made an excellent study system for this research. Both the River Continuum Concept and the Riverine Ecosystem
Synthesis provide hypotheses for free-flowing, natural (i.e., with minimal human disturbance) watersheds. Although modifications that incorporate changes with regard to these conditions exist for both these concepts, this project has sought to test the RCC and
RES based on their essential proposals.
The Kanawha River Basin (31,691 km2) is located in the Mid-Atlantic region of the
USA. It has a relatively pristine catchment containing mostly forested land cover.
Agriculture, industry, and resource extraction make up a small proportion (< 20%) of the overall land cover in the drainage basin. The basin falls within three physiographic provinces (i.e., Appalachian Plateaus, Valley and Ridge, Blue Ridge) and is underlain by a mixture of sandstone, shale, limestone, dolostone, chert, and alluvium. The Kanawha River
Basin exists between 164 m at the river mouth and its highest elevation of 1,355 m. Long- term mean annual precipitation (1961-2000) for the basin ranges from 910-1,500 mm resulting in mean daily discharge at Charleston, WV of 432 m3 s-1 with a range of 31 - 6,116 m3 s-1 (Messinger and Hughes 2000). The Kanawha River is a sixth order stream. Six distinct Functional Process Zones exist within the Kanawha River Basin including the
Lowland Alluvial, Lowland Constrained, Upland Constrained, Open-Valley Upland, Upland
High-Energy, and Reservoir zones (Williams et al. 2013).
Sites sampled as part of this project were shown to contain many distinct features.
The riverbed substratum of sites within each FPZ was unique. Underlying geology and valley floor width most influenced the composition of riverbed substratum within each
100 FPZ. Variation in flow regimes (measured as daily average streamflow) was also present within this system. Measures of spatial and temporal heterogeneity are known to affect biological communities and biogeochemical processes, and sites within this basin are variable in terms of physical habitat.
Overall, the Kanawha River Basin is relatively oligotrophic, and rates of primary productivity may be suppressed as a result of limited nutrient concentrations, limited light availability, or low water temperatures. Measured rates of many inorganic nutrients required for building primary producer biomass have shown that 85-90% of sites within
3- this system are oligotrophic with regard to phosphorus (as estimated by PO4 ) and nitrogen (as estimated by TDN), respectively. This suppressed trophic state may have resulted in observed low rates of primary productivity; rates of primary productivity were low, relative to respiration, at nearly all sites. Limitations in available organic carbon as a result of low aquatic primary production may have affected the function and structure of biological communities. Many sites were predicted to be dependent upon a mixture of in- stream and terrestrially organic carbon, and this prediction was confirmed by measurements of ecosystem metabolism. This prediction was further supported by estimates of the major food items in the diets of consumers. In nearly all SSOs and FPZs, terrestrial and aquatic organic carbon sources were used by consumers. Results from this study show that, while the Kanawha River Basin was an excellent system for testing hypotheses proposed by the RCC and RES, many sites had similar characteristics in terms of the functions and processes contained within them.
101 On models Conceptual models are prevalent across many fields of biological and ecological sciences. Models serve a specific purpose, that is, to provide a framework for understanding the functions and processes that occur within ecosystems, between populations, or within organisms themselves. Freshwater aquatic ecology offers no exception. In fact, many limnological models exist, and each provides a structured system for understanding a particular aspect (or multiple aspects) of those ecosystems. The process of building a conceptual model is crucial for developing an understanding for complex systems because model builders are forced to recognize and apply the principles which govern or regulate systems. Without this perspective, the advancement of ideas and concepts may be limited. Models are also useful because, as they are developed to explain processes and functions, they can also be assessed for their utility by those same processes and functions.
Many models are heuristic by nature, and testing conceptual models is just as important as creating models. Frequently, conceptual models propose to describe a suite of processes within aquatic ecosystems (e.g., nutrient cycling, system metabolism, reliance on autochthonous or allochthonous production). Though many models can correctly describe some functions and processes, properly defining the way processes and functions change within an entire system is often impossible. This process of model testing can lead to improvement of existing models or the development of new ones.
Additions or improvements can make models more globally applicable; however, additional variables can also complicate models. Several models in aquatic ecology build off of, provide amendments to, or test the applicability of other models (e.g., Bruns et al. 1984,
102 Sedell et al. 1989). While these additions can be beneficial, they can also provide unrealistic expectations. Collecting all the necessary information to successfully assess the predictions of multifaceted models may be impossible. Further, instead of simplifying ecological processes and functions into easily understandable sections, the models themselves may become overly complex.
Here, I have compared two conceptual models, the River Continuum Concept and the Riverine Ecosystem Synthesis, based on predictions of ecosystem metabolism, nutrient concentrations, food web dynamics, and the utilization of organic carbon sources by organisms. Each model uses a different set of variables to make predictions about processes and functions within streams and rivers. In some cases, the RCC was a better predictor; however, the RES also served as the better model at times. A review of the central tenets, as well as the applicability, of each model shows how each performed in the
Kanawha River Basin.
River Continuum Concept To review, the River Continuum Concept (Vannote et al. 1980) describes changes in biological communities, ecosystem processes, physical habitat, and water chemistry throughout a watershed. Changes are linear and exist along a gradient from headwaters to great rivers (i.e., from low- to high-order streams). In small streams (1st – 3rd order), great importance is placed on terrestrial subsidies (in the form of coarse particulate organic matter) as in-stream primary productivity may be limited by light or nutrients. Biological communities should be composed of individuals capable of assimilating a combination of organic carbon from both autochthonous and allochthonous sources, and physical habitats contain high amounts of spatial heterogeneity. The RCC proposes that in mid-order streams
103 (4th – 5th order), in-stream primary productivity meets the demands of overall community respiration due to high levels of light availability and few (if any) nutrient limitations.
Again, this relates to changes in biological community composition. A reduction in the complexity of physical habitats should also occur in these streams. Finally, in large rivers (>
6th order), primary productivity may be limited as a result of light attenuation within the water column, fine particulate organic matter transported from upstream areas is common, biological communities reflect these change in available food resources, and physical habitats are further reduced in environmental heterogeneity.
My research shows that while some of the proposed changes along a continuum exist, in many cases predictions of the RCC are not met. Variation in nutrient
3- + concentrations were not consistent as some nutrients (i.e., PO4 , NH4 ) increased with increasing stream order while others (i.e., TDN) had different trends or showed seasonal differences (i.e., DOC) (Figure 4.2). This may be due to differences in nutrient cycling rates or other biological processes (e.g., uptake of organic carbon). The predicted ratio of primary productivity and ecosystem respiration was also not met by empirical data in some cases. While headwater streams were correctly predicted to have less production than respiration, predictions for mid-order streams and large rivers were not met (Table
4.2). Low inorganic nutrient concentration within the basis is a potential reason for relatively low primary productivity. Overall, 85-90% of the sampled sites within the
Kanawha River Basin fell below the oligotrophic-mesotrophic boundary with regard to phosphorus and nitrogen, respectively.
Changes in the structure of biological communities and their dependence on terrestrial and aquatic organic carbon sources also occurred within the basin. Stable
104 isotope analysis of primary producers and consumers showed that some of the predictions of the RCC were met. Headwater streams had the smallest δ13C range, and this confirms that basal resources are limited in small streams (Figure 5.2a). Further, community members in headwater streams used a combination of in-stream and terrestrially derived organic carbon sources. Nearest neighbor distances, that is the amount of niche overlap between community members, showed that there was relatively little overlap only in the smallest streams (i.e., 2nd order). Many other community metrics did not vary between mid-order and large streams and rivers. Although the RCC provides a framework for changes in community metrics and changes in basal resources, the mechanisms underlying these changes are often not accurate. For example, based on the prediction that mid-order streams have high primary productivity, biological communities should be solely dependent on in-stream derived carbon. Results from the stable isotope analysis of consumers supported this hypothesis (Table 5.2). However, analysis of system metabolism shows that rates of primary productivity in mid-order streams are greatly exceeded by community respiration (Table 4.2), and thus conflicting results arise. Although several concepts from the RCC are supported by this research, more questions about the underlying mechanisms have been raised.
Riverine Ecosystem Synthesis The other conceptual model analyzed in this study was the Riverine Ecosystem
Synthesis (Thorp et al. 2006, 2008). The RES predicts changes within a watershed based on hydrogeomorphic factors, not longitudinal position. Several variables including elevation and topography, river channel sinuosity, precipitation, and slope are used to designate
Functional Process Zones. FPZs represent one level within a hierarchical patch model for
105 understanding riverine processes and functions. Between FPZs, predictions of the RES are based upon varying levels of environmental heterogeneity (e.g., physical habitat, flow velocities) and other unique factors that may influence the behavior of stream ecosystems.
Zones with narrow valleys and relatively higher amounts environmental heterogeneity should have lower amounts of primary productivity. This may be due to limited light availability and low concentrations of inorganic nutrients within these zones. Changes in the structure of biological communities should also exist between zones based on rates of productivity and ecosystem size. Overall, though, differences in processes and functions between FPZs exist as a result of changes in environmental attributes that define those zones.
Difference in the physical structure of each FPZ is a critical component of some predictions of the RES. Riverbed substratum is a commonly used measure of physical complexity and can affect biological communities (i.e., spawning grounds, refugia) and biogeochemical processes (i.e., nutrient cycling) that occur at the water-substratum interface. We developed a technique for rapidly characterizing the substrate of streams and rivers (Collins and Flotemersch 2013) and used that technique to assess the FPZs within the Kanawha River Basin. Differences in average sediment composition of the riverbed substratum between the four sampled FPZs exist (x2=13.3335, d.f.=3, P=0.003968; Figure
3.2). Further, silt and clay were the particles primarily responsible for similarities within the Lowland Alluvial and Lowland Constrained zones, while cobble, boulder, and bedrock were responsible for similarities within the Upland Constrained and Upland High-Energy zones. Principal axis correlation showed that two catchment variables, geology (a surrogate for sediment supply) and valley floor width (a surrogate for energy associated with high
106 flow), were responsible for the location of FPZ groupings within ordination space.
Altogether, the riverbed substratum of FPZs within the Kanawha River Basin is different.
My research has shown that some nutrient concentrations and system metabolism change with regard to FPZs within the Kanawha River Basin. Predictions of the RES for the ratio of primary productivity and respiration were met in all zones (Table 4.2). In zones with relatively higher amounts of environmental heterogeneity (i.e., Lowland Constrained,
Upland Constrained, Upland High-Energy), rates of primary productivity were lower than rates of community respiration. In the Lowland Alluvial zone, hypothesized to have higher amounts of light and higher amounts of inorganic nutrients, P/R was not significantly different than one (one-sample t-test: t = -1.06, P = 0.39, 2 degrees of freedom). Changes in nutrient concentrations within the Kanawha basin, though, were not all well-explained by
FPZs. Significant differences in dissolved organic carbon and total dissolved nitrogen exist; however, these changes do not always align with predictions from the RES.
Stable isotope analysis also showed how the RES can be used to explain changes in the food web structure and nutrient sources between FPZs. There are significant differences in food chain length (as measured by δ15N range) between FPZs and total hull area. However, significant differences between FPZs did not exist for other metrics (Figure
5.2b). However, the RES does provide predictions for the use of aquatic and terrestrial carbon sources that are met. In those zones where primary productivity is lower than respiration, terrestrial subsidies are expected. In each zone where P/R is less than one (i.e.,
Lowland Constrained, Upland Constrained, Upland High-Energy) organic carbon from terrestrial sources showed up in the diet of consumers (Table 5.1). While there were not
107 significant differences between each community metric, predictions of the RES for the use of aquatic and terrestrial organic carbon sources were accurate in each FPZ.
Comparing models The RCC and RES both provide a framework for understanding changes that occur between biological communities, physical habitats, and ecological processes and functions
(Table 6.1). Although each model uses a different suite of characteristics for understanding these changes, some predictions from either model were met in the Kanawha River Basin.
In some cases, the River Continuum Concept was shown to be a better model (e.g., lower
AIC scores for two nutrients). However, the Riverine Ecosystem Synthesis was a better model for some other processes (e.g., system metabolism). The mechanisms that underly the predictions of these models, though, must also be examined to determine if either model provides a more complete idea of how and why functions and processes change within the basin.
Stream order is the main concept behind all the hypotheses of the River Continuum
Concept. Several hydrogeomorphologic explanatory variables are used as the basis for the predictions of the Riverine Ecosystem Synthesis. In this study, we have shown that mechanistically, the River Continuum Concept could be flawed; however the “sliding scale” approach to the RCC broadens its applicability. Although most of the RCC’s predictions for the use of in-stream and terrestrially derived organic carbon by consumers were met, the ratio of primary productivity to community respiration in those stream sites did not support the original predictions. On the other hand, at times, the RES did not provide a robust framework for describing changes between FPZs (e.g., community metrics). Again, the mechanisms behind the conceptual hypotheses may be flawed.
108 Table 6.1 River Continuum Concept and Riverine Ecosystem Synthesis interpretations and results from each study question. Overall, a combination of models was often the best predictor for ecosystem processes and functions within the Kanawha River Basin.
Question RCC Interpretation RES Interpretation Outcome Sediment Reduction in size of Variety of sediment size Riverbed substrate Size sediments with related to Functional variability aligns with increasing stream order Process Zones prediction from RES System P/R << 1 in headwaters; P/R < 1 in highly Predictions from RES Metabolism P/R ≈ 1 in mid-order complex, heterogeneous were supported; streams; habitats; P/R ≈ 1 in however predictions P/R < 1 in large rivers; habitats with low from RCC were not; Rationale: light complexity; Rationale: environmental limitation due to canopy high flow variability may heterogeneity was cover (headwaters) or scour primary likely more important turbidity (large rivers) producers; complex than longitudinal limits production habitats have less position available light Nutrients Low concentrations of Low concentrations of Neither interpretation inorganic nutrients in inorganic nutrients in validated; RCC and highly productive highly productive RES do not provide streams as nutrients are streams as nutrients are robust framework for consumed during consumed during understanding photosynthesis; high photosynthesis; high changes in nutrient concentrations of concentrations of concentrations; dissolved organic carbon dissolved organic carbon combination of in highly productive in highly productive models better than streams streams either individually; fast-flowing system with likely low residence time Trophic Highly productive Highly heterogeneous Longitudinal Structure streams (i.e., mid-order) streams should have classification (RCC) should have highest highest trophic performed better than trophic complexity complexity patch model (RES); however combination was best Organic Highly productive Highly heterogeneous Neither interpretation Carbon streams (i.e., mid-order) streams should depend validated; carbon should depend less on more on terrestrial source also seemingly terrestrial carbon carbon sources unrelated to P/R; sources combination of models better
109 It is relevant to note that “all models are wrong but some are useful” (Box and
Draper 1987). Perhaps, a better model may be created from a combination of factors from the RCC and RES. In some cases, a statistical model describing changes in nutrient concentrations containing both these concepts was shown to explain more of the variation within the dataset. Caution must be exercised, however, in creating overly complex models.
