13.

(gm . .

..¢

 

Va- '5 1: GP "WW-M5”:

«.

"z, zyn u

w.
f.

.,A ”“m '97P"

.3

h‘

flu,»

‘2 -‘ . a
l’ffigfilhfi
x; ‘ o

1
‘

r 3"
£26

 

 

 

 

 

' V .
.- w
”5,335“- A
ll”

 

.'"'-~~db

(1:553
549/7'145

 

_ |

UBRAHYH
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

ECOLOGICAL ASSESSMENT OF HUMAN DISTURBANCE TO

STREAM ECOSYSTEMS USING ALGAE

presented by

Yi-Kuang Wang

has been accepted towards fulfillment
of the requirements for the '

degree in Zoology/EEBB Program

 

7 or“ 82%

 

 

Major Professor’s Signature

l4éwdi¢léwm3

Date

 

MSU is an Affirmative Action/Equal Opportunity Institution

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

gm 1 (l 2010

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRC/DataDue.p65-p.15

ECOLOGICAL ASSESSMENT OF HUMAN DISTURBANCE TO STREAM
ECOSYSTEMS USING ALGAE
By

Yi-Kuang Wang

A DISSERTATION
Submitted to Michigan State University
In partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Zoology and the
Ecology, Evolutionary Biology, and Behavior Program

2003

ABSTRACT

ECOLOGICAL ASSESSMENT OF HUMAN DISTURBANCE TO STREAM
ECOSYSTEMS USING ALGAE

By
Yi-Kuang Wang

Algal attributes responded to land use and environmental variables related to low-
level human disturbance in streams of glaciated northern Michigan, USA. Diatom species
composition and dominant species changed with a human disturbance gradient (HDG),
which was defined by covariation among nutrient concentrations and land uses. The
percentage of native diatom species was negatively correlated with the HDG. Simpson
dominance index and diversity indices were not correlated with the HDG. Average
diatom biovolume was positively correlated with the HDG. Relative abundance of
species sensitive to trophic status decreased and % trophic tolerant species increased with
the HDG, Cl', and conductivity, but did not correlate with nutrient concentrations. Thus
changes in diatom assemblages due to nutrient enrichment could be detected, but not
related to the narrow range in nutrient concentrations.

Of the three major benthic algal habitats in streams, diatoms on plants were the
most sensitive assemblages to the human disturbance. Epiphytes also had the smallest
beta diversity and fastest species turnover along an HDG, while epipelon and epilithon
had about the same beta diversity and species turnover rate. Epiphytes had the most
variance in species composition and species-environment relations explained by CCA.
Epiphyton was more sensitive than epilithon and epipelon based on changes in trophic

status indicators along the HDG. Species composition of epipelon was distinctly different

from the epilithon and epiphyton. Heterogeneity in species composition among habitats
did not decrease along the HDG.

A regional pattern in nutrient limitation of periphyton was not observed with
experiments in 16 streams despite very low P concentrations. However, species
composition did respond to experimental nutrient enrichment. Further examination
showed that nutrient enrichments did not change species composition in experiments in
the same way that composition varied among streams with different nutrient
concentrations. Diatom assemblages from treatment and control clay pots at the same site
were more similar than assemblages from the same treatments between Sites. Therefore,
extrapolation of experimental results from small-scale experiments to large regional
scales may be constrained by the local species pools.

A diatom-based Index of Biotic Integrity (IBI) for the Interior Plateau Ecoregion
(IPE) of the US was developed using data collected by the Kentucky Division of Water.
Diatom attributes at reference and impacted sites were selected for the IBI based on
significant differences between site groups, high separation power, and low CV. Seven
metrics were selected from a list of 59 attributes to form a diatom IBI, which successfully
distinguished reference and impaired sites in both development and test datasets. The IBI
scores were significantly related to measures of human disturbance in watersheds. This

IBI had more separation power than the other diatom IBIS.

ACKNOWLEDGMENTS

I thank my committee members, Dr. R. Jan Stevenson, Dr. Thomas M. Burton, Dr.
Stephen K. Hamilton, and Dr. Bryan C. Pijanowski, for their help on various aspects. Dr.
Brian Pijanowski has kindly provided me help on GIS related questions. Drs. Dave Long
and Brian Pijanowski have offered assistance on the Grand Traverse Bay Watershed
project. I can’t complete this dissertation without my major Professor, Dr. R. Jan
Stevenson. Jan shares his knowledge, stimulates discussion and thinking, and especially
teaches me writing. I have great appreciation to Jan’s great support throughout my Ph.D.
years.

I like to thank members in my lab, Julie Heinlein, Dr. Vannessa Lougheed, Kalina
Manoylova, Chris Parker, Dr. Steve Rier, Scott Rollins, and Lei Zheng, for providing
comments on my dissertation. Special thanks to Kalina Manoylova, she taught me algal
taxonomy and answered my questions with kids on one hand and a microscope on the
other hand.

Lastly, without the support and encouragement of my parents through the long
rough years, the completion of this dissertation is totally impossible. I dedicate this
dissertation to my parents, Shiu-Yin Chin and Shin-Liang Wang. They have always

believed in me and provided warmth to my heart.

iv

TABLE OF CONTENTS

LIST OF TABLES ............................................................................... vii
LIST OF FIGURES ................................................................................ ix
CHAPTER 1. INTRODUCTION .................................................................. 1

CHAPTER 2. ECOLOGICAL RESPONSE OF STREAM ALGAL COMMUNITIES TO
HUMAN DISTURBANCE IN THE GRAND TRAVERSE BAY WATERSHED..... 7

Abstract ........................................................................................ 7
Introduction .................................................................................... 8
Materials and Methods ....................................................................... 11
Results .......................................................................................... 16
Discussion ..................................................................................... 20

CHAPTER 3. RESPONSES AND SIMILARITIES OF DIATOM ASSEMBLAGES
FROM MULTIPLE HABITATS ALONG ENVIRONMENTAL GRADIENTS IN

MICHIGAN STREAMS ............................................................................ 33
Abstract ........................................................................................ 33
Introduction .................................................................................... 34
Materials and Methods ....................................................................... 36
Results .......................................................................................... 42
Discussion ..................................................................................... 55

CHAPTER 4. EFFECTS OF SMALL-SCALE NUTRIENT ENRICHMENT ON
ALGAL COMMUNITY COMPOSITION IN THE GRAND TRAVERSE BAY

WATERSHED ....................................................................................... 61
Abstract ........................................................................................ 61
Introduction .................................................................................... 62
Materials and Methods ....................................................................... 64
Results .......................................................................................... 69
Discussion ..................................................................................... 79

CHAPTER 5. DEVELOPMENT AND EVALUATION OF A DIATOM-BASED
INDEX OF BIOTIC INTEGRITY FORTHE INTERIOR PLATEAU ECOREGION,

KENTUCKY ........................................................................................ 87
Abstract ...................................................................................... 87
Introduction .................................................................................. 88
Materials and Methods .................................................................... 90
Results ........................................................................................ 102
Discussion .................................................................................. 1 13

CHAPTER 6. CONCLUSION .................................................................. 120

APPENDICES ..................................................................................... 1 24

BIBLIOGRAPHY .............................................................................. 131

vi

LIST OF TABLES

Table 2.1. Mean, SD, and range of land use and environmental variables. N .D. is short for
non-detectable ...................................................................................... 18

Table 2.2. The correlations between land use types and environmental variables. All
selected correlations have a p < 0.10. r is the Pearson correlation coefficient. All
correlations Shown have p < 0.10. * indicates p <0.05, ** for p < 0.01,

*** for p < 0.001 ................................................................................... 19

Table 2.3. RDA of environmental variables with land use and the correlation between
RDA axes and relevant variables. * indicates statistical significance, p < 0.05 ...... 20

Table 2.4. The correlations between diversity measures, evenness, dominance, average
Size, native species, indicator guilds with RDA 1St axis scores, land use and
environmental variables. All correlation shown have p < 0.10. * indicates p <0.05, ** for

p < 0.01, *** for p < 0.001 ................................................................... 24
Table 2.4. Continued ............................................................................... 25
Table 3.1. The minimum, maximum, 25‘“, median, and 75th percentile of environmental

variables ............................................................................................. 43
Table 3.2. Indicator species, indicative habitats, and their indicator values (IV) .......... 45
Table 3.2. Continued ............................................................................... 46

Table 3.3. Variances explained by CCA axes in different habitats. Fifteen sites were
randomly selected from epipelon and epilithon 1,000 times for CCA runs. Cum is the
short for cumulative ................................................................................ 49

Table 3.4. Correlations of PCA axes with environmental variables and % variance
explained for PCA axes ........................................................................... 51

Table 3.5. Regression statistics for similarity changes along the MEG. Sample sizes (n)
were the numbers of possible pair-wise comparison. The MEG axis represents an
environmental gradient from high to low nutrients, chloride, and conductivity. Bray-
Curtis (BC) and is the similarity measure ...................................................... 53

Table 4.1. The summary statistics for land use, environmental variables, and nutrient
ratios. ND. is the short for non-detectable ..................................................... 71