As factors are added, an increase in the explanatory power should be assumed. Simply adding factors (or complexity) to a model to attempt to describe more of the observed variation can be detrimental. Understanding the first principles behind changes in functions and processes is of the utmost importance. It is through this process of model conceptualization and testing that true advancements in knowledge can take place.
In this study, a statistical model using both the RCC and RES was built to explain changes in nutrient concentrations, system metabolism, and community metrics separately. Using the RCC and RES in conjunction provided a better model than either the
RCC or RES alone in some cases for macronutrients (Table 4.2). However, a combined model was not significantly better than either the RCC or RES alone at explaining changes in system metabolism. Finally, the combined model using both RCC and RES did not out- perform either model independently when explaining changes in community metrics
(Table 5.1). These results indicate that, while a combination of both models may provide a better framework for understanding changes in some variables within the Kanawha River
Basin, either individual model works well in some cases, too. Although the mechanisms underlying these models differ, it appears that the statistical combination of these mechanisms does not provide for a more robust understanding of changes in ecological processes and functions within this system.
110 From this work, it is evident that neither the RCC nor the RES provide a complete basis for understanding the complicated processes that occur within a river network. As has been shown, a more useful model may be one that combines parts of both these conceptual models. I propose that some of the most important features from the RCC and
RES may be used together to create a new framework for understanding how processes and functions change within a stream system. The RCC highlights the importance of a longitudinal connection from headwaters to river mouth. While the RCC uses Strahler stream order as a measure for this connectivity, a more robust method for characterizing streams may be more applicable. For example, catchment size (an general estimate of stream size) or linear distance from the mouth (a measure of spatial connectivity) could be more appropriate than Strahler stream order for understanding changes in some processes within a stream network. While the RES uses a suite of hydrogeomorphic variables to characterize streams segments, selecting a group of these variables (e.g., channel width, down-valley slope, floodplain width, underlying geology, precipitation) that are related to ecological and physiochemical processes of interest may aid in creating a more useful model. Combining these aspects from both models could improve overall predictive power for riverine ecosystems. As an example where this combined model may out-perform either model individually, biogeochemical processing and nutrient cycling rates are influenced by several factors including underlying geology (a surrogate for substrate; see Chapter 3), down-valley slope (or potential flow rate), and distance travelled along a watercourse (downstream flux), among other things. Thus, a model which incorporates a more complete combination of meaningful deterministic variables will be more useful than either the RCC or RES alone.
111 Bibliography Anderson, D.M., P.M. Glibert, and J.M. Burkholder. 2002. Harmful algal blooms and
eutrophication: nutrient sources, composition, and consequences. Estuaries 25:
704-726.
Angradi, T.R. 1994. Trophic linkages in the lower Colorado River: multiple stable isotope
evidence. Journal of the North American Benthological Society 13: 479-495.
Angradi, T.R. (ed.). 2006. Environmental Monitoring and Assessment Program: Great River
Ecosystems, Field Operations Manual. U.S. Environmental Protection Agency:
Washington, DC; EPA-620/R-06/002.
Babler, A.L., A. Pilati, and M.J. Vanni. 2011. Terrestrial support of detritivorous fish
populations decreases with watershed size. Ecosphere 2: 1-23.
Barbour, M.T., J. Gerritsen, B.D. Snyder, and J.B. Stribling. 1999. Rapid Bioassessment
Protocols for use in Streams and Wadeable Rivers: Periphyton, Benthic
Macroinvertebrates and Fish. 2nd edition. U.S. Environmental Protection Agency,
Office of Water: Washington, DC; EPA 841/B-99/002.
Bayley, P.B. 1995. Understanding large river-floodplain ecosystems. BioScience 45: 153-
158.
Beaulieu, J.J., C.P. Arango, D.A. Balz, and W.D. Shuster. 2013. Continuous monitoring reveals
multiple controls on ecosystem metabolism in a suburban stream. Freshwater
Biology 58: 918-937.
Belbin, L. 1993. Environmental representativeness: regional partitioning and reserve
selection. Biological Conservation 66: 223-230.
112 Benda, L., L. R. Poff, D. Miller, T. Duune, G. Reeves, M. Pollock, and G. Pess. 2004. Network
dynamics hypothesis: spatial and temporal organization of physical heterogeneity in
rivers. BioScience 54: 413-427.
Bevenger, G.S. and R.M. King. 1995. A pebble count procedure for assessing watershed
cumulative effects. Res. Pap. RM-RP-319. Fort Collins, CO: U.S. Deptartment of
Agriculture, Forest Service.
Bjorkland, R., C.M. Pringle, and B. Newton. 1999. A stream visual assessment protocol
(SVAP) for riparian landowners. Environmental Monitoring and Assessment 68: 99-
125.
Blocksom K.A., E. Emery, and J. Thomas. 2006. Comparison of random and systematic site
selection for assessing attainment of aquatic life uses in segments of the Ohio River.
U.S. Environmental Protection Agency, National Exposure Research Laboratory:
Cincinnati, OH; EPA 600/R-06/089.
Blocksom, K., E. Emery, and J. Thomas. 2009. Sampling effort needed to estimate condition
and species richness in the Ohio River, USA. Environmental Monitoring and
Assessment 155: 157-167.
Boecklen, W.J., C.T. Yarnes, B.A. Cook, and A.C. James. 2011. On the use of stable isotopes in
trophic ecology. Annual Review of Ecology, Evolution, and Systematics 42: 411-440.
Bott, T.L., J.T. Brock, C.E. Cushing, S.V. Gregory, and R.C. Petersen. 1978. A comparison of
methods for measuring primary productivity and community respiration.
Hydrobiologia 60: 3-12.
113 Bradley, W.C., R.K. Fahnestock, and E.T. Rowekamp. 1972. Coarse sediment transport by
flood flows on the Kink River, Alaska. Geological Society of America Bulletin 83:
1261-1284.
Bray, J.R. and J.T. Curtis. 1957. An ordination of upland forest communities of southern
Wisconsin. Ecological Monographs 27: 325-349.
Brookes, A. 1988. Channelized rivers: Perspectives for environmental management. John
Wiley and Sons Ltd: Chichester, UK.
Bruns, D.A., G.W. Minshall, J.T. Brock, C.E. Cushing, K.W. Cummins, and R.L. Vannote. 1982.
Ordination of functional groups and organic matter parameters from the Middle
Fork of the Salmon River, Idaho. Freshwater Biology 1: 2-12.
Bruns, D.A., G.W. Minshall, C.E. Cushing, K.W. Cummins, J.T. Brock, and R.L. Vannote. 1984.
Tributaries as modifiers to the river continuum concept: analysis by polar
ordination and regressionmodels. Archiv fur Hydrobiologie 99: 208-220.
Bunten K. and S.R. Abt. 2001. Sampling surface and subsurface particle-size distributions in
wadeable gravel-and cobble-bed streams for analyses in sediment transport,
hydraulics, and streambed monitoring. U.S. Forest Service; General Technical Report
RMRS-GTR-74.
Cardinale, B.J., M.A. Palmer, C.M. Swan, S. Brooks, and N.L. Poff. 2002. The influence of
substrate heterogeneity on biofilm metabolism in a stream ecosystem. Ecology 83:
412-422.
Carling, P.A. 1983. Threshold of coarse sediment transport in broad and narrow natural
streams. Earth Surface Processes and Landforms 8: 1-18.
114 Church, M.D. and R. Kellerhals. 1978. On the statistics of grain size variation along a gravel
river. - Canadian Journal of Earth Science 15: 1151-1160.
Church, M.A., D.G. McLean, and J.F. Wolcott. 1987. River bed gravels: sampling and analysis.
In: Thorne, C.R., J.C. Bathurst, and R.D. Hey (eds.), Sediment Transport in Gravel-Bed
Rivers. John Wiley and Sons, New York, NY., pp. 43-88.
Clarke, K.R. 1993. Non-parametric multivariate analyses of changes in community
structure. Australian Journal of Ecology 18: 117-143.
Cohen, J.E. and C.M. Newman. 1991. Community area and food-chain length: theoretical
predictions. American Naturalist 138: 1542-1554.
Cole, J.J., S.R. Carpenter, J. Kitchell, M.L. Pace, C.T. Solomon, and B. Weidel. 2011. Strong
evidence for terrestrial support of zooplankton in small lakes based on stable
isotopes of carbon, nitrogen, and hydrogen. PNAS 108: 1975-1980.
Collins, S.E. and J.E. Flotemersch. 2013. Evaluating the sounding rod method for sampling
coarse river-bed materials in non-wadeable streams and rivers. River Research and
Applications. DOI: 10.1002/rra.2697.
Collins, S.E., M.C. Thoms, and J.E. Flotemesrch. (Accepted). Hydrogeomorphic zones
characterize riverbed sediment patterns within a river network. River Systems.
Costanza, R., R. d’Arge, R. de-Groot, S. Farber, M. Grasso, B. Hannon, K. Limburg, S. Naeem, R.
O’Neill, J. Paruelo, R. Raskin, P. Sutton, and J. van den Belt. 1998. The value of
ecosystem services: putting the issues in perspective. Ecological Economics 25: 67-
72.
115 Cushing, C.E., K.W. Cummins, G.W. Minshall, and R.L. Vannote. 1983. Periphyton,
chlorophyll a, and diatoms of the Middle Fork of the Salmon River, Idaho. Holarctic
Ecology 6: 221-227.
Dansgaard, W. 1964. Stable isotopes in precipitation. Tellus 16: 436-468.
Davis, W.S. and T.P. Simon (eds). 1995. Biological Assessment and Criteria: Tools for Water
Resource Planning and Decision Making. Lewis Publishers: Boca Raton, FL. de Ruiter, P.C., V. Wolters, and J.C. Moore. 2005. Dynamic Food Webs. In: de Ruiter, P.C., V.
Wolters, and J. C. Moore (eds.), Dynamic Foodwebs: Multispecies Assemblages,
Ecosystem Development and Environmental Change. Academic Press. San Diego, CA:
pp. 3-9.
Doane, T.A. and W.R. Horwath. 2003. Spectrophotometric determination of nitrate with a
single reagent. Analytical Letters 36: 2713-2722.
Dodds, W.K., V.H. Smith, and B. Zander. 1997. Developing nutrient targets to control benthic
chlorophyll levels in streams: a case study of the Clark Fork River. Water Research.
31: 1738-1750.
Dodds, W.K., J.R. Jones, and E.B. Welch. 1998. Suggested classification of stream trophic
state: distributions of temperate stream types by chlorophyll, total nitrogen, and
phosphorus. Water Research. 32: 1455-1462.
Dodds, W.K., C.T. Robinson, E.E. Gaiser, G.J. A. Hansen, H. Powell, J.M. Smith, N.B. Morse, S.L.
Johnson, S.V. Gregory, T. Bell, T.M. Kratz, and W.H. McDowell. 2012. Surprises and
insights from long-term aquatic data sets and experiments. BioScience 62: 709-721.
116 Dollar, E.S.J., C.S. James, K.H. Rogers, and M.C. Thoms. 2007. A framework for
interdisciplinary understanding of rivers as ecosystems. Geomorphology 89: 147-
162.
Doucett, R.R., G. Power, D.R. Barton, R.J. Drimmie, and R.A. Cunjak. 1996. Stable isotope
analysis of nutrient pathways leading to Atlantic salmon. Canadian Journal of
Fisheries and Aquatic Science 53: 2058-2066.
Doucett, R.R., J.C. Marks, D.W. Blinn, M. Caron, and B.A. Hungate. 2007. Measuring
terrestrial subsidies to aquatic food webs using stable isotopes of hydrogen. Ecology
88: 1587-1592.
Duarte, C.M. 1995. Submerged aquatic vegetation in relation to different nutrient regimes.
Ophelia 41: 87-112.
Duarte, C.M., N. Marbà, E. Gacia, J.W. Fourqurean, J. Beggins, C. Barrόn, and E.T. Apostolaki.
2010. Seagrass community metabolism: assessing the carbon sink capacity of
seagrass meadows. Global Biogeochemical Cycles 24: GB4032.
doi:10.1029/2010GB003793.
Dynesius, M. and C. Nilsson. 1994. Fragmentation and flow regulation of river systems in
the northern third of the world. Science 266: 753-762.
Eaton A.D., Clesceri L.S., and Greenberg A.E., (eds.) 2005. Standard methods for the
examination of water and wastewater. American Public Health Association,
Washington, DC.
Ehleringer, J.R. and T.E. Dawson. 1992. Water uptake by plants: perspectives from stable
isotope composition. Plant, Cell and Environment 15: 1073-1082.
117 Elser, J.J., W.F. Fagan, A.J. Kerkoff, N.G. Swenson, and B.J. Enquist. 2010. Biological
stoichiometry of plant production: metabolism, scaling and ecological response to
global change. New Phytologist 186: 593-608.
Elwood, J.W., J.D. Newbold, R.V. O’Neill, and W. VanWinkle. 1983. Resource spiraling: an
operational paradigm for analyzing lotic ecosystems. In: Fontaine, T.D. and S.M.
Bartell (eds.), The Dynamics of Lotic Ecosystems. Ann Arbor Science Publishers, Inc,
Ann Arbor, MI.
European Commission. 2007. Interpretation Manual of European Union Habitats:EUR 27.
European Commission DG Environment: Habitats Council Directive
CE/1992/43/EEC.
Fernandez, W., R.M. Vogel, and A. Sankaramsubramanian. 2000. Regional calibration of a
watershed model. Hydrological Sciences 45: 689-707.
Fisher, S.G. 1997. Pattern, process, and scale in freshwater systems: some unifying
thoughts. In: Giller, P.S., A.G. Hildrew, and D.G. Raffaelli (eds.), Aquatic Ecology:
Scale, Pattern, and Process. Blackwell Scientific, Oxford, UK.
Fitzpatrick, F.A., I.R. Waite, P.J. D’Arconte, M.R. Meador, M.A. Maupin, and M.E. Gurtz. 1998.
Revised methods for characterizing stream habitat in the National Water-Quality
Assessment Program. U.S. Geological Survey; Water-Resources Investigations
Report 98-4052.
Flotemersch, J.E., J.B. Stribling, and M.J. Paul. 2006. Concepts and Approaches for the
Bioassessment of Non-wadeable Streams and Rivers. U.S. Environmental Protection
Agency: Cincinnati, OH; EPA 600/R-06/127.
118 Flotemersch, J.E., J.H. Thorp, and B.S. Williams. 2010. Application of the riverine ecosystem
synthesis (RES) and the functional process zone (FPZ) approach to EPA
environmental mission tasks for rivers. U.S. Environmental Protection Agency,
Washington, DC, EPA/600/R-11/089.
Folk, R.L. 1966. A review of grain-size parameters. Sedimentology 6: 73-93.
Forrest, J. and N.R. Clark. 1989. Characterizing grain size distributions: evaluation of a new
approach using a multivariate extension of entropy analysis. Sedimentology 36:
711-722.
Fry, B. 1991. Stable isotope diagrams of freshwater food webs. Ecology 72: 2293-2297.
Gaeuman, D. and R.B. Jacobson. 2005. Aquatic habitat mapping with an acoustic Doppler
current profiler: considerations for data quality. U.S. Geological Survey; Open File
Report 2005-1163, 24 pp.
Galat, D.L. and A.G. Frazier. 1996. Overview of river-floodplain ecology in the upper
Mississippi River basin. In: Kelmelis, J.A. (ed.), Science for Floodplain Management
into the 21st Century, Volume 3. Scientific Assessment and Strategy Team, U.S.
Government Printing Office: Washington, DC.
Gat, J.R. 1971. Comments on the stable isotope method in regional groundwater
investigations. Water Resource Research 7: 980-993.
Gillooly, J.F., J.H. Brown, G.B. West, V.M. Savage, and E.L. Charnov. 2001. Effects of size and
temperature on metabolic rate. Science 293: 2248-2251.
Gilvear D., P. Hunter, and T. Higgins. 2007. An experimental approach to the measurement
of the effects of water depth and substrate on optical and near infra-red reflectance:
119 a field-based assessment of the feasibility of mapping submerged instream habitat.
International Journal of Remote Sensing 28: 2241-2256.
Gordon, N.D., T.A. McMahon, B.L. Finlayson, C.J. Gippel, and R.J. Nathan. 2004. Stream
Hydrology: An Introduction for Ecologists. 2nd ed. John Wiley and Sons. West
Sussex, England.
Gurnell, A., J. Goodson, K. Thompson, N. Clifford, and P. Armitage. 2007. The river-bed: a
dynamic store for plant propagules? Earth Surface Processes and Landforms 32:
1257-1272.
Heritage, G.L. and D.J. Milan. 2009. Terrestrial Laser Scanning of grain roughness in a
gravel-bed river. Geomorphology 113: 4-11.
Herwig, B.R., D.H. Wahl, J.M. Dettmers, and D.A. Soluk. 2007. Spatial and temporal patterns
in the food web structure of a large floodplain river assessed using stable isotopes.
Canadian Journal of Fisheries and Aquatic Science 64: 495-508.
Horowitz, A.J. 1985. A primer on trace-metal sediment chemistry. U.S. Geological Survey
Water-Supply Paper. 2277.
Hynes, H.B.N. 1963. Imported organic matter and secondary productivity in streams.
Proceedings of the XVI International Congress of Zoology 4: 324-329.
Izagirre, O., M. Bermejo, J. Pozo, and A. Elosegi. 2007. RIVERMET ©: an Excel-based tool to
calculate river metabolism from oxygen concentration curves. Environmental
Modeling & Software 22: 24-32.
Jackson, P.C., J. Cavelier, G. Goldstein, F.C. Meinzer, and N.M. Holbrook. 1995. Partitioning of
water resources among plants of a lowland tropical forest. Oecologia 101: 197–203.
120 Jackson, A.L., R. Inger, A.C. Parnell, and S. Bearhop. 2011. Comparing isotopic niche widths
among and within communities: SIBER – stable isotope Bayesian ellipses in R.
Journal of Animal Ecology 80: 595-602.
Julian, J.P, M.W. Doyle, and E.H. Stanley. 2008. Empirical modeling of light availability in
rivers. Journal of Geophysical Research 3: G03022, doi:10.1029/2007JG000601.
Junger, M. and D. Planas. 1994. Quantitative use of stable carbon isotope analysis to
determine the trophic base of invertebrate communities in a boreal forest lotic
system. Canadian Journal of Fisheries and Aquatic Science 51: 52-61.
Junk, W.J., P.B. Bayley, and R.E. Sparks. 1989. The flood pulse concept in river-floodplain
systems. In: Dodge, D. P. (ed.), Proceedings of the International Large River
Symposium. Canadian Special Publications in Fisheries and Aquatic Sciences 106:
110-127.
Karr, J.R. and D.R. Dudley. 1981. Ecological perspective on water quality goals.
Environmental Management 5: 55-68.
Karr, J.R., K.D. Fausch, P.L. Angermeier, P.R. Yant, and I.J. Schlosser. 1986. Assessing
biological integrity in running waters: a method and its rationale. Illinois Natural
History Survey Special Publication No. 5: Champaign, IL.
Kaeser, A.J., T.L. Litts, and W. Tracy. 2012. Using low-cost side-scan sonar for benthic
mapping throughout the lower Flint River, Georgia, USA. River Research and
Applications. DOI: 10.1002/rra.2556.
Kaiser, M.J., S.I. Rogers, and J.R. Ellis. 1999. Importance of benthic habitat complexity for
demersal fish assemblages in fish habitat: essential fish habitat and rehabilitation.
121 In: Benaka, L.R. (ed.), American Fisheries Society, Symposium 22. pp. 212–223.
American Fisheries Society.
Kaufmann, P.R. and E.G. Robison. 1997. Physical habitat assessment. In: Lazorchak, J.L., D.J.
Klemm, and D.V. Peck (eds.), Environmental Monitoring and Assessment Program –
Surface Waters: Field Operations and Methods for Measuring the Ecological
Condition of Wadeable Streams, U.S. Environmental Protection Agency: Washington,
DC; EPA 620/R-94/004F.
Kaufmann, P.R. and T.R. Whittier. 1997. Habitat Assessment. Pages 5-1 to 5-26 In: Baker,
J.R., D.V. Peck, and D.W. Sutton (eds.), Environmental Monitoring and Assessment
Program -Surface Waters: Field Operations Manual for Lakes, U.S. Environmental
Protection Agency: Washington, DC; EPA-620/R-97/001.
Kaunzinger, C.M.K. and P.J. Morin. 1998. Productivity controls food-chain properties in
microbial communities. Nature 395: 495-497.
King, J.M. and C.A. Brown. 2006. Environmental flows: striking the balance between
development and resource protection. Ecology and Society 11: 26.
Kite, G.W. 1991. A watershed model using satellite data applied to a mountain basin in
Canada. Journal of Hydrology 128: 157-169.
Klingeman, P.C. and W.W. Emmett 1982. Gravel bedload transport processes. In: Hey, R.D.,
J.C. Bathurst, and C.R. Thorne (eds.), Gravel-Bed Rivers. John Wiley and Sons, New
York, NY., pp. 141-169.
Knighton, A.D. 1980. Longitudinal changes in size and sorting of stream-bed material in
four English rivers. Geological Society of America Bulletin 91: 55-62.
122 Kondolf, G.M. 2000. Assessing salmonid spawning gravel quality. Transactions of the
American Fisheries Society 129: 262-281.
Layman, C.A., D.A. Arrington, C.G. Montaña, and D.M. Post. 2007. Can stable isotope ratios
provide for community-wide measures of trophic structure? Ecology 88: 42-48.
Lazorchak, J.M., B.H. Hill, D.K. Averill, D.V. Peck, and D.J. Klemm (eds). 2000. Environmental
Monitoring and Assessment Program -Surface Waters: Field Operations and
Methods for Measuring the Ecological Condition of Non-Wadeable Rivers and
Streams. U.S. Environmental Protection Agency: Cincinnati OH; EPA-620/R-06/00.
Leavesley, G.H. and L.G. Stannard. 1995. The Precipitation-Runoff Modeling System – PRMS.
In: Singh, V.P. (ed.), Computer Models of Watershed Hydrology. Water Resources
Publications.
Leopold, L.B. and M.O. Marchand. 1968. On the quantitative inventory of the riverscape.
Water Resources Research 4: 709-717.
Leopold, L.B. and M.G. Wolman. 1957. River channel patterns: braided, meandering, and
straight. Washington, DC: US Government Printing Office, pp. 37-86.
Lindenmayer, D.B. and G.E. Likens. 2010. Effective Ecological Monitoring. CSIRO Publishing.
Lisle, T.E., J.M Nelson, J. Pitlick, M.A. Madeji, and B.L. Barkett. 2000. Variability of bed
mobility in natural, gravel-bed channels and adjustments to sediment load at local
and reach scales. Water Resources Research 36: 3743-3755.
Lohman, K., J.R. Jones, and B.D. Perkins. 1992. Effects of nutrient enrichment and flood
frequency on periphyton biomass in northern Ozark streams. Canadian Journal of
Fisheries and Aquatic Sciences. 49: 1198-1205.
123 (MA) Millennium Ecosystem Assessment. 2005. Ecosystems and Human Well-Being:
Synthesis. Island Press, Washington. 155pp.
May, R.M. and R.H. MacArthur. 1972. Niche overlap as a function of environmental
variability. PNAS 69: 1109-1113.
McConkey, B., P. Ayers, S. Bakaletz, A. Fiscor, J.R. Candlish, K. Swinson, and R. Emmott.
2009. Georeferenced riverine habitat mapping (surface & bathymetric) Big South
Fork National River and Recreation Area. American Society of Agricultural and
Biological Engineers, 2009 Reno, Nevada, June 21 - June 24, 2009.
McCoy, E.D. and S.S. Bell. 1991. Habitat structure: the evolution and diversification of a
complex topic. In: Bell, S.S., E.D. McCoy, and H.R. Mushinsky (eds.), Habitat Structure:
The Physical Arrangement of Objects in Space. pp. 3–27. Chapman and Hall, London.
Messinger, T. and C.A. Hughes. 2000: Environmental setting and its relations to water
quality in the Kanwaha River Basin. Report 00-4020, 66 pp. Water Resources
Investigations. U.S. Geological Survey.
Milan, D.J., D. Hetherington, and G.L. Heritage. 2007. Application of a 3D laser scanner in the
assessment of erosion and deposition volumes and channel change in a proglacial
river. Earth Surface Processes and Landforms 32: 1657-1674.
Miller, A.J. 1995. Valley morphology and boundary conditions influencing spatial patterns
of flood flow. In: Costa, J.E., A.J. Miller, K.W. Potter, and P.R. Wilcock (eds.), Natural and
Anthropogenic Influences in Fluvial Geomorphology: The Wolman Volume, American
Geophysical Union, Washington, pp. 57-82.
124 Minagawa, M. and E. Wada. 1984. Stepwise enrichment of 15N along food chains: further
evidence and the relation between δ15N and animal age. Geochimica et
Cosmochimica Acta 48: 1135–1140.
Minshall, G.W. 1978. Autotrophy in stream ecosystems. BioScience 28: 767-771.
Minshall, G.W. 1984. Aquatic insect-substratum relationships. In: Resh, V.H. & D.M.
Rosenberg (eds.), The Ecology of Aquatic Insects. Praeger Publishers, New York, NY.
Minshall, G.W. and J.N. Minshall. 1977. Microdistribution of benthic invertebrates in a
Rocky Mountain (U.S.A.) stream. Hydrobiologia 55: 231-249.
Minshall, G.W., R.C. Petersen, K.W. Cummins, T.L. Bott, J.R. Sedell, C.E. Cushing, and R.L.
Vannote. 1983. Interbiome comparison of stream ecosystem dynamics. Ecological
Monographs 53: 1-25.
Minshall, G.W., K.W. Cummins, R.C. Petersen, C.E. Cushing, D.A. Bruns, J.R. Sedell, and R.L.
Vannote. 1985. Developments in stream ecosystem theory. Canadian Journal of
Fisheries and Aquatic Sciences 42: 1045-1055.
Montgomery, D.R. 1999. Process domains and the river continuum. Journal of the American
Water Resources Association 35: 397-410.
Moore, I.D., R.B. Grayson and A.R. Ladson. 1991. Digital terrain modelling: a review of
hydrological, geomorphological, and biological applications. Hydrological Processes
5: 3-30.
Mosley, M.P. and D.S. Tindale. 1985. Sediment variability and bed material sampling in
gravel-bed rivers. Earth Surface Processes and Landforms 10: 465-482.
Mulholland, P.J., C.S. Fellows, J.L. Tank, N.B. Grimm, J.R. Webster, S.K. Hamilton, E. Martí, L.
Ashkenas, W.B. Bowden, W.K. Dodds, W.H. Mcdowell, M.J. Paul, and B.J. Peterson.
125 2001. Inter-biome comparison of factors controlling stream metabolism. Freshwater
Biology 46: 1503–1517.
Newbold, J.D., J.W. Elwood, R.V. O’Neill and W.V. Winkle. 1981. Measuring nutrient spiraling
in streams. Canadian Journal of Fisheries and Aquatic Sciences 38: 860-863.
Newbold, J.D., P.J. Mulholland, J.W. Elwood, and R.V. O’Neill. 1982a. Organic carbon spiraling
in stream ecosystems. Oikos 38: 266-272.
Newbold, J.D., R.V. O’Neill, J.W. Elwood, and W.V. Winkle. 1982b. Nutrient spiraling in
streams: implications for nutrient limitation and invertebrate activity. American
Naturalist 120: 628-652.
Newson, M.D. and C.L. Newson. 2000. Geomorphology, ecology and river channel habitat:
mesoscale approaches to basin-scale challenges. Progress in Physical Geography 24:
195-217.
Norris, R.H. and M.C. Thoms. 1999. What is river health? Freshwater Biology 41: 197-209.
O’Connor, N.A. 1991. The effects of habitat complexity on the macroinvertebrates
colonizing wood substrates in a lowland stream. Oecologia 85: 504-512.
Owens, P.N., R.J. Batalla, A.J. Collins, B. Gomez, D.M. Hicks, A.J. Horowitz, G.M. Kondolf, M.
Marden, M.J. Page, D.H. Peacock, E.L. Petticrew, W. Salomons, and N.A. Trustrum.
2005. Fine-grained sediment in river systems: environmental significance and
management issues. River Research and Applications 21: 693-717.
Paerl, H.W. and J. Huisman. 2009. Climate change: a catalyst for global expansion of harmful
cyanobacterial blooms. Environmental Microbiology Reports 1: 27-37.
Parnell, A.C., R. Inger, S. Bearhop, and A.L. Jackson. 2010. Source partitioning using stable
isotopes: coping with too much variation. PLoS ONE 5: e9672.
126 Plafkin, J.L., M.T. Barbour, K.D. Porter, S.K. Gross, and R.M. Hughes. 1989. Rapid
bioassessment protocols for use in streams and rivers: benthic macroinvertebrates
and fish. U.S. Environmental Protection Agency, Office of Water Regulations and
Standards: Washington, DC; EPA 440/4-89/001.
Petts, G.E. 1989. Historical analysis of fluvial hydrosystems. In: Petts, G.E., H. Moller and A.L.
Roux (eds.), Historical Change of Large Alluvial Rivers: Western Europe. John Wiley
and Sons, New York, NY,. pp.1-18.
Phillips, D.L. and J.W. Gregg. 2003. Source partitioning using stable isotopes: coping with
too many sources. Oecologia 136: 261-269.