Table 4.2. Correlation between CA axes of clay pot diatom species and environmental

variables. CA1 is the short for CA lst axis; CA2 for CA 2nd axis. * indicates 0.01 < p <
0.05; ** indicates 0.001 < p < 0.01 .............................................................. 75

vii

Table 4.3. The comparison of the difference between a treatment and the control to 0 on 2
DCA axes with t test ................................................................................ 76
Table 4.4. The responses of species to nutrient enrichment treatments comparing to
control clay pot. T-tests were performed on the relative abundance difference between
treatments and the control. * indicates p < 0.05 and ** represents p < 0.01 .............. 77

Table 4.4. Continued .............................................................................. 78

Table 5.1. Algal population and community attributes, their descriptions, and their

expected response to impairment. + represents attribute increasing with impairment, -
represents attribute decreasing with impairment, and V represents variable ............ 94

Table 5.1. Continued ........................................................................... 95

Table 5.2. Diatom genera growth form and motility. P = prostrate, E = erect, S = stalked,
U = unattached, V = variable, M = highly motile diatoms .................................. 97

Table 5.3. Selected algal metrics and their separation power, coefficient of variation,
response to impairment, and reasons for exclusion. + represents metric increasing with
impairment caused by human disturbance, - represents metric decreasing with
impairment ......................................................................................... 103

Table 5.4. Scoring range and 90‘h percentile and minimum values for 10 selected metrics
of IBI ............................................................................................... 105

Table 5.5. Pearson correlation coefficients between IBI metrics and % forest cover in
watersheds. * denotes statistical significance, (1 < 0.05. Ach. is short for Achnanthes; Nav.
for Navicula ..................................................................................... 110

Table 5.6. Box separation ratios of developed IBIS, Bahls IBI, KY diatom IBI, and Hill’s
IBI from test dataset ............................................................................... 114

viii

LIST OF FIGURES

Figure 2.1 The map of sampling sites in the Grand Traverse Bay Watersheds and their
locations in Michigan .............................................................................. 13

Figure 2.2 The relationships between dominant species and the HDG. A. Cymbella
delicatula, r = 0.95, p = 0.003; B. Gomphonema rmmilum, r = 0.65, p = 0.049 ........... 26

Figure 2.3. Differential responses of % trophic sensitive and native species and % native
species to the major environmental gradient. The lower line is % trophic sensitive and
native species, % native species is the top line. Dashed lines represent the corresponding
environmental gradient values when attribute values are at 50% ............................ 32

Figure 3.1. Locations of the sampling sites in Michigan ...................................... 37

Figure 3.2. CA plot for all sites from 3 habitats. Arepresents epipelic samples; I for
epilithic samples; * for epiphyte samples ...................................................... 44

Figure 3.3. Box plots for comparing CCA results from 3 habitats. A. % species variance
explained; B. % species-environment variance explained. 0 represents outliers. . 48

Figure 3.4. A plot of PCA axis 1 vs. axis 2. 0 represents for epiphytic sites, 0 is for non-
epiphytic sites ...................................................................................... 52

Figure 3.5. The comparison of epiphyton’s beta diversity and 1,000 re-sampled beta
diversity of epipelon and epilithon with box plots. A. pair-wise beta diversity; B. beta
diversity calculated from average similarity with reference approach. 0 represents
extreme values, and O is for outliers .......................................................... 54

Figure 3.6. A comparison of the slopes of 3 habitat assemblages with box plots. The
slopes were derived from the regression between MAIA trophic index and the MEG.
Epilithon and epipelon were randomly sampled for 15 sites 1,000 times. 0 represents for
outliers ............................................................................................ 56

Figure 4.1. Site locations in the Grand Traverse Bay Watersheds, Michigan .......... 66

Figure 4.2. Box plots of nutrient limitation indices. 0 represents outliers, while 0
represents extreme values ...................................................................... 72

Figure 4.3. A plot of the first 2 DCA axes scores of all clay pots. Symbols represent
different sites ................................................................................... 74

Figure 4.4. Relationship between environmental optima and Species responses to
treatments. The lines indicate 1 to 1 ratio .................................................... 80

ix

Figure 4.5. The relationship between ranks of species Cl optima from a regional survey
with A. ranks of species response to the NP treatment, B. ranks of Species response to the
NPSi treatment. The lines indicate 1 to 1 ratio ............................................... 81

Figure 4.6. The relationship between van Dam’s category and A. ranks of species in NP
treatments, and B. ranks of Species in NPSi treatments .................................... 82

Figure 5.1. Study area and locations of sampling sites in Kentucky (USA) and the Interior
Plateau Ecoregion ............................................................................... 91

Figure 5.2. Box plots of diatom metrics for reference (Ref) and impaired (Imp) Sites of
development dataset. Boxes are interquartile ranges (between 25th percentile and 75th
percentile); middle lines are medians; and whiskers are 1.5 interquartile ranges beyond
the boxes ......................................................................................... 106

Figure 5.3. Box plots of total IBI scores of reference sites versus impaired sites from
development dataset. A. 1-3-5 scaling system. B. 0-10 scaling system ............... 107

Figure 5.4. Relation between index values of biotic integrity from the KDOW data set
and % forest cover in watersheds (adjusted r2 = 0.616, p < 0.001). Percent forest was
arcsine squared-root transformed .............................................................. 109

Figure 5.5. Box plots of diatom metrics for reference (Ref) and impaired (Imp) sites of
test dataset. Boxes are interquartile ranges (between 25th percentile and 75th percentile);
middle lines are medians. o = outliers ....................................................... 111

Figure 5.6. Box plots of total IBI scores of reference sites versus impaired sites from test
dataset. A. 1-3-5 scaling system. B. 0-10 scaling system ................................... 112

CHAPTER 1
INTRODUCTION

Human activities in watersheds can alter factors directly controlling the structure and
function of stream ecosystems (Karr and Dudley 1981, Malmqvist and Rundle 2002). The
growing human population and associated water demands accelerate the degradation of
stream ecosystems (Baron et a1. 2002, Richter et a1. 2003). Both urban and agricultural
land uses contribute a large amount of nutrients to streams (Paul and Meyer 2001, Smith
2003), subsequently causing algal blooms in rivers, lakes, reservoirs, and even estuaries
(Mitsch et al. 2001, Smith 2003). Therefore, an ecologically sound and effective
assessment system is needed to evaluate the status and change of stream ecosystems,
diagnose stressors, and offer management options to decision makers (Wang and
Stevenson 2002). Bioassessment can also provide early warning signs before catastrophic
events occur (Cairns and McCormick 1992).

Human induced stresses often change biotic community structure in different ways.
Stresses can reverse the autogenic community development (Odum 1985) and induce a
series of stress syndromes on ecosystems (Rapport et al. 1985). Decrease in organism
size, decrease in species diversity, increase in species dominance, and species prone to
diseases are often observed in stressed ecosystems (Odum 1985, Rapport et a1. 1985).
Stresses also favor r—strategists (Odum 1985), which often are opportunistic species
(Rapport et a1. 1985, Schindler 1987). Changes in indicator species (tolerant species
increase and sensitive species decrease) (Sheehan 1984) and shifts in community
structure (Rapport and Whitford 1999, Gunderson 2001) are often observed in stressed

ecosystems. From whole lake experiments, Schindler (1987) found that taxonomic

changes are more important than diversity indices, short lifecycle, low dispersal species
are most affected, and keystone predators are often lost due to stresses. Loss of native
species, invasion of exotic Species (Rapport et al. 1985, Rapport and Whitford 1999), and
increases in the amplitude of population fluctuations (Rapport 1985) are getting
increasing attention in stressed ecosystems.

Biological assessment is actively adopted by agencies around the world for several
reasons. It integrates physical, chemical, and biological environments. Organisms
incorporate a longer duration of environmental information than a one-time snap-shot
water sampling. Biological integrity and biodiversity, important components of
bioassessment, are important social values. Bioassessment can detect intermittent sources
of pollution and non-point source pollution (Yoder 1995).

Algae have a high intrinsic value in aquatic ecosystems because they are primary
sources of energy for grazers (Gullberg et a1. 1997, Finlay 2001), play an important role
in nutrient cycling (Dodds et a1. 2000), and facilitate the growth of microbes (Petit et a1.
1999, Romani and Sabater 1999). On the other hand, algal blooms can decrease dissolved
oxygen levels, create taste and odor problems, increase difficulty for drinking water
processing, produce toxins, deter recreation, and decrease aesthetic values. Therefore,
algae have valued ecological attributes from social-economic perspectives and need to be
evaluated for bioassessment and management.