Pickett, S.T.A., W.R. Burch, and J.M. Grove. 1999. Interdisciplinary research: maintaining the
constructive impulse in a culture of criticism. Ecosystems 2: 302-307.
Pickup, G. 1983. Sedimentation processes in the Purari River upstream of the delta. In: Petr,
T. (ed.), The Purari: Tropical Environment of a High Rainfall River Basin. Dr. W. Junk
Publishers, The Hague. pp. 205-225.
Pickup, G. and R.F. Warner. 1984. Geomorphology of tropical rivers: II. Channel adjustment
to sediment load and discharge in the Fly and Lower Purari, Papua-New-Guinea.
Catena Supplement 5: 19-41.
Pimm, S.L. 1982. Food Webs. Chapman and Hall, London.
Pingram, M.A., K.J. Collier, D.P. Hamilton, B.J. Hicks, and B.O. David. 2012. Spatial and
temporal patterns in carbon flow in a temperate, large river food web.
Hydrobiologia DOI 10.1007/s10750-012-1408-2.
127 Poff, N.L., J.D. Olden, D. Merritt, and D. Pepin. 2007. Homogenization of regional river
dynamics by dams and global biodiversity implications. Proceedings of the National
Academy of Sciences 104: 5732-5737. DOI: 10.1073/pnas.0609812104.
Post, D.M. 2002. The long and short of food-chain length. TRENDS in Ecology and Evolution
17: 269–277.
Post, D.M., M.L. Pace, and N.G. Hairston. 2000. Ecosystem size determines food-chain length
in lakes. Nature 405: 1047-1049.
Power, M.E. 1992. Habitat heterogeneity and the functional significance of fish in river food
webs. Ecology 73: 1675-1688.
Power, M.E. and W.E. Dietrich. 2002. Food webs in river networks. Ecological Research
17:451-471.
R Core Team. 2013. R: A language and environment for statistical computing. R Foundation
for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL
http://www.Rproject.org/.
Rau, G.H. 1980. Carbon-13/carbon-12 variation in subalpine lake aquatic insects: food
source implications. Canadian Journal of Fisheries and Aquatic Science 37: 742-746.
Raven, P.J., N.T.H. Holmes, F.H. Dawson, P.J.A. Fox, M. Everard, I.R. Fozzard, and K.J. Romen.
1998. River habitat quality: the physical character of rivers and streams in the UK
and Isle of Man. ISBN 1-873760-429. Environment Agency; Bristol, UK.
Ringuet, S., L. Sassano, and Z.I. Johnson. 2010. A suite of microplate reader-based
colorimetric methods to quantify ammonium, nitrate, orthophosphate and silicate
concentrations for aquatic nutrient monitoring. Journal of Environmental
Monitoring 13: 370-376.
128 Rosgen, D. and H.L. Silvey. 1996. Applied River Morphology. Wildland Hydrology: Pagosa
Springs, CO.
Rounick, J.S., M.J. Winterbourn, and G.L. Lyon. 1982. Differential utilization of allochthonous
and autochthonous inputs by aquatic invertebrates in some New Zealand streams: a
carbon isotope study. Oikos 39: 191-198.
Saleska, S., H. Da Rocha, B. Kruijt, and A. Nobre. 2013 Ecosystem carbon fluxes and
Amazonian forest metabolism. In: Keller, M., M. Bustamante, J. Gash, and P. Silva Dias
(eds.), Amazonia and Global Change. American Geophysical Union, Washington, D. C.
doi: 10.1029/2008GM000739.
Schoener, T.W. 1989. Food webs from the small to the large. Ecology 70: 1559-1589.
Scott, K.M. 1967. Downstream changes in sedimentological parameters illustrated by
particle distribution from a breached rockfill dam. U.S. Geological Survey,
Sacramento, California.
Schindler, D.W. 1998. Whole-ecosystem experiments: replication versus realism: the need
for ecosystem-scale experiments. Ecosystems 1: 323-334.
Schindler, D.W., K.H. Mills, D.F. Malley, D.L. Findlay, J.A. Shearer, I.J. Davies, M.A. Turner, G.A.
Linsey, and D.R. Cruikshank. 1985. Long-term ecosystem stress: the effects of years
of experimental acidification on a small lake. Science 228: 1395-1401.
Sedell, J.R. and J.L. Froggart. 1984. Importance of streamside forests to large rivers: the
isolation of the Williamette River, Oregon, U.S.A., from its floodplain by snagging and
streamside forest removal. Verhandlungen des Internationalen Verein Limnologie
22: 1828-1834.
129 Sedell, J.R., J.E. Richey and F.J. Swanson. 1989. The River Continuum Concept: a basis for the
expected ecosystem behavior of very large rivers? In: Dodge, D. P. (ed.), Proceedings
of the International Large River Symposium. Canadian Special Publications in
Fisheries and Aquatic Sciences 106: 49-55.
Shaw, J. and R. Kellerhals. 1982. The composition of recent alluvial gravels in the Alberta
river beds. Alberta Research Council Bulletin 41: 151pp.
Simonsen, J.F. and P. Harremoës. 1978. Oxygen and pH fluctuations in rivers. Water
Research 12: 477-489.
Sinsabaugh, R.L. 1997. Large-scale trends for stream benthic respiration. Journal of the
North American Bethological Society. 16: 119-122.
Slobodkin, L. B. 1961. Growth and Regulation of Animal Populations. Holt, Rinehart and
Wilson, New York.
Smith, R.A., R.B. Alexander, and G.E. Schwartz. 2003. Natural background concentrations of
nutrients in streams and rivers of the conterminous United States. Environmental
Science & Technology 37: 3039-3047.
Sorensen, T. 1948. A method of establishing groups of equal amplitude in plant sociology
based on similarity of species and its application to the analyses of vegetation on
Danish commons. Kongelige Danske Videnskabernes Selskab 5: 1-34.
Sternberg, H. 1875. Untersuchen über längen und querprofil gescheibeführende Flüsse. Z
Bauwesen 25: 483-506.
Strahler, A.N. 1964. Quantitative geomorphology of drainage basins. In: Chow, V. T. (ed.)
Handbook of Applied Hydrology. McGraw-Hill Book Co.
130 Statzner B. and B. Higler. 1986. Stream hydraulics as a major determinant of benthic
invertebrate zonation patterns. Freshwater Biology 16: 127–139.
Takimoto, G. and D.M. Post. 2013. Environmental determinants of food-chain length: a
meta-analysis. Ecological Research 28: 675-681.
Thoms, M.C. 1992. A comparison of grab- and freeze-sampling techniques in the collection
of gravel-bed sediment. Sedimentary Geology 78: 191-200.
Thoms, M.C., S.M. Hill, M.J. Spry, X.J. Chen, T.J. Mount, and F. Sheldon. 2004. The
geomorphology of the Darling River. In: Breckwodt, R., R. Boden, and J. Andrews
(eds.), The Murray Darling Basin Commission, pp. 68-105.
Thoms, M.C., M.E. Parsons, and J.M. Foster. 2007. The use of multivariate statistics to
elucidate patterns of floodplain sedimentation at different scales. Earth Surface
Processes and Landforms 32: 672-686.
Thorp, J.H., M.C. Thoms, and M.D. Delong. 2006. The riverine ecosystem synthesis:
biocomplexity in river networks across space and time. River Research and
Applications 22: 123-147.
Thorp, J.H, M.C. Thoms, and M.D. Delong. 2008. The Riverine Ecosystem Synthesis: Toward
a Conceptual Cohesiveness in River Science. Academic Press: London, UK.
Thorp, J.H., J.E. Flotemersch, M.D. Delong, A.F. Casper, M.C. Thoms, F. Ballantyne, B.S.
Williams, B.J. O’Neill, and C.S. Haase. 2010. Linking ecosystem services,
rehabilitation, and river hydrogeomorphology. BioScience 60: 67-74.
Townsend, C.R. 1989. The patch dynamics concept of stream community ecology. Journal of
the North American Benthological Society 8: 36-50.
131 Townsend, S.A. and A.V. Padovan. 2005. The seasonal accrual and loss of benthic algae
(Spirogyra) in the Daly River, an oligotrophic river in tropical Australia. Marine and
Freshwater Research 56: 317-327.
USEPA. 1990. Biological Criteria: National Program Guidance for Surface Waters. U.S.
Environmental Protection Agency, Office of Water Regulations and Standards:
Washington, DC; EPA 440/5-90/004.
USEPA. 1992. Procedures for Initiating Narrative Biological Criteria. U.S. Environmental
Protection Agency, Office of Science and Technology: Washington, DC; EPA 822/B-
92/002.
USEPA. 1998. Lake and Reservoir Bioassessment and Biocriteria: Technical Guidance
Document. EPA 841-B-98-007. Office of Water, U.S. Environmental Protection
Agency, Washington, DC, 191 pp.
USEPA. 2002. Biological Assessments and Criteria: Crucial Components of Water Quality
Programs. EPA 822-F-02-006. Office of Water. U.S. U.S. Environmental Protection
Agency, Washington, DC, 6 pp.
USEPA. 2005. Use of Biological Information to Tier Designated Aquatic Life Uses in State
and Tribal Water Quality Standards. U.S. Environmental Protection Agency, Office of
Water: Washington, DC; EPA-822/R-05/001.
USEPA. 2007. National Rivers and Streams Assessment: Field Operations Manual. EPA-841-
B-07-009 U.S. Environmental Protection Agency: Washington, DC.
USGPO. 1989. FederalWater Pollution Control Act (33 U.S.C. 1251 et seq.) as amended by a
P.L. 92-500. In: Compilation of Selected Water Resources and Water Pollution
132 Control Laws, Printed for use of the Committee on Public Works and Transportation
U.S. Government Printing Office: Washington, DC; 561 pp.
Van Donk, E., R.D. Gulati and M.P. Grimm. 1989. Food web manipulation in Lake Zwemlust:
positive and negative effects during the first two years. Hydrobiological Bulletin 23:
19-34.
Vannote, R.L., G.W. Minshall, K.W. Cummins, J.R. Sedell, and C.O. Cushing. 1980. The river
continuum concept. Canadian Journal of Fisheries and Aquatic Sciences 37: 130-137.
Vinson, M.R. and C.P. Hawkins. 1998. Biodiversity of stream insects: variation at local,
basin, and regional scales. Annual Review of Entomology 43: 271-293.
Walling, D.E. 1995. Suspended sediment yields in a changing environment. In: Gurnell, A.
and G. Petts (eds.), Changing River Channels. Wiley, Chichester, pp. 149-176.
Walling D.E. and D. Fang. 2003. Recent trends in the suspended sediment loads of the
world’s rivers. Global and Planetary Change 39: 111–126.
Ward, J.V. and J.A. Stanford. 1983. The serial discontinuity concept of lotic ecosystems. In:
Fontaine, III, T.D. and S.M. Bartell (eds.), Dynamics of Lotic Ecosystems. Ann Arbor
Science, Ann Arbor, MI.
Ward, J.V. and J.A. Standford. 1995. Ecological connectivity in alluvial river ecosystems and
its disruption by flow regulation. River Research and Applications 11: 105-119.
Warwick, R.M. and K.R. Clarke. 1993. Comparing the severity of disturbance: a meta-
analysis of marine macrobenthic community data. Marine Ecology Progress Series
92: 221-231.
Weatherburn, M.W. 1967. Phenol-hypochlorite reaction for determination of ammonia.
Analytical Chemistry 39: 971-974.
133 Webster, J. 1975. Analysis of potassium and calcium dynamics in stream ecosystems on
three Southern Appalachian watersheds of contrasting vegetation. Ph.D.
dissertation, Univ. of Ga. at Athens.
Welch, E.B., J.M. Quinn, and C.W. Hickey. 1992. Periphyton biomass related to point-source
nutrient enrichment in seven New Zealand streams. Water Research. 26: 669-675.
Welcomme, R.L. 1985. River fisheries. Food and Agriculture Organization Fisheries
Technical Paper 262. Food and Agriculture Organization of the United Nations,
Rome, Italy.
Wentworth, C.K. 1922. A scale of grade and class terms for clastic sediments. The Journal of
Geology. 30: 377-392.
Wetzel, R.G. 1993. Microcommunities and microgradients: linking nutrient regeneration,
microbial mutualism, and high sustained aquatic primary production. Aquatic
Ecology 27: 3-9.
Wilhelm, J.G.O. 2002. A habitat rating system for non-wadeable rivers of Michigan. Master’s
Thesis, School of Natural Resources and Environment, University of Michigan, Ann
Arbor, MI.
Wilhelm, J.G.O., J.D. Allan, K.J. Wessell, R.W. Merritt, and K.W. Cummins. 2005. Habitat
assessment of non-wadeable rivers in Michigan. Environmental Management 36:
592-609.
Williams, B.S., E. D'Amico, J.H. Kastens, J.H. Thorp, J.E. Flotemersch, and M.C. Thoms. 2013.
Automated riverine landscape characterization: GIS-based tools for watershed-scale
research, assessment, and management. Environmental Monitoring and Assessment
185: 7485-7499.
134 Williams, D.D. 1980. Some relationships between stream benthos and substrate
heterogeneity. Limnology and Oceanography 25: 166-172.
Williamson, C.E., W.K. Dodds, T.K. Kratz, and M.A. Palmer. 2008. Lakes and streams as
sentinels of environmental change in terrestrial and atmospheric processes.
Frontiers in Ecology and the Environment 6: 247-254.
Wilson, M.A. and S.R. Carpenter. 1999. Economic valuation of freshwater ecosystem
services in the United States: 1971-1997. Ecological Applications 9: 772-783.
Winterbourne, M.J., J.S. Rounick, and B. Cowie. 1989. Are New Zealand stream ecosystems
really different? New Zealand Journal of Marine and Freshwater Research 15: 321-
328.
Wolman, M.G. 1954. A method of sampling coarse river-bed material. Transactions of the
American Geophysical Union 35: 951-956.
Woodward, G. and A.G. Hildrew. 2002. Food web structure in riverine landscapes.
Freshwater Biology 47: 777-798.
Wu, J. and O.L. Loucks. 1995. From balance of nature to hierarchical patch dynamics: a
paradigm shift in ecology. Quarterly Review of Biology 70: 439-466.
Yatsu, E. 1955. On the longitudinal profile of the graded river. Transactions American
Geophysical Union 36: 655-663.
Young, R.G., C.D. Matthaei, and C.R. Townsend. 2008. Organic matter breakdown and
ecosystem metabolism: functional indicators for assessing river ecosystem health.
Journal of the North American Benthological Society 27: 605-625.
135 Zweig, L.D, and C.F. Rabeni. 2001. Biomonitoring for deposited sediments using benthic
invertebrates: a test on 4 Missouri streams. Journal of the North American
Benthological Society 20: 643-657.
136 Appendix I – Habitat heterogeneity between Functional Process Zones
Introduction Habitat heterogeneity is a commonly used measure of overall ecosystem function and health, particularly in aquatic habitats. The US Environmental Protection Agency uses a variety of measures when employing biological monitoring programs including morphological variability (i.e., riffle, run, pool, glide), physical habitat (i.e., riverbed substrate composition), and flow velocity (Barbour et al. 1999).
Changes in physical habitat composition have been shown to affect biological communities in many ways. High habitat heterogeneity is thought to increase overall biodiversity (McCoy and Bell 1991, Kaiser et al. 1999). Changes in predator-prey relationships can also have cascading effects throughout the food web. For example, selective predation on insectivorous macroinvertebrates by fishes in rivers with boulder- bedrock substrates releases algivorous macroinvertebrates from the pressure of predation, and this reduces algal standing crop (Power 1992). This phenomenon was not seen in rivers with fine-gravel substrates, demonstrating that an increase in spatial heterogeneity can drastically affect food web dynamics.
Similarly, changes in river flow regime can alter a variety of riverine processes.
Primary productivity can be altered by changes in flow (Julian et al. 2008). High flow velocities are typically associated with low primary productivity because of limited nutrient availability. Disruption of the flow regime can have deleterious effects on ecological connectivity between rivers and their floodplains, can reduce flood peak and increase flood duration, and can alter successional patterns within aquatic systems (Ward and Stanford 1995).
137 Here, I determine if Functional Process Zones (FPZs) within the Kanawha River
Basin differ with regard to physical habitat (as determined by riverbed substratum composition) and flow. Understanding if and how FPZs differ in environmental heterogeneity may help to explain other processes examined within the basin.
Methods Riverbed substrate composition data were collected using the sounding rod method
(Collins and Flotemersch 2013) at 35 sites within four FPZs of the Kanawha River Basin. In brief, a capped, hollow copper pole was used to estimate the size of 100 particles for each site. The method roughly follows the Wolman (1954) method with the addition of a haphazard zigzag pattern (instead of linear transects) as suggested by Bevenger and King
(1995). Mean particle size (D50) for each site was determined using the limit of each estimated size class, and this value was used to compare FPZs. Additionally, coefficient of variation (CV) was calculated for each site and compared across FPZs. Data for neither D50 nor CV were not normally distributed, so a non-parametric Kruskal-Wallis one-way ANOVA was used to test for significance in both cases.
Daily average streamflow (cubic feet per second; cfs) data were collected from all available U.S. Geological Survey gaging stations within the Kanawha River Basin for the time period 01 January 2010 through 31 December 2010 (or the most recent year available for gaging stations no longer in operation). Data were categorized by FPZ depending on where gaging stations were located. CV was determined for each station. ANOVA was used to determine if significant differences existed in CV between FPZs, and Tukey’s HSD was used to determine similar groups when significant differences were detected.
138 Results and Discussion Each FPZ of the Kanawha River Basin has a unique textural character in terms of the relative composition of the six textural groups measured (Figure AI.1). Significant differences exist between FPZs in terms of the average sediment composition (x2=13.3335, d.f.=3, P=0.003968) as well as CV (x2=11.333, d.f.=3, P=0.01155). The bedrock textural group was not present in every zone. The Lowland Alluvial and Lowland Constrained FPZs did not contain bedrock. Further, fine sediments (silt/clay) were relatively under- represented in the Upland Constrained and Upland High-Energy zones; however, the silt/clay textural group was dominant in the Lowland Alluvial FPZ contributing 39.2% to the substratum textural composition. Cobbles were the dominant textural group in all FPZs except the Lowland Alluvial FPZ. This textural group represented 29.2%, 37.8% and 31.5% of the substratum composition, on average, in the Lowland Constrained, and Upland High- energy FPZs, respectively.
Variation in flow also exists within the Kanawha River Basin. Significant differences in flow exist between FPZs (Figure AI.2; F3,17=4.999, P=0.0115). Lowland Alluvial and
Lowland Constrained zones have the lowest variation in flow. The Upland High-Energy zone is not significantly different than these two zones; however, the CV is higher in the
Upland High-Energy zone. CV in flow was highest in the Upland Constrained zone.
139
Figure AI.1 Pebble size for 100 particles was measured at each site. Distributions for each FPZ were made based on percent composition for all sites within that FPZ. D50 measurement was determined using the highest value for each category (e.g., fine sediment = 0.01 mm, sand = 2 mm, etc.) and calculating the mean particle size for each site. Coefficient of variation was also determined for each site using the highest value for each category. In both cases, data failed the assumption of normality, so a Kruskal-Wallis test was used. Significant differences exist between FPZs in both D50 and CV (x2=13.3335, d.f.=3, P=0.003968; x2=11.333, d.f.=3, P=0.01155, respectively).
140
2.5 B AB 2
A A 1.5
1
0.5 Coefficient of Variation of Coefficient
0 Lowland alluvial Lowland Upland Upland high- constricted constricted energy FPZ
Figure AI.2 Coefficient of variation was determined for daily average flow data for a calendar year at USGS gaging stations within the Kanawha River Basin. Significant differences in this measure exist between Functional Process Zones (F3,17=4.999, P=0.0115) and Tukey’s HSD was used to determine differences between groups.
141 Appendix II – Temporal variation in stable isotopes of carbon and nitrogen in fish tissue
Introduction Stable isotope analysis (SIA) is a common technique for studying many environmental and ecological topics including rain and groundwater investigations
(Dansgaard 1964, Gat 1971), water uptake in plants (Ehleringer and Dawson 1992), resource partitioning (Jackson et al. 1995), and trophic ecology (Minagawa and Wada 1984,
Fry 1991, Post 2002). SIA has beneficially impacted aquatic food web and trophic ecology studies as it offers many advantages over traditional methods (Boecklen et al. 2010).
Typically these studies have relied upon impractical, time-consuming, and tedious methods such as gut content analysis or observational experiments that may be inconclusive in determining the importance of various food sources in diets. SIA offers a solution to these problems as it provides a relatively easy and reliable method for determining trophic position and food web structure.
It is common for ecological studies to occur over multiple field seasons and years
(e.g., Schindler et al. 1985, Van Donk et al. 1989,). A potential difficulty then becomes comparing data that were collected from distinct temporal periods as isotopic signatures may be variable. Atmospheric conditions and precipitation can affect stable isotopes due to evaporative fractionation (Stewart 1975) which may influence the isotopic content of primary producers. This, in turn, could alter the stable isotopes found in consumers.
142 Here, I determine whether stable isotope data (δ13C and δ15N) vary between samples collected in 2010 and 2011. I expect that samples collected in either year will have similar isotopic signatures due to relatively low environmental variability between years.
Methods One hundred twenty-two individuals comprising 15 species of fish from multiple feeding levels (i.e., herbivores, insectivores, piscivores) were collected from four sites within the Kanawha River Basin in 2010 and 2011. All specimens were identified to species. Dorsal muscle tissue samples were collected on site from individuals > 250 mm while smaller individuals were kept for laboratory processing. All fish and tissue samples were frozen for transportation back to the lab. Dorsal muscle tissue from each specimen was dried to constant mass at 60 C for 24-48 h and homogenized using a mortar and pestle.
Samples of 0.100-0.300 mg for each fish were packed in tin capsules for stable isotope analysis. Stable isotopes of carbon and nitrogen were analyzed on a Finnigan TC/EA isotopic ratio mass spectrometer at the Andrew W. Breidenbach Environmental Research
Center, U.S. EPA. Isotopic standards including ammonium sulfate, glutamic acid, acetanilide, and spiders were run concurrently to allow for data correction.
Data were corrected where necessary such that standards analyzed along with samples from 2011 were not significantly different than those run with samples from 2010.
Mean isotopic ratios of δ13C and δ15N from each species at each site for 2010 and 2011 were calculated, and these values were compared using a paired samples t-test in the statistical program R (version 2.15.1; R Core Team 2012).
143 Results and Discussion There were no significant differences from fish tissue in either δ13C (t = 1.56, 21 d.f.,
P = 0.13) or δ15N (t = -1.06, 21 d.f., P = 0.30) when compared between 2010 and 2011. This shows that tissue collected from a site over multiple years can be used in food web and trophic ecology studies without reservation. Although ecological studies typically encompass several years, stable isotopes do not appear to vary on the short time scale examined. However, a common standard must be used in all SIA to ensure that data corrections can be made if necessary.
SIA can be used for a variety of purposes including climate, atmospheric, groundwater, precipitation, and dietary studies. The utility of SIA in trophic and dietary studies is evident, and it can be relatively easy to collect and analyze and reliable over multiple-year studies.
144 Appendix III – Data
145 Table AIII.1 Description of thirty-five sampling locations within the Kanawha River Basin with location, stream order, Functional Process Zones, and catchment variables. LA is Lowland Alluvial, LC is Lowland Constrained, UC is Upland Constrained, and UH is Upland High- Energy. Precip is annual average precipitation in mm. Elevation in m.
Valley ValleyFloor DownValley Side Sine Plan SiteID Latitude Longitude FPZ SSO Precip Geology Elevation Width Width VW:VFW Slope Slope RC Form 1.1 38.48 -81.83 LA 6 1086.1 alluvium 171.89 2152.1 1481.51 1.452633 0.000025 0.05 1.224 1 1.2 38.22 -81.43 LA 6 1155.1 alluvium 180.50 2609.3 603.2925 4.325022 -0.000061 0.36 1.386 2 1.3 38.61 -81.99 LA 6 1068.1 alluvium 164.33 1985.6 1446.611 1.372561 0.000085 0.29 1.061 1 1.4 38.84 -82.14 LA 6 1053.8 alluvium 164.40 3133.8 1967.523 1.592777 -0.000024 0.09 1.048 1 1.5 38.30 -81.56 LA 6 1151.5 alluvium 173.57 1319.3 850.8049 1.550637 -0.000079 0.53 1.252 1 1.6 38.33 -81.58 LA 6 1151.5 alluvium 173.57 1319.3 850.8049 1.550637 -0.000079 0.53 1.252 1 1.7 38.39 -81.83 LA 6 1112.2 alluvium 172.52 2527.3 1602.006 1.577576 -0.000062 0.16 1.201 1 1.8 38.00 -80.75 LA 4 1258.1 alluvium 730.78 3386.2 2518.326 1.344614 -0.00011 0.21 1.479 1 1.9 37.74 -80.63 LA 4 1019.2 alluvium 490.30 904.1 374.8192 2.412109 -0.00237 0.05 1.890 1 2.1 38.45 -81.78 LC 5 1103.0 shale 171.89 1510.6 924.0486 1.634729 0 0.30 2.190 1 2.2 38.61 -80.86 LC 5 1228.9 sandstone 238.99 827.4 155.1982 5.331008 -0.00028 0.28 1.379 2 2.3 38.07 -81.08 LC 6 1142.4 sandstone 288.61 1919.2 166.1791 11.54897 -0.00351 0.00 1.288 1 2.4 38.49 -81.33 LC 5 1200.4 sandstone 182.97 1234.9 237.422 5.201366 -0.00039 0.31 1.228 2 2.5 38.18 -81.71 LC 4 1184.4 sandstone 198.49 1089.0 151.7451 7.176435 -0.00091 0.21 1.512 1 2.6 38.06 -81.82 LC 5 1220.0 sandstone 214.24 927.3 124.8901 7.424806 -0.00124 0.40 1.374 1 2.7 38.52 -81.66 LC 5 1125.1 shale 181.56 4258.4 341.84 12.45738 -0.00036 0.01 2.223 1 2.8 38.50 -80.67 LC 2 1381.6 sandstone 476.36 1269.0 206.9243 6.132545 -0.01112 0.33 1.339 1 2.9 38.11 -81.14 LC 6 1140.1 sandstone 253.53 1763.5 122.6544 14.37816 -0.00265 0.40 1.425 4 3.1 38.49 -80.43 UC 5 1278.1 shale 430.69 1405.7 205.0769 6.8545 -0.00502 0.31 2.179 2 3.2 37.73 -80.59 UC 5 982.6 limestone 486.63 622.7 201.6128 3.088487 -0.00182 0.36 1.323 2 3.3 38.21 -80.06 UC 3 1180.5 siltstone 667.90 1165.3 185.7366 6.27419 -0.00239 0.37 1.845 1 3.4 37.30 -80.71 UC 5 1002.7 shale 501.47 613.3 367.2545 1.669842 -0.002 0.42 2.277 1 3.5 38.08 -80.89 UC 4 1224.0 sandstone 578.63 1462.3 232.1829 6.298091 -0.00297 0.17 1.487 2 3.6 37.16 -80.86 UC 5 1002.6 dolostone 600.91 1038.2 327.6678 3.168429 -0.00175 0.20 2.583 1 3.7 37.43 -81.11 UC 4 956.1 shale 628.22 1193.0 306.5595 3.891709 -0.0004 0.26 3.029 2 3.8 37.74 -81.14 UC 3 1080.2 sandstone 697.44 766.6 474.1394 1.616841 -0.00647 0.19 1.913 1 3.9 38.06 -80.37 UC 2 1180.5 shale 745.46 1228.0 288.5835 4.255427 -0.01167 0.35 1.693 1
146 Table AIII.1 (cont.) Description of thirty-five sampling locations within the Kanawha River Basin.
Valley ValeylFloor DownValley Side Sine Plan SiteID Latitude Longitude FPZ SSO Precip Geology Elevation Width Width VW:VFW Slope Slope RC Form 5.1 38.50 -80.37 UH 4 1457.0 sandstone 512.96 1405.5 129.83 10.82602 -0.00727 0.45 1.499 1 5.2 38.53 -80.18 UH 4 1597.1 shale 670.97 2182.8 127.3056 17.14576 -0.00704 0.37 1.494 1 5.3 38.43 -80.41 UH 3 1451.8 shale 696.23 1211.8 157.2856 7.704318 -0.00098 0.27 1.812 1 5.4 38.30 -80.52 UH 3 1470.9 shale 759.57 3960.1 380.2126 10.41557 -0.0102 0.24 1.496 2 5.5 36.74 -80.53 UH 3 1320.2 metaargillite 765.42 1293.0 128.2772 10.08002 -0.00404 0.18 2.168 1 5.6 38.23 -80.44 UH 2 1466.7 shale 856.22 1130.3 141.5973 7.982567 -0.01423 0.33 1.469 1 5.7 38.58 -79.71 UH 3 1307.7 siltstone 965.94 196.0 107.9407 1.815553 -0.00988 0.42 1.323 1 5.8 38.04 -81.09 UH 2 1176.4 sandstone 562.15 614.2 190.7148 3.220736 -0.03044 0.13 1.673 1
147 Table AIII.2 Description of substrate composition as determined by the sounding rod method (Collins and Flotemersch 2013) at each site. Substrate types include fine-particle sediment (clay/silt), sand (< 2 mm), gravel (2-64 mm), cobble (65-256 mm), boulder (257-500 mm), and bedrock (large, unbroken surface).