From the perspective of water quality assessment, algae have been used extensively to
monitor water pollution for over a century due to the following reasons: 1) algae are

highly sensitive to pollution (Kolkwitz and Marsson 1908, Sladeckova 1962, Patrick

1973, Lowe 1974, Descy 1976); 2) they have rapid reproduction rates and short life

cycles, therefore they respond quickly to perturbation (Stevenson and Lowe 1986); 3)
they are easily sampled and quantified (Watanabe et a1. 1990); 4) algae are well
documented in the taxonomic literature and are relatively easy to identify to species
(Stevenson and Pan 1999); 5) a large number of species are sensitive to various
environmental factors (Lowe 1974, Round 1981, van Dam et a1. 1994); 6) high diversity
and species-specific sensitivity to their environments confer diatoms with the power to
infer environmental conditions; and 7) sub-cosmopolitan characteristics of algae enable
use of indices in many regions (Kelly et a1. 1998).

Algal indices can be classified into 4 general types. Two types of indices are based on
environmental optima and tolerances of species and help diagnose environmental
stressors. The first type uses categorical characterizations of species autecology (often 1-
5 or 1-8 ordinal scales) and relative abundance of species to infer the physical and
chemical status of ecosystems (e. g. pH, organic enrichment, and nutrients) (Lowe 1974,
Watanabe 1990, Prygiel and Coste 1993, van Dam et al. 1994, Kelly and Whitton 1995).
The second approach, weighted average models, uses accurate characterization of the
optimal environmental conditions of species based on a continuous quantitative
environmental scale (ter Braak and van Dam 1989, Cumming and Smol 1993, Pan et al.
1996). The other 2 types of indices provide assessments of biotic integrity, rather than
physical and chemical stressors. Diversity indices, the third approach, have been widely
used to assess the number of species and evenness of their distributions (Archibald 1972,
Patrick 1973, Stevenson 1984). The fourth approach is the similarity in species
composition between assessed and designated reference sites and is another measure of

biotic integrity (Stevenson and Lowe 1986, Charles 1996, Stevenson and Pan 1999).

Algae indices of biotic integrity (IBIS) are relatively new and only a few examples
have been developed (Bahls 1993, Kentucky Division of Water 1993, Hill et al. 2000).
The multimetric approach using invertebrates and fish has been used widely for assessing
and communicating biotic integrity of streams across the US. (Barbour et al. 1996,
Barbour et al. 1999). The IBI emphasizes biotic integrity, and uses diversity and
similarity indices as well as functional guild metrics and higher taxonomic metrics to
form a multimetric scoring index. The IBI scores give a quantitative sense of the
ecological health of the biological community and can be statistically effective for
evaluating human disturbance (Fore et al. 1994).

Algal biodiversity in streams includes common and unique flora from different
habitats. The affinity of algal flora has been documented: for example Achnanthes
lanceolata, Achnanthes lauenburgiana, Fragilaria pin—neg, and Navicula ign_og var.
m are episammic species (Miller et al. 1987) and species in the genera Asterionella.
Melosirg, CLclotelg, and Stephanodiscus are planktonic species (Kingston et al. 1983).
Epilithon is usually assessed for ecological research and bioassessment. The differences
in algae among habitats and applicability of algae in different habitats for bioassessment
have seldom been examined.

Though the emphasis has been on community structure in algal bioassessment, some
functional assessment methods have been adopted, such as the alkaline phosphatase (Hill
et al. 2000), primary production (Niyogi et al. 2002), and nutrient limitation (Wold and
Hersey 1999). Nutrient limitation assays are often used to assess whether primary
production is limited by P, N, or both (Fairchild et al. 1985, Burton et al. 1991). Nitrogen

limitation is common in the Western United States (Grimm and Fisher 1986, Hill and

Knight 1988), while streams are often P limited in Michigan (Pringle and Bowers 1984,
Fajrchild et a1. 1985, Burton et al. 1991). Regional patterns in nutrient limitation may be
controlled by the biogeochemistry. Changes in the nutrient limitation pattern may affect
biotic integrity. Nutrient limitation assays often use nutrient diffusing substrates (NDS),
such as clay pots (Fairchild and Lowe 1985) and filter membranes on test tubes (Wold
and Hersey 1999).

Objectives

Many bioassessment studies have focused on highly impacted areas. In chapter 2 to
4, the Grand Traverse Bay Watershed (GTBW), a relatively low impacted area, was
studied. The GTBW is approximately 2600 km2 and is located in the northwestern part of
the lower Michigan peninsula. The Boardman River drains the majority of the GTBW.
The Pere Marquette National Forest is located in the upper- and mid-region of the
Boardman River watershed and contains a mixed deciduous-coniferous forest. Traverse
City, with a summer recreation population of 500,000, is the largest city in the GTBW.
Land use/cover in the GTBW is composed of 49% forest, 20% agriculture, 15%
herbaceous/shrub/grasslands, 10% water, and 6% urban. Despite a high degree of
development in some catchments, land use varies from urbanized and agricultural areas
to completely natural areas.

In Chapter 2, a set of hypotheses relating community structure to human
disturbance was tested with data from a field survey. The sensitivities of ecological
attributes were compared in this low level of human disturbance. Ecological attributes
that respond to low human disturbance have the potential to be used in bioassessment and

as early warning Signs of threats to stream ecosystem health.

In Chapter 3, the similarities among assemblages from multiple habitats were
evaluated and characteristic species for each habitat were derived. The sensitivities of
diatom assemblages in different habitats were compared with 3 methods. The
homogenization of beta diversity among habitats along a disturbance gradient was also
evaluated.

Chapter 4 explored the regional nutrient limitation pattern in the GTBW with
nutrient diffusing substrates (NDS). To evaluate if short-term small-scale experiments
can reproduce community structural responses as observed among assemblages in
nutrient enriched streams, similarities of assemblages on clay pots were assessed. The
responses of species on nutrient enriched clay pots were compared with their responses to
nutrient differences among streams in the region.

In Chapter 5, an index of biotic integrity (IBI) was developed for the Interior
Plateau Ecoregion in Kentucky. Attributes tested were selected from those commonly
used in bioassessment. The developed IBI was evaluated with land use information,

tested with another dataset, and compared to existing IBIS.

CHAPTER 2
ECOLOGICAL RESPONSE OF STREAM ALGAL COMMUNITIES TO HUMAN
DISTURBANCE IN THE GRAND TRAVERSE BAY WATERSHEDS
Abstract

Human disturbance in watersheds influences Stream ecosystems through alterations
of hydrological pathways, water chemistry, habitats, and food sources. A set of
hypotheses on ecological responses to human disturbance were tested with algal
community data. The following hypotheses were tested: 1) species composition changes
along a human disturbance gradient (HDG); 2) dominant species change along the HDG;
3) species diversity decreases along the HDG; 4) the size of organisms decreases along
the HDG; 5) native species decrease along the HDG; and 6) indicator species increase
along the HDG. Thirty-one sites were sampled in the Grand Traverse Bay Watershed
(GTBW) in Fall 2000. The GTBW, located in the Northwestern part of the lower
Michigan, is relatively low in development compared to southern lower Michigan. Land
use types were correlated with environmental variables. The HDG was formulated as the
lst axis of redundancy analysis and was positively correlated with % urban and %
agriculture land uses.

Algal species composition and dominant species changed with the HDG. Diatoms
were the dominant algal group. Diatom dominance index and diversity indices were not
correlated with the HDG. Average diatom biovolume was positively correlated with the
HDG. Relative abundance of trophic sensitive species significantly decreased with the
HDG, while % trophic tolerant species significantly increased with the HDG. Both
trophic sensitive and tolerant species were significantly correlated with HDG, Cl', and

conductivity, but not with nutrient levels. Nutrient levels may be sequestered as resources

by the biota and are often associated with elevated Cl' and conductivity. Hence, diatom
species were a better indicator for human disturbance than nutrient concentrations in
water samples. The percentages of native species were negatively correlated with the
HDG, while % native and trophic sensitive species had a higher correlation with the
HDG than % native species. Overall, these algal attributes responded to land use and
environmental variables under low-level human disturbance, except for species diversity.
Key words: watersheds, human disturbance, species composition change, dominant
species, average diatom size, species diversity, indicator species, sensitive species,

tolerant species, native species, exotic species.

Introduction

The connections between land use/cover and stream ecosystems have been
observed in various watershed experiments (Likens and Bormann 1974, Swift and Swank
1981, Gregory et al. 1991). From field surveys, land use patterns were also strongly
correlated with stream ecosystem structure and function (Hynes 1975, Schlosser 1995,
Richards et al. 1996, Allen et al. 1997). Human activities in the watersheds can change
stream ecosystem structure and function through alteration of nutrients, hydrological
regimes, water quality, energy sources, habitat structure, and biotic interactions (Karr
1991).

The relationship between nutrients and algae has been widely studied (Vallentyne
1972, Welch et al. 1988, Chetelat et al. 1999, Biggs 2000). The linkage between

watershed disturbance and algae has received less attention until recently. Watershed

disturbance often induces a change in algal species composition (Leland 1995, Leland
and Porter 2000, Munn et al. 2002). The responses of algal assemblages to environmental
variables are often assessed by ordination and weighted average indicators of
environmental conditions. Few algal studies related the effects of human disturbance to
valued ecological attributes. A set of hypothesis, referred to as stress syndromes (Rapport
et al. 1985) and retrogression characteristics (Odum 1985), have been used to predict the
effects of human disturbance on ecosystems (Schindler 1987, Blair 1996, Havens and
Carlson 1998). These hypotheses can serve as a unifying ecological framework with
which responses of valued attributes of algae can be related to human disturbance.