Site Fine-particle Sand Gravel Cobble Boulder Bedrock 1.1 76 9 3 9 3 0 1.2 42 3 32 12 11 0 1.3 54 37 3 5 1 0 1.4 69 7 17 6 1 0 1.5 10 22 27 17 24 0 1.6 19 32 27 12 10 0 1.7 74 8 4 8 6 0 1.8 9 58 5 9 19 0 1.9 0 2 25 53 20 0 2.1 82 17 1 0 0 0 2.2 2 26 24 25 23 0 2.3 2 26 17 15 40 0 2.4 0 4 39 52 5 0 2.5 5 14 18 50 13 0 2.6 12 25 15 39 9 0 2.7 28 37 32 3 0 0 2.8 0 3 31 51 15 0 2.9 26 0 20 28 26 0 3.1 0 2 11 38 30 19 3.2 0 4 27 39 29 1 3.3 0 1 39 51 9 0 3.4 0 2 21 14 18 45 3.5 4 35 10 43 8 0 3.6 1 18 49 25 4 3 3.7 2 7 43 26 22 0 3.8 4 8 10 56 21 1 3.9 2 1 23 48 26 0 5.1 2 3 30 41 24 0 5.2 0 1 2 19 65 13 5.3 0 37 31 26 6 0 5.4 0 32 12 17 38 1 5.5 0 9 25 45 9 12 5.6 0 18 22 39 15 6 5.7 0 7 37 46 10 0 5.8 5 29 11 19 36 0
148 Table AIII.3 Nutrient concentrations measured at each site. Values are expressed as mg l-1. Where negative values exist, nutrient concentrations were below detection limit for the specified protocol. Sp or Su designation refers to spring or summer collection dates, respectively.
SiteID PO4Su PO4Sp TDPSu TDPSp NH4Su NH4Sp NO3Su NO3Sp TDNSu TDNSp DOCSu DOCSp 1.1 0.078 0.046 -0.030 -0.024 0.103 0.041 0.345 0.333 0.446 0.404 2.161 2.023 1.2 0.058 0.066 -0.033 -0.030 0.046 0.039 0.362 0.557 0.562 0.539 2.234 1.374 1.3 0.054 0.087 -0.036 -0.021 0.046 0.047 0.275 0.565 0.428 0.470 2.443 1.644 1.4 0.060 0.095 -0.028 -0.007 0.071 0.122 0.282 0.514 0.438 0.396 4.989 1.865 1.5 0.063 0.062 -0.036 -0.024 0.079 0.043 0.210 0.516 0.359 0.326 2.056 1.411 1.6 0.067 0.083 -0.034 -0.027 0.158 0.058 0.188 0.499 0.310 0.663 2.087 1.458 1.7 0.061 0.070 -0.032 -0.019 0.042 0.099 0.345 0.582 0.500 0.589 2.184 1.497 1.8 0.047 0.042 -0.039 -0.043 0.046 0.008 0.225 0.577 0.475 0.527 3.459 1.674 1.9 0.044 0.044 -0.036 -0.041 0.002 0.004 0.078 0.678 0.425 0.631 3.088 2.058 2.1 0.065 0.068 -0.035 -0.020 0.133 0.082 0.049 0.196 0.283 0.161 3.793 4.609 2.2 0.065 0.058 -0.037 -0.029 0.020 0.041 0.226 0.431 0.194 0.402 1.850 1.170 2.3 0.065 0.066 -0.027 -0.020 0.026 0.039 0.445 0.576 0.586 0.601 3.597 1.305 2.4 0.060 0.069 -0.041 -0.028 0.033 0.040 0.074 0.349 0.162 0.392 1.948 1.220 2.5 0.056 0.078 -0.040 -0.040 0.034 0.027 2.153 0.848 2.674 1.012 1.809 1.178 2.6 0.065 0.060 -0.031 -0.035 0.023 0.039 0.410 0.892 0.559 1.072 1.849 1.177 2.7 0.086 0.096 -0.026 -0.023 0.093 0.025 0.323 0.188 0.386 0.225 3.787 3.055 2.8 0.045 0.043 -0.041 -0.043 0.008 0.002 0.786 1.021 2.083 1.526 1.545 1.231 2.9 0.052 0.043 -0.025 -0.039 0.011 0.101 0.106 0.581 0.354 0.507 4.431 1.448 3.1 0.053 0.056 -0.042 -0.035 0.070 0.030 0.153 0.457 0.693 0.545 6.383 1.327 3.2 0.055 0.070 -0.037 -0.024 0.031 0.046 0.189 0.409 0.377 0.542 2.353 1.642 3.3 0.063 0.064 -0.041 -0.035 0.012 0.038 0.032 0.192 0.591 0.229 6.580 1.462 3.4 0.061 0.087 -0.039 -0.032 0.021 0.052 0.177 0.299 0.309 0.076 1.576 0.971 3.5 0.043 0.050 -0.041 -0.031 0.017 0.036 0.156 0.301 0.299 0.334 1.975 0.829 3.6 0.070 0.062 -0.026 -0.032 0.031 0.045 0.224 0.378 0.117 0.424 1.846 1.241 3.7 0.100 0.074 -0.004 -0.023 0.045 0.022 0.149 0.440 0.198 0.328 2.283 1.262 3.8 0.044 0.040 -0.034 -0.041 0.038 0.003 0.287 0.411 0.400 0.394 2.792 1.974 3.9 0.043 0.039 -0.035 -0.043 0.008 0.000 0.392 0.747 0.468 0.656 1.769 1.202 5.1 0.059 0.050 -0.042 -0.036 0.064 0.019 0.215 0.369 0.243 0.381 1.397 0.759 5.2 0.076 0.067 -0.041 -0.038 0.068 0.018 0.318 0.521 0.029 0.586 1.908 1.400 5.3 0.061 0.060 -0.041 -0.035 0.058 0.012 0.121 0.319 0.170 0.363 1.208 0.967 5.4 0.064 0.070 -0.036 -0.035 0.025 0.018 0.072 0.242 0.123 0.329 1.839 3.688 5.5 0.057 0.111 -0.039 -0.031 0.058 0.018 0.316 0.387 0.325 0.420 1.003 0.879 5.6 0.068 0.062 -0.041 -0.036 0.040 0.005 0.235 0.352 0.171 0.413 1.466 1.647 5.7 0.057 0.070 -0.040 -0.035 0.022 0.020 0.318 0.257 0.281 0.267 0.895 1.104 5.8 0.048 0.045 -0.034 -0.041 0.067 0.019 0.522 0.399 0.555 0.408 3.043 1.858
149 AIII.4 Ecosystem metabolism (i.e., gross primary productivity and ecosystem respiration) estimates for a subsample of 21 sites. Units for gross primary productivity (GPP) and ecosystem respiration (ER) are presented as oxygen concentration in g m-2 d-1.
SiteID GPP ER 1.1 28.99 145.28 1.2 1.25 0.91 1.3 1.57 10.39 2.1 13.21 66.15 2.2 1.64 8.61 2.3 1.94 2.56 2.4 2.87 5.56 2.5 2.89 4.00 2.6 7.37 9.83 3.1 4.10 9.63 3.2 4.79 9.29 3.3 5.00 10.79 3.4 21.62 34.90 3.5 4.62 9.05 3.6 7.90 10.01 3.7 12.55 26.70 5.1 3.86 9.83 5.2 3.12 8.71 5.3 3.12 12.67 5.4 2.28 14.49 5.5 3.87 14.38
150 Table AIII.5 Stable isotope signature for aquatic and terrestrial primary producers (organic carbon sources) sampled from each site within the Kanawha River Basin. Mean and standard deviation (SD) for isotopes of nitrogen and carbon are given.
Source Mean d15N SD d15N Mean d13C SD d13C Tree Leaves 4.6961 2.2195 -30.9572 1.3433 Aquatic Vascular 11.7113 1.6999 -29.7424 3.7910 Benthic Algae 6.3736 4.4743 -23.4395 6.1005 Terrestrial Detritus 4.7433 2.0697 -30.2910 3.5130
151 Table AIII.6 Stable isotope analysis results for all food web metrics. Metrics include nitrogen (δ15N) range, carbon range, hull area, centroid distance, and nearest neighbor distance (NND; mean and standard deviation). Based on Jackson et al. (2011), SIBER (a function within SIAR) was used to calculated these variables and the 95% confidence interval (CI) around them.
SIBER (SIAR) 95% CI Site NR CR HullArea CentDist MeanNND SDNND NR CR HullArea CentDist MeanNND SDNND 1.1 11.65 13.92 82.03 3.32 1.63 2.06 0.22 0.23 1.96 0.03 0.02 0.07 1.2 13.82 13.08 92.49 3.80 1.89 2.14 0.27 0.27 2.08 0.04 0.02 0.09 1.3 12.96 14.23 96.32 3.14 1.43 1.96 0.20 0.23 2.61 0.02 0.01 0.05 1.4 13.00 15.59 103.87 3.46 1.66 2.15 0.28 0.28 4.58 0.03 0.02 0.08 1.5 12.23 14.91 98.73 3.42 1.69 2.15 0.23 0.26 3.85 0.03 0.02 0.07 1.6 15.94 12.81 111.71 4.74 2.34 2.49 0.19 0.21 2.15 0.04 0.02 0.07 1.7 15.01 9.44 76.71 5.03 2.70 2.25 0.14 0.17 1.39 0.04 0.03 0.06 1.8 13.68 11.79 84.54 4.62 2.67 2.22 0.17 0.20 1.73 0.04 0.03 0.07 1.9 15.16 13.24 105.83 4.61 2.16 2.19 0.21 0.21 2.01 0.03 0.02 0.07 2.1 15.84 19.60 167.48 3.67 1.35 2.10 0.34 0.38 4.76 0.02 0.01 0.08 2.2 16.82 15.53 135.96 3.39 1.39 2.15 0.35 0.32 4.12 0.03 0.02 0.08 2.3 12.46 16.79 112.85 3.79 1.68 2.01 0.21 0.24 2.84 0.02 0.02 0.06 2.4 15.36 17.58 144.26 4.15 1.72 2.21 0.38 0.51 5.86 0.05 0.03 0.13 2.5 15.72 14.75 118.18 3.45 1.42 2.07 0.24 0.30 2.99 0.02 0.02 0.07 2.6 16.85 16.35 143.26 3.63 1.41 2.12 0.31 0.29 3.61 0.02 0.01 0.07 2.7 15.45 12.27 101.12 3.85 2.01 2.42 0.17 0.22 1.99 0.03 0.02 0.06 2.8 15.74 10.75 84.93 4.66 2.70 2.44 0.16 0.17 1.70 0.04 0.02 0.06 2.9 10.27 9.68 49.97 3.18 1.94 2.10 0.16 0.19 1.29 0.04 0.02 0.07 3.1 14.47 11.56 91.41 5.01 2.45 2.00 0.18 0.26 1.81 0.05 0.03 0.09 3.2 15.02 13.80 108.80 3.75 1.63 2.11 0.64 0.52 4.59 0.07 0.04 0.18 3.3 14.00 14.48 105.74 3.80 1.75 2.11 0.61 0.27 4.05 0.06 0.04 0.15 3.4 18.29 14.24 142.12 5.16 1.78 2.06 0.25 0.27 2.92 0.03 0.02 0.07 3.5 11.80 13.48 85.98 4.10 2.17 2.08 0.19 0.21 1.86 0.04 0.02 0.07 3.6 14.81 15.89 125.88 4.30 1.73 2.07 0.26 0.31 2.81 0.03 0.02 0.08 3.7 13.47 13.19 92.85 3.58 1.63 2.04 0.20 0.26 2.45 0.03 0.02 0.07 3.8 12.29 12.01 78.77 3.85 2.11 2.10 0.18 0.18 1.57 0.03 0.02 0.06 3.9 8.09 5.76 18.72 3.64 3.65 2.36 0.12 0.10 0.44 0.05 0.04 0.06 5.1 13.22 16.77 116.17 3.81 1.59 2.00 0.25 0.26 2.53 0.03 0.02 0.07 5.2 5.21 7.29 16.22 3.02 2.48 1.77 0.11 0.13 0.47 0.05 0.04 0.07 5.3 12.29 13.93 88.85 3.55 1.74 2.12 0.32 0.22 3.32 0.04 0.02 0.09 5.4 11.45 15.79 81.58 4.23 2.10 2.30 0.15 0.25 1.80 0.04 0.02 0.08 5.5 11.49 12.62 72.76 3.22 1.59 2.02 0.30 0.29 2.56 0.04 0.03 0.10 5.6 11.67 8.80 49.30 4.37 2.39 1.99 0.17 0.27 1.46 0.06 0.04 0.11 5.7 11.50 16.79 96.33 4.52 2.03 1.98 0.18 0.17 1.74 0.03 0.02 0.06 5.8 7.23 5.52 15.61 3.37 3.33 1.87 0.10 0.14 0.51 0.06 0.04 0.07
152 Table AIII.7 Stable isotopes of carbon and nitrogen for each consumer sampled by site. Units are expressed as ‰ relative to a reference standard.