Six hypotheses or trends of community structural responses to human disturbance
were tested in this study. These hypotheses have great literature support and sound
ecological bases. They have been observed in different types of ecosystems and under
different kinds of stressors and also have both ecological and societal values.

The first hypothesis is that species composition changes along the human
disturbance gradient (HDG) (Schindler 1987, Rapport and Whitford 1999, Gunderson
2001). The second hypothesis is that dominant species change along the disturbance
gradient (Sheehan 1984). The first hypothesis differs from the second hypothesis by the
inclusion of rare species. The basis of these two hypotheses is a combination of
environmental requirements of species (niche) (Hutchinson 1957) and differential
competitive ability along environmental or resource gradients (Southwood 1977).
Changes in species composition along environmental gradients have been observed since
Whittaker (1956) and are recognized as a major theme of ecological study (Austin 1985,

ter Braak and Prentice 1988).

The third hypothesis is that species diversity decreases with increasing human
disturbance, as dominance by one or a few species increases (Odum 1985). Species
diversity is a valuable ecological attribute and conservation target (Rozdilsky et al. 2001).
Species diversity can have a hump-shaped relationship along the productivity gradient
(Mittelbach et al. 2001). Non-resource toxic stressors often have adverse effects on
diversity. However, the mix of resources and toxic stressors may have complex effects on
diversity. The fourth hypothesis is that organism size decreases with increasing human
disturbance gradients (Odum 1985). Human disturbance has been observed to eliminate
large species and especially predators (Havens 1994).

Fifth, indicator Species have been observed to increase with the increase of relevant
environmental gradients (Sheehan 1984). The idea that the presence of certain species
can indicate specific environmental conditions has been widely recognized (Cairns 1974).
Indicator species can be indicative of the status of other species, key habitats, or the
impacts of a stressor (Dale et al. 2000). The environmental preference of species also has
its foundation in niche theory (Hutchinson 1957), and has been used in saprobien
indicator systems for almost a century and toxicity tests for environmental assessment.
Indicator species are often categorized as sensitive or tolerant species to a specific
stressor, for example acid tolerant or sensitive species in the plankton (Havens and
Carlson 1998).

The sixth hypothesis is that human disturbance can cause loss of native species
(Schindler 1987, Rapport et al. 1985, Blair 1996, Rapport and Whitford 1999).
Anthropogenically induced changes in environmental conditions can create competitive

advantages for non-indigenous species over native species (Byers 2002). Invasive species

10

have caused serious economic and ecological damage (Leung et al. 2002, Perrings et al.
2002). The relative abundances of native species are often reduced after the invasion of
alien species, which is often facilitated by human disturbance. Therefore, invasive species
increase their relative abundances along the disturbance gradient.

In this paper, I studied algal community and evaluate the responses of the Six
ecological attributes listed above to human disturbance. The algal data were collected
from a field survey. These attributes have been observed to change under human
disturbance, but have seldom been studied together at one time for stream ecosystems,

especially in a relatively pristine area such as GTBW.

Materials and Methods

Study Area

Grand Traverse Bay Watershed (GTBW) has an area approximately 2,600 km2 and
is located in the northwestern part of the lower peninsula of Michigan, US. It has one of
the fastest growing populations in the Midwest United States (Grand Traverse County
1996). Traverse City, with a summer recreation population of 500,000, is the largest city
in GTBW. Land use/cover in GTBW is composed of 49% forest, 20% agricultural, 15%
herbaceous/shrub/grasslands, 10% water, and 6% urban. Despite high development in
some catchments, overall the region is mildly developed. The Boardman River drains the
majority of GTBW with the Pere Marquette National Forest located in the upper- and
mid-region of the Boardman Watershed. Thirty-one sites were selected across a gradient

of urban and agriculture land use intensities (Figure 2.1).

11

Sampling and Sample Processing

Sampling sites were visited between 13 and 20 October 2000, when hydrology is
seasonally relatively stable. Biological, physical, and chemical samples were collected.
Algae were scraped from 6 rocks with a toothbrush for each site and stored on ice for
further lab analysis. The scraped areas on rocks were measured with a ruler. Canopy
cover was estimated with a densiometer. Flow regimes were measured with an electronic
portable flowmeter (Flo-mate model 2000, Marsh-McBirney Inc.). Conductivity, water
temperature, and pH were also measured in the field with a multipurpose probe (Denver
Instrument).

Algal samples were subsampled in the lab for chlorophyll analysis and algae
enumeration. Diatoms were digested with the boiling nitric acid method and catalyzed
with potassium dichromate. Soft algae were preserved with M3 solution (APHA 1998).

Diatoms were identified based on Kramrner and Lang-Bertalot (1986, 1988, 1991a,
1991b); Patrick and Reimer (1966, 1975) were used as supplements. Prescott (1951) was
used as the primary reference for soft algae identification; Taft and Taft (1990) was used
as a supplement. Both diatom and soft algae were identified to the lowest level possible
with a Leica microscope under 1000X and 400x magnification respectively. Biovolumes
of soft algae were measured from samples. The biovolumes of common diatom species
were calculated from size measurements based on designated shapes (Charles et al. 2002),
while the biovolumes of rare species came from past lab data or average sizes from
taxonomic references.

Water chemistry analysis was done with a Technicon Auto-analyzer III following

standard methods (APHA 1998). NH3-N was analyzed with an automated phenate

12

 

 

e Shes

Streams
f: g Wetlands
- Lakes
N

 

 

 

 

20 0 20 40 Ulla.

Figure 2.1. The map of sampling sites in the Grand Traverse Bay Watersheds and their

locations in Michigan.

13

method, NO3-N with an automated Cd-reduction method, and total N (TN) was digested
with persulfate before using the nitrate method. Soluble reactive phosphorus (SRP) was
measured with the automated ascorbic acid method, while total P (TP) was measured
using the SRP method after persulfate digestion and the sample was not filtered. Cl' was
measured by the automated ferricyanide method, while SiOz-Si was measured by an
automated method for molybdate-reactive silica. Chlorophyll a (Chl a) was extracted by
dissolving the membrane filter in 90% aqueous acetone and measured following standard
methods (APHA 1998) with a spectrophotometer (Spectro Instrument). Ch] 3
concentration was calculated as mg/cmz.

Land use/cover information was obtained from the 1980 Michigan Resource
Information System (MIRIS) with Anderson land use/land cover level 1 classification. A
watershed was delineated as the sourceshed above the sampling point in the catchment.

Digital elevation modeling with ArcInfo software was used for watershed delineation.

A major environmental gradient (MEG) was obtained from redundancy analysis
(RDA) of environmental variables with land use as constraining variables to maintain the
independence of the MEG from biological data. RDA is derived from multiple linear
regression, followed by principal component analysis (Legendre and Legendre 1998).
The MEG represented the major human disturbance gradient (HDG) in RDA and was
used for future statistical analysis in this paper.

Data Analysis

Shifts in species composition along the HDG were evaluated by similarity to

reference sites. The top 25% of low-developed sites along the HDG were selected as

reference Sites. The HDG was formulated from land use and environmental variables, and

14

no species data were involved to avoid circularity. The average Bray-Curtis similarity
with reference sites was correlated with the HDG to test for significant relationship.

The change of dominant species along the HDG was evaluated with the top 2
species from each site. The correlation between relative abundances of top 2 species and
the HDG was evaluated with Pearson correlation analysis. Dominance pattern was also
evaluated by the correlation between Simpson’s dominance index (Simpson 1949). The
correlations between the HDG and relative biovolumes of major algal groups, such as %
filamentous green and % blue green algae, were also evaluated with Pearson correlation
analysis.

The relationships between organism size and species diversity measures with the
HDG were assessed by Pearson correlation analysis. Diatom species diversity was
separated into species richness and evenness measures. Species richness was defined as
the average number of species per 100 diatom count, to offset the phenomenon of
increasing number of species with increasing diatom counted. Evenness was calculated
using Pielou’s evenness index (Pielou 1975).

Sensitive species were defined as species that decreased in relative abundances
with increasing disturbance; on the contrary, tolerant species increased in their
abundances with increasing disturbance. Trophic sensitive and tolerant species were
taken from 3 compiled lists: Bahls (1993), van Dam et al. (1994), and MAIA trophic
index (Stevenson’s unpublished US EPA Mid-Atlantic region diatom trophic index).
Species with score 3 in Bahls’s list, 1 in van Dam’s trophic and saprobity lists, and l and
2 (< 10 ug/L TP) in Stevenson’s list were selected as trophic sensitive species. Species

with a score of l on Bahls’s list, 5 and 6 in van Dam’s trophic list (eutrophic species), 4

15

and 5 in van Dam’s saprobity list (poly-saprobous species), and 6, 7, and 8 (> 25 ug/L TP)
in Stevenson’s list were selected as trophic tolerant species. Species listed in both
sensitive and tolerant lists were thrown out. The tendencies of relative abundance of
trophic sensitive and tolerant species with the HDG were evaluated with Pearson
correlation analysis. Nitrogen, pH, and salinity indicator species were also tested and
were derived from van Dam’s N, pH, and salinity indices respectively. Species in the
lower 2 categories in these van Dam’s indices were selected as sensitive species, while
those in the top 2 categories were considered as tolerant species. Percent nitrogen, pH,
and salinity sensitive and tolerant species and % motile diatom genera (Wang and
Stevenson unpublished manuscript) were evaluated along the HDG with Pearson
correlation analysis.