Site ID 1.1 1.2 1.3 FPZ/SSO LA/6 LA/6 LA/6
Family Genus Species Common Name δ13C δ15N δ13C δ15N δ13C δ15N Catostomidae Carpiodes carpio River carpsucker -25.4 10.6 -25.4 14.0 Catostomidae Ictiobus bubalus Smallmouth buffalo -25.4 10.1 -26.5 13.4 -25.6 10.7 Catostomidae Moxostoma carinatum River redhorse -26.7 12.0 Catostomidae Moxostoma erythrurum Golden redhorse -26.6 11.8 -25.8 9.7 Centrarchidae Lepomis cyanellus Green sunfish -25.0 11.8 Centrarchidae Lepomis macrochirus Bluegill -26.2 11.7 -27.2 13.0 Centrarchidae Lepomis megalotis Longear sunfish -27.2 10.9 -27.2 12.6 -25.9 10.8 Centrarchidae Lepomis spp. Juvenile sunfish -26.7 11.9 -28.1 13.1 -24.4 9.6 Centrarchidae Micropterus dolomieu Smallmouth bass -26.2 12.0 -27.2 13.9 -25.8 11.9 Centrarchidae Micropterus dolomieu Smallmouth bass (juv) -27.5 12.2 -26.2 11.5 Centrarchidae Micropterus punctulatus Spotted bass -26.9 11.8 -27.7 12.5 -25.7 12.6 Centrarchidae Micropterus punctulatus Spotted bass (juv) -26.9 11.5 Centrarchidae Micropterus salmoides Largemouth bass -28.6 12.6 Clupeidae Dorosoma cepedianum Gizzard shad -27.5 9.9 -27.7 10.4 Corbiculidae Corbicula fluminea Asian clam -29.8 8.6 -27.3 6.6 -29.9 8.4 Cyprinidae Campostoma anomalum Central stoneroller -23.5 11.6 Cyprinidae Cyprinella spiloptera Spotfin shiner -26.2 9.3 -25.8 8.8 Cyprinidae Erimystax dissimilis Streamline chub -21.7 10.9 Cyprinidae Notropis Unknown shiner -26.9 11.8 -26.4 10.5 -27.0 10.9 Cyprinidae Pimphales notatus Bluntnose minnow -21.7 10.4 -23.8 10.4 Heptageniidae Flat-headed mayfly -27.4 8.1 -24.6 11.7 -26.9 13.8 Lepisosteidae Lepisosteus osseus Longnose gar -26.1 15.0 Percidae Etheostoma blennioides Greenside darter -23.3 12.5 Percidae Etheostoma caeruleum Rainbow darter -24.0 11.8 Percidae Etheostoma nigrum Johnny darter -24.5 11.3 Percidae Percina caprodes Logperch -23.7 10.8 -25.6 12.6 -25.1 11.0 Pleuroceridae Pleurocera -22.3 9.5 -25.1 10.7 Sciaenidae Aplodinotus grunniens Freshwater drum -25.3 12.0 -26.8 12.4 Viviparidae Campeloma -24.4 5.3
153 Site ID 1.4 1.5 1.6 FPZ/SSO LA/6 LA/6 LA/6
Family Genus Species Common Name δ13C δ15N δ13C δ15N δ13C δ15N Atherinidae Labidesthes sicculus Brook silverside -28.2 16.0 Caenidae Small squaregill mayfly -25.0 7.2 Catostomidae Carpiodes carpio River carpsucker -25.0 10.9 -26.2 10.8 Catostomidae Ictiobus bubalus Smallmouth buffalo -26.9 13.1 -26.0 10.8 -27.6 15.4 Catostomidae Moxostoma erythrurum Golden redhorse -25.1 10.6 -25.8 10.5 Centrarchidae Ambloplites rupestris Rock bass -24.2 11.3 -27.1 17.2 Centrarchidae Lepomis cyanellus Green sunfish -24.5 12.3 -23.4 10.5 Centrarchidae Lepomis macrochirus Bluegill -26.9 11.9 -25.9 10.9 -28.8 16.2 Centrarchidae Lepomis megalotis Longear sunfish -26.6 11.3 -23.7 10.9 -26.2 16.5 Centrarchidae Lepomis spp. Juvenile sunfish -22.9 11.4 Centrarchidae Micropterus dolomieu Smallmouth bass -27.1 13.1 -25.2 12.0 -27.3 17.4 Centrarchidae Micropterus punctulatus Spotted bass -24.8 12.1 Centrarchidae Micropterus punctulatus Spotted bass (juv) -27.0 13.7 Centrarchidae Pomoxis annularis White crappie -29.5 18.5 Centrarchidae Poxomis nigromaculatus Black crappie -28.4 18.0 Clupeidae Dorosoma cepedianum Gizzard shad -26.3 11.8 Corbiculidae Corbicula fluminea Asian clam -28.7 8.6 -29.2 6.9 -29.1 7.1 Cyprinidae Campostoma anomalum Central stoneroller -25.1 9.3 Cyprinidae Cyprinella spiloptera Spotfin shiner -25.7 10.5 -24.5 11.1 Cyprinidae Cyprinus carpio Common carp -27.8 10.6 Cyprinidae Notropis atherinoides Emerald shiner -28.1 15.7 Cyprinidae Notropis Unknown shiner -26.9 11.8 -26.0 10.8 Cyprinidae Pimphales notatus Bluntnose minnow -28.2 16.0 Heptageniidae Flat-headed mayfly -26.5 10.8 -21.2 7.2 -23.2 7.1 Percidae Etheostoma blennioides Greenside darter -19.6 11.5 Percidae Percina caprodes Logperch -24.8 10.8 -22.2 11.3 Percidae Percina phoxocephala Slenderhead darter -25.6 11.5 Physidae Physella -18.9 9.2 Pleuroceridae Pleurocera -23.8 10.2 -22.4 9.3 Sciaenidae Aplodinotus grunniens Freshwater drum -27.6 11.8 Viviparidae Campeloma -26.2 6.9
154 Site ID 1.7 1.8 1.9 FPZ/SSO LA/6 LA/4 LA/4
Family Genus Species Common Name δ13C δ15N δ13C δ15N δ13C δ15N Catostomidae Hypentelium nigricans Northern hogsucker -30.3 14.9 Catostomidae Ictiobus bubalus Smallmouth buffalo -25.8 14.2 Centrarchidae Ambloplites rupestris Rock bass -29.0 16.1 -25.6 16.1 Centrarchidae Lepomis cyanellus Green sunfish -29.1 14.6 Centrarchidae Lepomis macrochirus Bluegill -26.9 16.3 Centrarchidae Lepomis megalotis Longear sunfish -25.9 16.8 Centrarchidae Lepomis spp. Juvenile sunfish -30.5 14.7 Centrarchidae Micropterus dolomieu Smallmouth bass -27.0 17.0 -26.5 16.5 Corbiculidae Corbicula fluminea Asian clam -28.8 7.1 -32.3 6.7 -27.2 5.9 Cyprinidae Campostoma anomalum Central stoneroller -23.9 15.0 Cyprinidae Notropis atherinoides Emerald shiner -27.0 15.8 Cyprinidae Notropis telescopus Telescope shiner -26.7 15.1 Cyprinidae Notropis Unknown shiner -25.1 15.4 Cyprinidae Pimphales notatus Bluntnose minnow -24.6 15.5 -26.1 14.4 Cyprinidae Semotilus atromaculatus Creek chub -26.9 15.0 -27.7 15.1 Heptageniidae Flat-headed mayfly -27.1 5.1 -32.4 12.0 -32.1 11.5 Limnephilidae Tube-case caddisfly -29.4 7.5 -29.3 9.2 Percidae Etheostoma blennioides Greenside darter -28.7 14.7 Percidae Etheostoma flabellare Fantail darter -30.3 16.4 Percidae Etheostoma zonale Banded darter -29.5 15.3 Pleuroceridae Leptoxis -29.0 6.8 Pleuroceridae Pleurocera -23.8 9.5
155 Site ID 2.1 2.2 2.3 FPZ/SSO LC/5 LC/5 LC/6
Family Genus Species Common Name δ13C δ15N δ13C δ15N δ13C δ15N Baetidae Small minnow mayfly -23.0 10.2 Catostomidae Catostomus commersonii White sucker -23.4 9.4 Catostomidae Hypentelium nigricans Northern hogsucker -24.3 10.5 -21.8 10.1 -20.4 9.7 Catostomidae Ictiobus bubalus Smallmouth buffalo -29.0 8.3 Catostomidae Moxostoma duquesni Black redhorse -22.9 10.5 Catostomidae Moxostoma erythrurum Golden redhorse -27.0 10.8 Catostomidae Moxostoma macrolepidotum Shorthead redhorse -22.9 10.3 Centrarchidae Ambloplites rupestris Rock bass -22.5 11.0 -23.6 11.6 Centrarchidae Lepomis cyanellus Green sunfish -26.0 10.3 -24.5 10.4 Centrarchidae Lepomis macrochirus Bluegill -27.2 9.6 -25.0 9.1 Centrarchidae Lepomis megalotis Longear sunfish -26.4 9.4 -22.0 10.3 Centrarchidae Lepomis spp. Juvenile sunfish -30.6 10.0 Centrarchidae Lepomis Redbreast sunfish -27.2 9.4 Centrarchidae Micropterus dolomieu Smallmouth bass -25.3 12.0 -21.6 10.1 -20.7 11.0 Centrarchidae Micropterus punctulatus Spotted bass -30.9 11.3 Centrarchidae Micropterus salmoides Largemouth bass -31.2 12.0 -22.5 10.6 Centrarchidae Pomoxis annularis White crappie -31.7 11.8 Clupeidae Dorosoma cepedianum Gizzard shad -30.6 10.2 Corbiculidae Corbicula fluminea Asian clam -28.9 5.8 -26.3 5.7 -27.2 11.4 Cyprinidae Campostoma anomalum Central stoneroller -25.8 10.5 Cyprinidae Cyprinella spiloptera Spotfin shiner -28.9 9.2 -24.2 8.9 -24.8 7.8 Cyprinidae Cyprinus carpio Common carp -26.6 8.4 Cyprinidae Erimystax dissimilis Streamline chub -22.6 10.6 Cyprinidae Luxilus cornutus Common shiner -24.7 8.2 Cyprinidae Notropis Unknown shiner -23.5 9.2 -24.1 8.8 Cyprinidae Pimphales notatus Bluntnose minnow -28.6 8.2 -24.8 9.5 Cyprinidae Pimphales promelas Fathead minnow -26.6 9.6 Esocidae Sander vitreus Walleye -21.5 13.0 Heptageniidae Flat-headed mayfly -34.2 6.3 -23.2 4.9 -24.2 10.7 Hydropsychidae Net-spinning caddisfly -23.1 7.3 Ictaluridae Ameiurus natalis Yellow bullhead -25.2 10.9 Ictaluridae Ictalurus punctatus Channel catfish -27.7 10.9 Ictaluridae Noturus miurus Brindled madtom -23.5 11.0 Ictaluridae Pylodictis olivaris Flathead catfish -23.2 10.8 Limnephilidae Tube-case caddisfly -25.3 3.3 Moronidae Morone chrysops White bass -30.6 12.0 -22.4 10.5 Percidae Etheostoma blennioides Greenside darter -25.9 11.4 -20.4 10.8 Percidae Etheostoma caeruleum Rainbow darter -26.0 11.3 Percidae Etheostoma flabellare Fantail darter -28.0 11.7 -22.3 11.6 Percidae Etheostoma nigrum Johnny darter -24.8 11.3 Percidae Etheostoma zonale Banded darter Percidae Percina caprodes Logperch -26.0 10.1 Philopotamidae Finger-net caddisfly -23.3 6.8 -27.1 6.4 Physidae Physella -18.7 7.9 Planorbidae Planorbella -16.3 8.0 Polycentropodidae Trumpet -net caddisfly -22.1 5.5 Sciaenidae Aplodinotus grunniens Freshwater drum -26.9 10.4 -22.0 10.9 Viviparidae Campeloma -20.6 8.1
156 Site ID 2.4 2.5 2.6 FPZ/SSO LC/5 LC/4 LC/5
Family Genus Species Common Name δ13C δ15N δ13C δ15N δ13C δ15N Brachycentridae Humpless casemaker caddisfly -28.3 10.2 Caenidae Small squaregill mayfly -27.5 5.1 Catostomidae Hypentelium nigricans Northern hogsucker -23.1 9.6 -24.7 10.9 -26.4 14.2 Catostomidae Ictiobus bubalus Smallmouth buffalo -24.1 9.2 Catostomidae Moxostoma carinatum River redhorse -24.4 9.9 -23.8 11.7 -24.6 12.2 Catostomidae Moxostoma duquesni Black redhorse -23.7 13.5 -26.2 14.8 Catostomidae Moxostoma erythrurum Golden redhorse -25.3 9.6 -22.8 10.9 -26.2 13.2 Catostomidae Moxostoma macrolepidotum Shorthead redhorse -23.0 11.0 -24.6 14.9 Centrarchidae Ambloplites rupestris Rock bass -24.0 9.3 -24.4 11.5 -24.5 12.8 Centrarchidae Lepomis macrochirus Bluegill -26.3 10.6 -26.1 12.9 Centrarchidae Lepomis megalotis Longear sunfish -25.1 9.4 -24.0 12.4 Centrarchidae Micropterus dolomieu Smallmouth bass -22.9 13.2 -28.2 16.4 Corbiculidae Corbicula fluminea Asian clam -27.3 5.6 -24.6 7.1 -25.6 9.8 Cyprinidae Campostoma anomalum Central stoneroller -27.3 13.8 Cyprinidae Cyprinella spiloptera Spotfin shiner -23.1 10.1 -25.3 11.1 Cyprinidae Cyprinus carpio Common carp -24.6 10.3 Cyprinidae Luxilus cornutus Common shiner -25.0 7.4 Cyprinidae Notropis Unknown shiner -28.2 8.4 -26.4 10.5 -25.2 9.9 Cyprinidae Pimphales notatus Bluntnose minnow -24.7 9.1 -24.4 10.6 -26.1 13.0 Heptageniidae Flat-headed mayfly -32.6 4.3 -25.1 4.8 -27.4 8.1 Hydropsychidae Net-spinning caddisfly -25.8 7.0 -25.4 8.6 Ictaluridae Ictalurus punctatus Channel catfish -24.5 10.5 -25.0 10.1 -22.5 10.6 Ictaluridae Noturus miurus Brindeled madtom -25.1 14.8 Isonychiidae Brushlegged mayfly -23.3 3.3 -23.9 8.6 -26.3 7.1 Limnephilidae Tube-case caddisfly -26.0 4.0 Percidae Etheostoma blennioides Greenside darter -21.9 10.8 -24.6 13.2 Percidae Etheostoma caeruleum Rainbow darter -26.5 14.2 Percidae Percina caprodes Logperch -23.9 9.6 -24.4 14.9 Petromyzontidae Ichthyomyzon unicuspis Silver lamprey -23.4 15.1 Philopotamidae Finger-net caddisfly -30.3 5.1 Physidae Physella -25.7 8.7 -25.6 11.0 Planorbidae Planorbella -25.8 10.4 Pleuroceridae Pleurocera -22.6 8.4 Sciaenidae Aplodinotus grunniens Freshwater drum -25.7 12.2 -23.8 11.3 -24.4 13.7 Viviparidae Campeloma -23.5 8.2
157 Site ID 2.7 2.8 2.9 FPZ/SSO LC/5 LC/2 LC/6