Native species are those that live at a site prior to human disturbance. Non-
indigenous species are invaders from outside the region. Sites in top 25th percentile of
the first RDA site scores were selected as reference sites. Species at reference sites were
designated as native Species. Species on both native and sensitive species lists were
defined as sensitive and native species. Percent native species at reference sites were
evaluated with native species in other reference sites, and the same procedure was used
for % sensitive and native species. Relative abundance of non-indigenous species was
calculated as 1 minus the relative abundance of native species. The relationships between
relative abundances of native species, non-indigenous species, and sensitive and native

species with the HDG were evaluated with Pearson correlation analysis.

16

Results

Physical and chemical characteristics were quite variable among sites (Table 2.1).
Nonetheless, these variables were generally indicative of pristine to moderately impacted
sites. The watershed is largely dominated by forested land (49%) with lesser amounts of
agriculture (22%) and urban (5%) development. Water temperature ranged from 7.9 to
16.1 °C, while pH levels were mildly basic. NO3-N ranged from non-detectable to 2.01
mg/L, TN ranged from 0.13 to 2.7 mg/L with 0.65 mg/L average, which indicated mild
enrichment in some streams. TP and SRP levels were relatively low (< 5 jug/L). Cl', an
indicator of human presence in watersheds, ranged from 0.5 to 21.4 mg/L with a mean
7.9 mg/L. Conductivity ranged from 253 to 605 with a mean of 407 uS/cm. Stream order
spanned from lst to 4th order.

Urban land use was significantly correlated with conductivity and Cl‘ (Table 2.2).
Agricultural land use was positively correlated with TP, conductivity, water temperature,
and Cl‘, and negatively correlated with % forest, stream order and pH. Wetland cover in
the watershed was positively correlated with SRP and NH3-N. Forest cover was
positively correlated with stream order and pH, but negatively correlated with %
agriculture, % urban, conductivity, TP, TN, Cl‘, and water temperature.

The first 2 RDA axes explained 54.6 and 0.2% of environmental variable variance
respectively, so only the first RDA axis was used as the HDG (Table 2.3). The lst RDA
axis was positively correlated with % agriculture, % urban, TP, Si, Cl', and conductivity,
and negatively correlated with % forest and pH. The HDG served as a watershed

disturbance gradient from high forest cover to relatively high development and nutrient

17

Table 2.1. Mean, SD, and range of land use and environmental variables. ND. is the
short for non-detectable.

 

 

Mean SD Minimum Maximum

% Urban 6.97 9.95 0 43
% Agriculture 22.19 19.97 0 67
% Forest 49.00 26.78 9 93
% Wetland 4.26 5.75 0 27
Order 1.81 0.91 1 4
Canopy 38.08 28.62 0.5 95.5
Water Temp. (°C) 10.5 2.0 7.9 16.1
Conductivity (uS/cm) 407.06 96.17 253 605
pH 8.15 0.13 7.8 8.4
N03 (mg/L) 0.45 0.43 N.D. 2.01
NH4-H (mg/L) 0.01 0.02 N.D. 0.11
Total N (mg/L) 0.65 0.56 0.13 2.70
Soluble Reactive P (pg/L) 1.53 0.67 0.1 3.3
Total P (pg/L) 1.75 1.04 0.1 4.7
Si (mg SiOzlL) 8.89 2.12 3.2 15.7
C1 (mg/L) 7.87 5.43 0.5 21.4

 

18

 

de- somd- *Iud *Nvd- *mvd- *sovd *vaWd- *xuwmod- *Lcwmnd- *xxond- $9.3“— Awe

.m Z 2:3. In 95% .535 55.5 axe SEC 5 m :38. $383.50 Eszsotw< exc
mmd *mwd- :omd viiomd macs—83 eke

95a Leg a5. 32 mam
mmd *Nvd *3...C “553.6- tbmd- aincmd aimed whszzotwd‘. eke
Z 33% 6 9:3. 85>? IQ EEO Esobm brzcozccou m 730%

mmd *mmd *omd 53L: 05

a 6 335.328

 

83st?» Egon—usage“. can 8: 6:3 099 cm: unau—

Aood v 9 H8 .32.. .56 v a how i. .modv a 8:865 * .3 .o v a 26: .565 macaw—oboe
=< .Eomomtooo seam—8.80 copaom 2: mm a .2 .o v a a 3m: 2882280 3828
=< any—nut? 1355:0530 use 899 8: was 503.3 macaw—oboe 2F .NN 035.

19

Table 2.3. RDA of environmental variables with land use and the correlation between
RDA axes and relevant variables. * indicates statistical significance, p < 0.05.

 

 

Axis I All 4 axes
Cumulative percentage variance
of species data 54.6 54.8
of species-environment relation 99.6 100
F 28.86 7.28
p 0.001 0.001
Correlation with land use
% Urban 0.48*
% Agriculture 048*
% Forest 075*
% Wetlands 0.07
Correlation with environmental variables
N03 (mg/L) 0.34
NH4-H (mg/L) 0.13
Total N (mg/L) 0.30
Soluble reactive P (jig/L) 0.12
Total P (pg/L) 0.54*
Si (mg Si02/L) 0.43*
CI (mg/L) 0.52*
Water Temp. (°C) 0.35
Conductivity (uS/cm) 0.77*
pH -0.37*
Canopy -0.29

 

20

concentrations. The rest of the RDA axes explained a small fraction of variance in

environmental variables.

Diatom species composition, measured as % similarity to reference condition,
changed along the HDG (r = -0.48, p = 0.007) (Table 2.4), and was also negatively
correlated with % urban, conductivity, SRP, and Si. Diatoms were the dominant algal
group and had a mean relative biovolume of 94% (SD 8%) of cells in counts. Centric
diatoms were a relatively small proportion with a mean relative biovolume of 1%. No
major algal taxonomic group was significantly correlated with the HDG. Of diatom
species, Aclmanthidium minutissimum Ktitzing, Achnanthidium mm, and
Amphora pediculus (Ktltzing) Grunow were the 3 most common species based on
relative abundance, while m m (Nitzsch) Ehrenberg, Navicula tripunctata
(Muller) Bory, and Achnanthidium minutissimum were the 3 most dominant in relative
biovolume.

Only 2 of 21 dominant species were significantly correlated with the HDG (Figure
2.2). Percentages of vabellfia delicatula and Gomphonema pumilum were positively
correlated with the HDG (r = 0.95, p = 0.003; r = 0.65, p = 0.049). The Simpson
dominance index, ranging from 0.05 to 0.70, was not correlated with the HDG. Though
diatom species richness, evenness, and Shannon diversity were not significantly
correlated with the HDG, they were positively correlated with pH and negatively with Si
concentrations.

Average diatom biovolume ranged from 61.8 to 620.2 m3 and was positively
correlated with the HDG (r = 0.39, p = 0.04) (Table 2.4). The highest average biovolume

occurred at one site due to a large proportion of Navicula lanceolata (Agardh) Ehrenberg,

21

a relatively large diatom. Achnanthidium minutissimum, a very small and abundant
diatom, may have constrained the average biovolume at several sites.

The relative abundance of trophic sensitive species, ranging from 0.09 to 0.97,
significantly decreased with the HDG (r = -0.51, p = 0.005), was positively correlated
with % wetlands, but was negatively correlated with % agriculture, Cl', and conductivity
(Table 2.4). The relative abundance of trophic tolerant species, ranging from 0.09 to 0.84,
significantly increased with the HDG (Pearson correlation, r = 0.48, p = 0.009). Both
trophic sensitive and tolerant species were significantly correlated with Cl' and
conductivity, but not with nutrients. Percent motile diatom genera was significantly
correlated with the HDG (r = 0.43, p = 0.02) and conductivity, but negatively correlated
with % forest. While % high N species was not significantly correlated with the HDG, %
low N species was significantly correlated with the HDG (r = -0.45, p = 0.01). Low N
species positively correlated with % wetlands and negatively correlated with conductivity.
Percent high salinity species was correlated with the HDG (r = 0.44, p = 0.01), while %
low salinity species was not correlated with the HDG. Percent high salinity species was
positively correlated with % urban, Cl', and conductivity; on the contrary, low salinity
species was negatively correlated with those. None of pH indicators (% high pH, % low
pH species, and % acidophilic genera) was significantly correlated with the HDG.