Family Genus Species Common Name δ13C δ15N δ13C δ15N δ13C δ15N Atherinidae Labidesthes sicculus Brook silverside -23.2 10.5 Catostomidae Hypentelium nigricans Northern hogsucker -29.7 16.0 -20.0 13.0 Catostomidae Moxostoma erythrurum Golden redhorse -29.2 16.9 Centrarchidae Ambloplites rupestris Rock bass -23.8 10.8 Centrarchidae Lepomis macrochirus Bluegill -29.6 16.0 -23.5 10.5 Centrarchidae Lepomis megalotis Longear sunfish -29.4 16.4 -23.7 10.8 Centrarchidae Micropterus dolomieu Smallmouth bass -20.2 14.2 -22.2 12.1 Centrarchidae Micropterus punctulatus Spotted bass -28.8 17.4 -23.1 11.7 Centrarchidae Micropterus salmoides Largemouth bass -21.6 12.2 Clupeidae Dorosoma cepedianum Gizzard shad -30.2 14.1 Corbiculidae Corbicula fluminea Asian clam -33.7 7.1 -24.8 6.2 Cottidae Cottus bairdii Mottled sculpin -25.0 17.6 Cyprinidae Campostoma anomalum Central stoneroller -21.8 12.7 Cyprinidae Cyprinella spiloptera Spotfin shiner -30.1 15.2 -23.0 9.2 Cyprinidae Notropis atherinoides Emerald shiner -29.7 16.0 -22.2 10.0 Cyprinidae Notropis Unknown shiner -23.1 9.9 Cyprinidae Pimphales notatus Bluntnose minnow -30.1 15.2 Cyprinidae Semotilus atromaculatus Creek chub -23.8 8.6 Heptageniidae Flat-headed mayfly -29.9 8.5 -23.8 8.6 -20.9 7.7 Lepisosteidae Lepisosteus osseus Longnose gar -25.4 3.7 Percidae Etheostoma blennioides Greenside darter -29.2 17.2 -19.5 13.6 Percidae Etheostoma caeruleum Rainbow darter -20.2 13.1 Percidae Etheostoma nigrum Johnny darter -32.7 17.5 Percidae Etheostoma zonale Banded darter -29.1 16.7 Percidae Percina caprodes Logperch -29.7 17.0
158 Site ID 3.4 3.5 3.6 FPZ/SSO UC/5 UC/4 UC/5
Family Genus Species Common Name δ13C δ15N δ13C δ15N δ13C δ15N Humpless casemaker Brachycentridae caddisfly -27.0 2.4 -28.6 8.5 Catostomidae Hypentelium nigricans Northern hogsucker -16.8 9.8 -25.2 13.3 Centrarchidae Ambloplites rupestris Rock bass -23.9 13.7 -22.8 10.3 -26.3 14.5 Centrarchidae Lepomis auritus Redbreast sunfish -26.1 12.6 -26.9 13.2 Centrarchidae Lepomis macrochirus Bluegill -23.4 9.8 Centrarchidae Micropterus dolomieu Smallmouth bass -25.4 13.3 -21.0 10.5 -24.8 12.9 Corbiculidae Corbicula fluminea Asian clam -25.5 5.5 -24.2 6.6 Cottidae Cottus bairdii Mottled sculpin -26.8 11.7 -25.0 12.7 Cyprinidae Campostoma anomalum Central stoneroller -25.2 12.5 -22.9 12.7 Cyprinidae Cyprinella spiloptera Spotfin shiner -21.7 12.2 Cyprinidae Nocomis platyrhynchus Bigmouth chub -25.0 11.9 Cyprinidae Notropis Unknown shiner -22.6 12.4 -24.9 11.2 Cyprinidae Pimphales notatus Bluntnose minnow -23.9 13.3 -20.9 8.7 -28.7 11.2 Cyprinidae Rhinichthys obtusus Western blacknose dace -25.2 12.4 Cyprinidae Semotilus atromaculatus Creek chub -19.4 9.3 Glossosomatidae Saddle casemaker caddisfly -32.0 6.9 Helicopsychidae Snail-case caddisfly -25.5 2.5 Heptageniidae Flat-headed mayfly -26.0 2.6 -20.4 3.4 -28.6 6.6 Hydrobiidae Birgella -28.5 6.5 Hydropsychidae Net -spinning caddisfly -27.8 3.6 -19.6 5.3 -27.9 9.1 Ictaluridae Noturus insignis Margined madtom -26.4 13.4 Limnephilidae Tube-case caddisfly -26.2 5.1 Mortarjoint Odontoceridae casemaker caddisfly -25.9 5.1 -28.9 8.5 Percidae Etheostoma blennioides Greenside darter -27.6 12.3 Percidae Etheostoma nigrum Johnny darter -23.4 8.7 Pleuroceridae Elimia -24.7 6.9 -25.6 7.3 Uenoidae Stonecase caddisfly -31.2 6.3
159 Site ID 3.7 3.8 3.9 FPZ/SSO UC/4 UC/3 UC/2
Family Genus Species Common Name δ13C δ15N δ13C δ15N δ13C δ15N Humpless casemaker Brachycentridae caddisfly -26.9 7.3 Catostomidae Catostomus commersonii White sucker -19.0 5.6 Catostomidae Hypentelium nigricans Northern hogsucker -20.9 11.4 Catostomidae Moxostoma erythrurum Golden redhorse -24.1 10.0 Centrarchidae Ambloplites rupestris Rock bass -23.7 13.1 -23.3 11.8 Centrarchidae Lepomis cyanellus Green sunfish -25.8 9.7 Centrarchidae Lepomis megalotis Longear sunfish -23.4 13.0 Centrarchidae Lepomis spp. Juvenile sunfish -23.8 12.3 Centrarchidae Micropterus dolomieu Smallmouth bass -23.9 14.0 -22.4 12.5 Corbiculidae Corbicula fluminea Asian clam -24.0 7.6 Cyprinidae Campostoma anomalum Central stoneroller -25.0 12.0 -22.7 11.0 Cyprinidae Notropis Unknown shiner -23.7 9.8 Cyprinidae Pimphales notatus Bluntnose minnow -22.8 10.4 Cyprinidae Rhinichthys cataractae Longnose dace -25.5 12.4 Cyprinidae Rhinichthys obtusus Western blacknose dace -21.1 6.1 Cyprinidae Semotilus atromaculatus Creek chub -24.5 12.0 -25.1 8.6 -22.2 6.6 Heptageniidae Flat-headed mayfly -27.2 8.7 -23.4 7.3 Hydropsychidae Net-spinning caddisfly -27.5 9.1 Ictaluridae Ictalurus punctatus Channel catfish -23.2 12.3 Limnephilidae Tube-case caddisfly -27.1 4.5 Percidae Etheostoma blennioides Greenside darter -24.7 14.4 Percidae Etheostoma flabellare Fantail darter -22.6 12.3 Physidae Physella -21.8 7.1 Planorbidae Planorbella -19.7 7.1 Pleuroceridae Leptoxis -26.6 10.0 Uenoidae Stonecase caddisfly -22.3 0.2
160 Site ID 5.1 5.2 5.3 FPZ/SSO UH/4 UH/4 UH/3
Family Genus Species Common Name δ13C δ15N δ13C δ15N δ13C δ15N Atherinidae Labidesthes sicculus Brook silverside -22.2 8.6 Humpless casemaker Brachycentridae caddisfly -23.3 7.3 -22.6 3.5 Catostomidae Catostomus commersonii White sucker -26.3 6.7 -18.6 3.7 Centrarchidae Ambloplites rupestris Rock bass -19.6 8.3 -23.9 9.0 Centrarchidae Lepomis cyanellus Green sunfish -22.7 9.1 Centrarchidae Micropterus dolomieu Smallmouth bass -20.2 9.0 -23.8 8.0 Cottidae Cottus bairdii Mottled sculpin -18.5 6.4 Cyprinidae Luxilus cornutus Common shiner -20.3 7.1 -24.1 7.3 Cyprinidae Notropis atherinoides Emerald shiner -18.0 8.5 Cyprinidae Notropis Unknown shiner -23.4 8.5 Cyprinidae Pimphales notatus Bluntnose minnow -25.5 7.7 Cyprinidae Rhinichthys cataractae Longnose dace -17.7 5.9 -18.4 4.2 Heptageniidae Flat-headed mayfly -21.3 5.0 -21.4 4.6 -29.1 5.2 Hydropsychidae Net-spinning caddisfly -22.8 6.7 -29.9 7.3 Isonychiidae Brushlegged mayfly -22.7 4.8 Limnephilidae Tube-case caddisfly -27.5 5.2 -27.9 5.8 Neoephemeridae -24.9 2.8 Percidae Etheostoma caeruleum Rainbow darter -16.8 8.1 Percidae Etheostoma flabellare Fantail darter -18.4 8.1 -25.5 9.5 Percidae Etheostoma nigrum Johnny darter -25.4 8.3 Percidae Etheostoma zonale Banded darter -18.3 5.2 Percidae Etheostoma olmstedi Tesselated darted -22.3 7.2 Petromyzontidae Lampetra aepyptera Least brook lamprey -24.2 3.1 Philopotamidae Finger-net caddisfly -23.1 8.0 Polycentropodidae Trumpet-net caddisfly -24.5 8.3
161 Site ID 5.4 5.5 5.6 FPZ/SSO UH/3 UH/3 UH/2
Family Genus Species Common Name δ13C δ15N δ13C δ15N δ13C δ15N Catostomidae Catostomus commersonii White sucker -24.5 8.2 Catostomidae Hypentelium nigricans Northern hogsucker -19.7 6.2 -25.8 8.2 Centrarchidae Ambloplites rupestris Rock bass -20.8 8.5 Centrarchidae Micropterus dolomieu Smallmouth bass -19.4 7.4 Cottidae Cottus bairdii Mottled sculpin -23.3 7.9 Cyprinidae Notropis heterolepis Blacknose shiner -21.5 6.6 Cyprinidae Rhinichthys cataractae Longnose dace -20.2 6.2 Cyprinidae Rhinichthys obtusus Western blacknose dace -24.5 6.9 Cyprinidae Nocomis micropogon River chub -20.7 6.7 Ephemeridae Burrowing mayfly -25.0 5.4 Saddle casemaker Glossosomatidae caddisfly -24.2 2.2 -28.0 -0.1 Heptageniidae Flat-headed mayfly -25.9 4.1 -25.9 2.5 Hydropsychidae Net-spinning caddisfly -24.3 3.8 -25.4 6.8 -27.2 2.1 Ictaluridae Noturus miurus Brindled madtom -23.7 8.5 Isonychiidae Brushlegged mayfly -26.1 5.7 Limnephilidae Tube-case caddisfly -30.3 0.8 -26.0 0.1 Mortarjoint casemaker Odontoceridae caddisfly -24.3 4.3 Percidae Etheostoma flabellare Fantail darter -19.7 7.1 -25.3 7.2 Percidae Etheostoma nigrum Johnny darter -17.6 6.1 Polycentropodidae Trumpet-net caddisfly -23.8 3.6 Salmonidae Salmo trutta Brown trout -23.2 7.7 -24.8 7.8 Salmonidae Salvelinus fontinalis Brook trout -24.0 8.0
162 Site ID 5.7 5.8 FPZ/SSO UH/3 UH/2
Family Genus Species Common Name δ13C δ15N δ13C δ15N Catostomidae Catostomus commersonii White sucker -22.6 6.0 Centrarchidae Lepomis spp. Juvenile sunfish -27.0 11.2 Cottidae Cottus bairdii Mottled sculpin -23.9 6.5 Cyprinidae Campostoma anomalum Central stoneroller -17.3 6.2 Cyprinidae Semotilus atromaculatus Creek chub -26.4 9.2 Cyprinidae Rhinichthys cataractae Longnose dace -21.5 6.7 Cyprinidae Phoxinus erythrogaster Red bellied dace -21.7 6.1 Saddle casemaker Glossosomatidae caddisfly -30.8 3.3 Heptageniidae Flat-headed mayfly -27.7 2.6 -28.7 8.0 Hydropsychidae Net-spinning caddisfly -29.4 2.1 Limnephilidae Tube-case caddisfly -25.9 0.3 -27.8 5.3 Percidae Etheostoma flabellare Fantail darter -22.8 6.8 Philopotamidae Finger-net caddisfly -24.1 5.8 Salmonidae Salmo trutta Brown trout -22.7 7.1 Salmonidae Salvelinus fontinalis Brook trout -23.2 6.2
163
Table AIII.8 Stable isotopes of carbon and nitrogen for each producer sampled by site. Units are expressed as ‰ relative to a reference standard.
Benthic Algae Aquatic Vascular Tree Leaves Terrestrial Detritus Site FPZ SSO δ13C δ15N δ13C δ15N δ13C δ15N δ13C δ15N 1.1 LA 6 -23.47 10.39 -33.16 12.35 -32.45 6.98 -27.38 6.95 1.2 LA 6 -20.06 8.40 -33.13 11.40 -31.45 4.58 -26.52 4.70 1.3 LA 6 -17.11 -26.92 11.92 -30.60 4.19 -29.58 5.46 1.4 LA 6 -34.20 -27.02 12.87 -31.62 5.65 -29.84 6.62 1.5 LA 6 -18.18 -25.97 13.83 -32.38 5.92 -28.31 7.58 1.6 LA 6 -25.54 9.21 -31.48 3.23 -31.51 6.21 1.7 LA 6 -13.78 10.11 -31.38 10.69 -32.31 4.74 -30.46 5.80 1.8 LA 4 -26.39 10.61 -29.09 11.70 -30.93 4.38 -29.57 4.27 1.9 LA 4 -25.97 9.59 -31.35 4.76 -29.94 2.1 LC 5 -30.16 -31.15 3.28 -33.80 2.52 2.2 LC 5 -29.53 9.13 -32.41 11.68 -29.47 2.31 -36.72 4.98 2.3 LC 6 -28.54 10.80 -31.25 4.40 -30.45 9.58 2.4 LC 5 -26.73 -32.75 12.08 -31.04 4.16 -30.52 3.32 2.5 LC 4 -26.03 -32.96 14.42 -32.06 7.32 -30.03 3.93 2.6 LC 5 -17.53 11.07 -28.93 15.13 -30.99 5.63 -21.25 4.77 2.7 LC 5 -30.43 11.67 -32.33 12.33 -29.91 7.32 -36.05 7.77 2.8 LC 2 -24.13 -28.33 3.77 -28.14 4.37 2.9 LC 6 -19.22 11.91 -19.46 8.48 -32.12 6.87 -29.95 2.49 3.1 UC 5 -17.85 6.60 -31.63 10.81 -34.87 2.70 3.2 UC 5 -12.67 6.91 -24.50 9.62 -27.17 3.3 UC 3 -16.90 7.67 -32.03 4.43 -31.24 3.83 3.4 UC 5 -4.99 8.20 -32.22 10.21 -31.03 8.51 -30.87 4.13 3.5 UC 4 -29.07 6.71 -31.85 12.27 -29.91 5.09 -31.17 3.18 3.6 UC 5 -26.79 10.26 -35.08 11.01 -31.48 8.06 -29.95 7.24 3.7 UC 4 -24.67 12.99 -30.84 9.23 -29.18 7.73 3.8 UC 3 -26.48 7.61 -30.16 4.80 -34.51 3.9 UC 2 -31.05 6.45 -30.27 3.05 -31.37 5.1 UH 4 -21.15 4.56 -30.56 3.20 -37.97 5.2 UH 4 -20.82 7.39 -30.92 5.15 -23.82 3.11 5.3 UH 3 -29.54 4.90 -31.79 10.06 -29.99 1.97 -23.15 2.38 5.4 UH 3 -24.11 3.88 -31.72 4.30 -31.30 5.5 UH 3 -25.92 5.14 -29.29 2.69 -28.37 5.6 UH 2 -26.58 7.28 -31.28 2.44 -32.14 1.32 5.7 UH 3 -23.20 6.72 -31.15 0.74 -32.09 3.46 5.8 UH 2 -26.71 10.11 -30.57 3.99 -30.99 2.44
164