Percent native species, ranging from 62 to 100, was negatively correlated with the
HDG (r = -0.46, p = 0.01), while % non-native species was positively correlated with the
HDG (r = 0.46, p = 0.01). Percent native species at all sites had a mean of 94% and a SD
of 9%. After restricted to trophic sensitive species, % trophic sensitive and native species,

ranging from 62 to 1, had a highly significant correlation with the HDG (r = -0.55, p =

22

0.002). Percent native and sensitive species wwere positively correlated with % forest

and negatively with % agriculture, conductivity and Cl'.
Discussion

Results of this study demonstrated that algal assemblages responded to human
disturbance in watersheds, even with the low levels of disturbance in the Grand Traverse
Bay Watershed. All tested community retrogressive characteristics and stress syndromes
responded to the low gradient of human disturbance except species diversity measures.
These attributes have been widely used for ecological studies and assessments, and have
seldom been tested together for stream ecosystems. Current bioassessment tools often use
these characteristics without addressing their ecological bases. Putting bioassessment into
an ecological framework can strengthen its foundation, increase transferability among
types of ecosystems, and facilitate acceptance and communication among biologists.

All hypotheses were tested with field survey data, which is often the scale of study where
ecological patterns are observed (MacArthur 1957, Fischer 1960, Brown 1999) and
generalizations about nature can be made (Naeem 2001). The results of lab bioassays and
microcosm studies are difficult to extrapolate to nature (Cairns and McCormick 1992).
Ecosystem experiments often are not replicated, which makes generalization difficult
(Hurlbert 1984, Carpenter et al. 1998). However, the experimental approach can provide
causal relationships to support survey findings. This study provides the first intensive

survey of algae in low nutrient streams in this region.

23

 

 

 

and EB- 8o 1.35 see 83m saga E: s
$3.? :3 twee no- twee- seam z 33 s
fried _m.0- ivd- *mvd 5331 0:32 0»

seed 839% 252.82 s

. . . . . £89. 2283.
fawn 0- “Rm 0 *mv 0 *0m 0- .3.an 0- yuan—g 05 27562 95
reed- scan 2:32 as

$3 twee $8.: .5223 92:89 as
tone- tone So and :3. :23. 8627. 3283. 9295. as

.23 3o- tone «no- and a =38..ch

Rd .53 85.22.. 833.. away},

8.0- 85585

2 .o- bus»:— 8:55

00.0 man—:95..—

ad- :3 ES N3- scan: 938%

. . . . . . . 52838

x: 0- **0n 0 mm 0 mm 0- 1M 0- *Nv 0 Aim? 0- woman—m u: mung—U
asfiasao 32.8 .85 85.33 s SE: s .83... s 2.5.3:? s 29.

.53 v a .8 t... .S.o v a

.8 .1. .modv a 8:865 * d— .o v a 3a: :32? mecca—9:8 =< any—net? 3:085:20

use em: 98— .388 mus .L «5% FEB 83% .8365 .86on 633: .36
omfiozw 628588 £8585 $2888 @6556 5953 macaw—oboe 2F .vd 033.

24

 

 

reed seas. bias as: a

Leo. N2 sea... 2 :3 s

8336 2532 ex.

Rd 83on 25.3.5.2 9A.

. - . «28% 359.3

*3 0 3.0m 0 93:9: 6:: 0332 §

3o- 862.» 9:32 as

.1de 988% 28.8.8 3.389 as

sane- £2 828% 3283. 93%; s

:3 :13 a Egg—82.6

*wmd 2:295:- 8336 owfio><

*omd 51%.? 09—55569

med- 38¢ Sagan 853%

*hmd- trend wee—Eokm

singed- “1.000 mafia—numb Eugen—m

. - . - . - SEES—Eco

NM O *mm C *mm C . 8min kc OMEGJU
6 a .— 35 .55 2 39a 2.32 2-82 E.

 

fioscueoo .vd 2an

25

 

.7‘

.5I

.4I

% CMDELCAT

 

 

 

M1
.—
O

4.5 -1.0 -.'5 0.0 1.5

HDG

 

1.2

1.0i e

% GOPUMILU
O‘
I

 

 

 

2.0

—-
LII
p—e
O
LII‘
.0
O
LII
p..—
O
LII

HDG

Figure 2.2. The relationships between dominant species and the HDG. A. Cymbella
delicatula, r = 0.95, p = 0.003; B. Gomphonema pumilum, r = 0.65, p = 0.049.

26

In GTBW, human disturbance was a driver of species composition change. The
change of community structure or species composition along environmental gradients has
become one of the most studied relationships in ecology (Whittaker 1975, Austin 1985,
ter Braak and Prentice 1988). Changes in community composition along environmental
gradients have been observed widely in various ecosystems, such as coral reefs
(Ostrander et al. 2002), coastal plants (La Peyre et al. 2001), fish assemblages (Rodriguez
and Lewis 1997), oak savannas (Leach and Givnish 1999), and arthropod communities
(Progar and Schowalter 2002). It has also been observed in stream ecosystems (Leland
and Porter 2000, Griffith et al. 2001, Munn et al. 2002).

Comparison of species composition at reference and test sites was used to
determine community composition change along the HDG. The pollutants from
agriculture and urban land use often include both resources and physiological stressors
(Cooper 1993, Paul and Meyer 2000). Therefore, multivariate statistics (ordination) were
used to characterize the HDG and to evaluate the species composition changes along
multivariate environmental gradients. The assessment of the overall change in species
composition relative to reference condition is advantageous not only in indicating the
degree of changes at community and ecosystem levels, but also can be easily conveyed to
the general public if quantified. Biological attributes described with complex multivariate
ordination are difficult for the public to understand.

Community composition changes can also be broken down into the responses of
characteristic species or indicator taxonomic groups for statistical tests. Both sensitive
and tolerant indicator species were responsive to human disturbance in this study. Species

often have preferences for different levels of environmental conditions, such as oxygen,

27

salinity, pH, and temperature. Sensitivity and tolerance are species traits with roots in
basic biology and ecology, such as shade tolerant species in forests (Canham 1988, Porter
1999), pH sensitive and tolerant zooplankton species in lakes (Brezonik et al. 1993,
Locke and Sprules 1993, Fischer et al. 2001) and salinity and anoxia tolerant species in
salt marshes (Bertness and Ellison 1987). N053 (1990) and Kremen (1994) have proposed
using indicator species to measure environmental changes or impacts for conservation
programs. A well-developed database of species autecology may provide valuable
information for ecological and environmental research.

Trophic sensitive, trophic tolerant species, and low N Species were Significantly
correlated with the HDG and conductivity, but not with any specific nutrient measured.
Streams in the study region are groundwater fed and relatively stable hydrologically
(Boutt et al. 2001). Consequently, mild addition of nutrients in watersheds may not Show
up in stream water because of hyporheic microbial activities, riparian vegetation uptake,
and especially rapid uptake by instream biota. Hence, algal trophic indicators indicated
increased nutrients, even though correlations with nutrients were not observed. N on-
resource pollutants associated with nutrient sources, such as conductivity, pH, and Cl‘,
were also significantly correlated with nutrient indicator species. The change in relative
abundances of species indicative of nutrients enabled the diagnosis that nutrients were
important factors affecting diatom assemblages in this region.

Percent native and % invasive species were significantly responsive to the HDG.
When % native species was constrained with sensitive species, it was more sensitive to
the HDG than % native species alone. Anthropogenic changes in environmental

conditions can increase invasion success of non-indigenous species by creating new

28

habitats for colonization, introducing propagules, decreasing populations of native
species, and shifting the competitive advantage to non-indigenous over native species
(Byers 2002). Native species respond to stressors in watersheds and are considered a
valuable ecological attribute for conservation and assessment purposes (Zampella and
Bunnell 1998, Diamond and Serveiss 2001). However, caution should be exercised to
recognize that not all native Species are sensitive and some species may tolerate the
human disturbance. Tolerant native species may positively respond to the HDG and mask
the important loss of sensitive native species.

Diatom species richness and diversity measures were not significantly correlated
with the HDG. Algal species richness can be negatively correlated with SRP and soluble
inorganic N, especially in the lower range of nutrients (Biggs and Smith 2002). On the
contrary, Dodson et al. (2000) found a unimodal response of phytoplankton richness to
pelagic productivity. van Dam (1982), however, found that algal species diversity
decreased with nutrients. The fixed count (300 algal colonies or 600 diatom valves) of
algal species does not fully assess algal species diversity (Stevenson and Lowe 1986).
Types of pollutants (Patrick 1973) and levels of pollutants may also control the change in
species richness, evenness, and dominance.

The average diatom size was positively correlated with the HDG. Contrary to the
hypothesis of Odum (1985), Peters (1983) stated that increased resources may lead to
individuals in communities with larger sizes. The difference in responses of organism
Size to human disturbance may come from the different effects between resources and
toxic stressors. Pollutants from urban and agriculture land use often include nutrients.

Contrary to toxic stressors such as heavy metals and herbicides, nutrients are resources to

29

algae and stimulate algal growth. Hence, cell size may increase instead of decrease due to
enrichment. Nutrient enrichment can even ease the effect of a toxic stressor (Lozano and
Pratt 1994). Average diatom size was also observed to increase along a phosphorus
gradient in streams and lake littoral zones (Cattaneo 1987, Cattaneo et a1. 1995). Nutrient
enrichment can shift community composition from diatom dominated to big-cell
filamentous green algae (Dodds and Gudder 1992). However, nutrients may have
opposite effects on different taxonomic groups. Excessive cover of algae usually reduces
02 levels, so sensitive and large aquatic insects and fish may respond differently than
algae.

Valued ecological attributes are desired assessment endpoints and should be
applicable to set ecological protection and resource conservation criteria (Campbell 2001).
Stream water quality protection depends on the designated use, as different streams have
different designated usages and are subject to different standards. Different community
attributes respond to environmental gradients with different sensitivities. Hence, different
valued ecological attributes have the potential to be used as the criteria for different
designated uses. For example, % trophic sensitive and native species and % native
species had different sensitivities along the HDG. For comparison, values of these
attributes were plotted against the HDG with a standardized range from 0 to 1 (Figure
2.3). If a 50% decrease in the attribute value is the criterion, % native species had a
higher corresponding level on the environmental gradient than % trophic sensitive and
native species species. Percent native and trophic sensitive Species decreased faster along
the HDG and was more sensitive to the HDG than % native Species. In other words, %

trophic sensitive and native species can be used to set a higher standard than % native

30

species. This is useful to set different criteria to protect valued ecological attributes when

there are different desired protection levels.

In summary, algal assemblages responded to the HDG according to patterns
proposed by Odum (1985) and others in this low impacted area, except for diversity
measures. Average organism size responded positively to the HDG. Species composition
change as well as native species and non-indigenous species were good indicators of
human disturbance. Indicator species, widely used by ecological studies and
environmental assessment, have also been supported as sensitive to human disturbance in
this study. Comparison of species autecology across regions may reveal some interesting
ecological and taxonomical properties and may result in indicator species lists that are
applicable to wide regions. Despite our structural focus, we encourage the test of
responses of functional attributes to human disturbance as proposed by Odum (1985) and
the study of linkages among structural and functional attributes and ecosystem services

that are important for protecting stream ecosystems.

31

 

0
Ln

Attribute Value

 

 

 

 

0 2 4 6 8 10

Human disturbance gradient

Figure 2.3. Differential responses of % trophic sensitive and native species and % native
species to the human disturbance gradient. The lower line is % trophic sensitive and
native species, % native species is the top line. Dashed lines represent the corresponding
environmental gradient values when attribute values are at 50%.

32

CHAPTER 3
RESPONSES AND SIMILARITIES OF DIATOM ASSEMBLAGES FROM
MULTIPLE HABITATS ALONG ENVIRONMENTAL GRADIENTS FROM
MICHIGAN STREAMS
Abstract
Biodiversity in stream ecosystems may be supported by habitat diversity. Epilithon

is the most studied algal assemblage for ecological and bioassessment research in streams.
In this study, diatoms were sampled from multiple habitats, including plants, rocks, and
sediments, in streams through the Grand Traverse Bay Watershed during October 2000.
The similarities among the 3 assemblages were evaluated with detrended correspondence
analysis (DCA), while the responses of assemblages in multiple habitats to an
environmental gradient were evaluated with canonical correspondence analysis (CCA),
beta diversity, and the MAIA trophic index. The reduction of beta diversity along the
disturbance gradient was also evaluated. The major disturbance gradient (MEG) was
defined as the site scores on the lst axis of a principal component analysis, which was
positively correlated with total N, total P, Si, Cl’, and conductivity. Epipelon was
distinctly different from the epilithon and epiphyton cluster in a DCA plot. Assemblages
in each habitat had characteristic species, while epipelon had the most unique species.
Epiphytes had the most % species variance explained and % species-environment relation
variance explained from CCA, while epipelon was ranked 2nd among the 3 assemblages.
Epiphytes also had the smallest beta diversity and fastest species turnover along the MEG,
while epipelon and epilithon had about the same beta diversity and species turnover rate.
Further tests showed that epiphyton was more sensitive than epilithon and epipelon when

the MAIA trophic index was applied to 3 assemblages. Similarities among habitats were

33

not increased along the MEG. Natural habitats, besides rock, deserve more study for

biodiversity and bioassessment.

Introduction

Streambeds are composed of many types of habitats for the colonization and growth
of algae, such as fine organic matter, sand, rock, and plant. Each of these habitats
supports the diversity of algae that contribute to the overall function of stream
ecosystems. Rocks in riffles are the most studied habitat for algae in streams (Round
1991). In many streams, however, other substrata are dominant, and riffles often are rare.
The study of algal communities on substrata other than rock has been limited (Miller et al.
1987, Stevenson and Hashim 1989). Different habitats often have unique algal flora
(Douglas 1958, Kingston et a1. 1983, Stevenson and Hashim 1989, Jtittner et al. 1996,
Lim et al. 2001). However, some studies have shown no consistent distinct assemblages
among different habitats (Sullivan 1982, Winter and Duthie 2000).

Eutrophication has the potential to reduce the heterogeneity of assemblages among
multiple habitats and homogenize inter-habitat diversity (sensu McKinney and Lockwood
1999, Rahel 2002). Differences in assemblages among multiple habitats come from the
differences in the nutrient environment (release of nutrients) and stability of substrata.
Elevated nutrients in stream water may reduce the difference in nutrients among habitats.
Therefore, eutrophication may increase the similarity among assemblages from different
habitats in each site. The total diversity in a site or region is composed of the average

species diversity in habitats and the change of species composition among habitats (inter-

34

habitat diversity) (Loreau 2000). The total diversity may suffer from the reduction of
inter-habitat diversity, while the local diversity in each habitat is intact (Loreau 2000,
Rahel 2002).

From a bioassessment perspective, sampling the habitat where algae are most
sensitive to environmental gradients is preferable for the effectiveness of the monitoring
program. The most sampled habitat for algal bioassessment is also the rocks. The
sensitivities of algal flora to stressors by human disturbance are largely unknown in
habitats other than rocks. Only a few studies compared the sensitivities of algal
assemblages in different habitats (e. g. Winter and Duthie 2000). Even when rocks are
present, epilithon may not be the most sensitive flora. Winter and Duthie (2000) found
that epiphyton was more sensitive than epilithon. Beta diversity, a component of the
regional diversity, is the variation in species composition among localities or the extent of
species replacement or biotic change along environmental gradients (Whittaker 1972). It
is a measurement of species turnover along environmental gradients, hence it can detect
the sensitivities of assemblages to the disturbance gradient.

In this study, algae were sampled in sediment, plant, and rock habitats. I studied the
distinctness of algal assemblages in 3 habitats and analyzed the characteristic diatom
assemblage from each habitat. The sensitivity of response to human disturbance among 3
habitats was compared with 3 methods: CCA, beta diversity, and MAIA trophic index. I
also tested the hypothesis that the similarity of diatom assemblages among 3 habitats

increased along a human disturbance gradient.

35

Materials and Methods

Sampling Area

The Grand Traverse Bay Watershed (GTBW) with an area of approximately 2,600
km2 is located in the northwestern part of the lower Michigan peninsula and has one of
the fastest growing populations in the Midwestern region of the US. (Grand Traverse
County 1996). Traverse City, with a summer recreation population of 500,000, is the
largest city in GTBW. Land use/cover in GTBW is composed of 49% forest, 20%
agricultural, 15% herbaceous/shrub/grasslands, 10% water, and 6% urban. Despite high
development in some catchments, a wide range of development exists throughout GTBW.
The Boardman River drains the majority of GTBW, while the Pere Marquette National
Forest is located in the upper- and mid-region of the Boardman River watershed.
Sampling and Sample Processing

Algae were sampled from rock, aquatic plant, and sediment (fine organic matter)
from stream margins, which often do not occur together at the same site. Sediments were
sampled at 37 sites, rocks at 35 sites, and plants at 15 sites during early October 2000
(Figure 3.1). Twelve sites had all 3 habitats sampled. Algae were collected from 6 rocks
in each site and were scraped with a toothbrush into a white pan. Aquatic plants were cut
off and stored in whirl-pac plastic bags for epiphyte sampling, and included a wide
variety of large plants, such as Myriophyllum spp., Potamogeton spp., and overhanging
grasses that dipped into the stream. Epipelon was sampled with a petri dish inverted to
the sediments, then sealed by a spatula, and stored in whirl-pac plastic bags. Diatoms

were sampled by swirling algae off sands in a container

36

 

 

o Shes

 

 

 

 

‘ I ‘ . ” , * 3.: , ' ‘ “Streams
‘ - ~ ‘ .. '~ F 2, Wetlands
, « . _ ,, . ‘ - - Lakes
w a
20 0 20 40 Miles
E s

 

 

 

 

Figure 3.1. Locations of the sampling sites in Michigan.

37

then decanting suspended algae into a beaker 10 times. Samples were stored on ice until
further analysis in the lab.

Samples were subsampled in the lab for diatom enumeration. Diatoms were digested
with boiling nitric acid method and catalyzed with potassium dichromate. Sampled plants
were also digested with the same method to get digested epiphytic diatoms. Identification
of diatoms was based on Kramrner and Lange-Bertalot (1986, 1988, 1991a, 1991b);
Patrick and Reimer (1966, 1975) were used as supplements. Diatoms were identified to
the lowest level possible and usually to species level.

Canopy cover in the streams was estimated with a densiometer. Flow regimes were
measured with an electronic portable flowmeter (Flo-mate model 2000, Marsh-McBirney
Inc.). Conductivity, water temperature, and pH were also measured in the field with a
multipurpose probe (Denver Instrument).

Water chemistry analysis was done in the lab with a Technicon Auto-analyzer 111
following standard methods (APHA 1998). NHf-N was analyzed with an automated
phenate method, NO3'-N with an automated Cd-Reduction method, and total N (TN) was
digested with persulfate before using the nitrate method. Soluble reactive phosphorus
(SRP) was measured with the automated ascorbic acid method, while total P (TP) was
measured using the SRP method after persulfate digestion and the sample was not filtered.
Cl’ was measured by the automated ferricyanide method, while SiOz-Si was measured by
an automated method for molybdate-reactive Silica.

Data Analysis
The distinctness of diatom assemblages in each habitat was analyzed with detrended

correspondence analysis (DCA). The site scores on the lst DCA axis were plotted against

38

those on the 2nd DCA axes with different symbols for each habitat. Similarity of
assemblages from different habitats can be evaluated from the DCA biplot.

Species unique to each habitat were analyzed by Indicator Species Analysis (ISA)
(Dufrene and Legendre 1997). The indicator value (IV) is calculated by dividing the
mean relative abundance of a species in one habitat by its average in all habitats, then
multiplied by the proportional frequency of the species in that habitat (Dufrene and
Legendre 1997). The highest IV among habitats (IVmax) represents the indicator value
from the associated indicator group for that species. For example, if a species has a
significant IVmax for the epiphyte group, then it is an indicator species of epiphyte.
Statistical significance of the indicator value was estimated by a Monte Carlo method. It
is done by randomly reassigning sites to groups 1,000 times, then calculating IV, and
selecting IVmax each time. The null hypothesis is that the simulated IVmax is no larger
than would be expected by chance. The probability of type I error is based on the
proportion of occurrences that the IVmax from the randomized data set equals or exceeds
the IVmax from the actual data set. The significance level was set at 0.05. In this study,
indicator species were further restricted with the following procedures because the
proportional frequency was not compared among groups. If the habitat in which a species
had its highest relative abundance was different from the habitat with the highest
proportional frequency of observation, then the species was eliminated from the indicator
species list. A second indicator value was calculated as the relative abundance in the
maximum group multiplied by the difference between the maximum proportion and

average proportion divided by two-thirds of the maximum proportion. If the 2nd indicator

39

value was < 23.5, which was the average of the lst indicator values at p = 0.05 level, the
species was discarded.

Environmental variables from all sampling sites were reduced to a composite
environmental variable by principal component analysis (PCA). The lst axis of PCA
usually explains the most variation in environmental data, hence it was used as the major
environmental gradient (MEG) among sampling sites. Site scores along PCA axis 1 were
used to characterize environmental conditions at each site.

The precision of diatom assemblage response from each habitat to the MEG was
compared with % variance explained by the first and all 4 axes from canonical
correspondence analysis (CCA). NO3' and Cl' were excluded from analysis due to a
variation inflation factor > 20 (ter Braak and Verdonschot 1995). Each habitat had a
different sample size; while epiphytes had the smallest sample size of 15. For a fair
comparison, 15 sites were randomly selected for each run from epilithic and epipelic
samples separately for CCA to calculate % variance explained until the means of %
variance explained stabilized. Sample selection for runs was performed with 2 random
functions to emulate the real random process. A stabilized mean was defined as when the
means of % variance explained changed by < 5 % for 10 consecutive runs after at least 50
runs.

Beta diversity was calculated to compare the sensitivities of different assemblages to
the MEG. For each habitat assemblage, pair-wise comparison was calculated with Bray-
Curtis similarity and the difference of scores on the MEG (Pielou 1975). Bray-Curtis
dissimilarity was converted to similarity values by subtracting dissimilarity from 1.0. The

rate of change in assemblage similarity along the MEG distance was estimated with

40

linear regression. Beta diversity was calculated as the change of distance on the MEG
that corresponded to a 50% change in community similarity. In a second approach,
community similarity was calculated as the mean Bray-Curtis similarity with reference
sites. Reference sites were those in the lowest one-third on the MEG for each habitat.
Regression analysis was also applied to estimate the rate of change in assemblage
similarity to the reference condition.

Due to unequal sample size among habitats, random sampling was used to compare
beta diversities from different substrata. Fifteen sites were randomly selected from sites
with epilithon and epipelon 1,000 times to calculate beta diversity. The probability of
eipelon or epilithon with beta diversity smaller than epiphyton’s beta was calculated as
the proportion of the number of times that those habitats had beta < epiphyton’s beta out
of 1,000 runs.

Sensitivity of habitats was further evaluated by the slopes of regression analysis
between MAIA trophic index (Stevenson’s unpublished US EPA Mid-Atlantic region
diatom trophic index) and site scores on the MEG. The slope of the regression model was
used as the indicator of sensitivity. For a fair comparison with epiphyton, 15 samples
from epipelon and epilithon were randomly selected for 1000 runs. The 1000 slopes
generated for epilithon and epipelon were compared with the slope of epiphyton.

To evaluate the reduction of inter-habitat diversity along the MEG, similarity of
species composition among the 3 habitats within each of 12 sites with all 3 habitats
sampled was calculated with Bray-Curtis and Jaccard indices. Bray-Curtis dissimilarity
was converted to similarin values by subtracting dissimilarity from 1.0. Changes in inter-

habitat similarity were examined along the MEG with Pearson correlation analysis. A

41

second inter-habitat diversity, Whittaker’s B diversity, was calculated as total species
diversity divided by the mean species diversity among habitats in each site (Whittaker
1960). The relationship between Whittaker’s B diversity and the MEG was examined

with Pearson correlation analysis.

Results

Generally, studied watersheds had low levels of human land use except for several
sites having either moderate agricultural activity or urban development (Table 3.1).
Nutrient levels were relatively low, especially phosphate. The median TP concentration
was 1.4 ug/L, while median SRP was 1.3 ug/L. TN concentration spanned from 0.10 to
2.70 mg/L, NO3'-N ranged from non-detectable to 2.01 mg/L, and NHf-N was generally
low with a maximum level of 0.11 mg/L. Most pH levels were above 8. Most sites had
conductivity above 300 us/cm. Si had a range from 3.2 to 15.7 mg/L; while Cl' was
between 0.5 and 21.5 mg/L.

Species compositions of samples from 3 habitats were grouped by DCA into two
clear clusters (Figure 3.2). Epiphytic and epilithic samples were inter-dispersed among
each other, however epipelic samples aggregated into a distinct cluster.

Indicator species analysis showed that each habitat had its own characteristic species
(Table 3.2). For example, epiphytes were characterized by Gomphonem pseudotenellum

L-B., Cocconeis placentula var. lineata (Ehr.) V. H. Sm, Eunotia arcus Ehr., and

 

Gomphonema affine Kiltz, while epilithon was represented by Fragilaria elliptica

42

Table 3.1. The minimum, maximum, 25th, median, and 75th percentile of environmental

variables. N = 41.

 

 

Variables Minimum Maximum Percentiles

25th 50th 75‘“
% Agriculture 0 67 9 18 40
% Forest 9 93 21 46 69
% Wetlands 0 27 l 2 5
% Urban 0 43 1 3 8
NO3'-N (mg/L) 0.000 2.010 0.155 0.280 0.740
NHH-N (mg/L) 0.000 0.110 0.002 0.004 0.008
TN (mg/L) 0.100 2.700 0.300 0.420 0.930
SRP (pg/L) 0.7 3.3 0.9 1.3 1.8
TP (pg/L) 0.0 4.7 1.0 1.4 2.0
SiOz-Si (mg/L) 3.20 15.70 7.75 9.10 9.95
Cl (mg/L) 0.50 21.50 4.85 6.00 9.15
Water Temp °C 7.9 16.1 9.1 10.5 11.3
Conductivity (pS/cm) 253 605 313 373 476
pH 7.8 8.4 8.1 8.2 8.2

 

43

 

 

 

 

A
A
‘ ‘
155‘
(\l * ‘ 31‘ ,
.‘Q . * * I “A
x a A
CU I . * . .0“, fl": ‘- ‘A‘
< . it. . ' n
U * ‘ ‘
D 'x O x
I g .
O ..*
0* O .
DCAansl

Figure 3.2. A DCA biplot for all sites from 3 habitats. Arepresents epipelic samples; I
for epilithic samples; * for epiphyte samples.

Table 3.2. Indicator species, indicative habitats, and their indicator values (IV). p

(probability) was for the first IV.

 

 

Species Habitat IV p 2nd IV
Gomphonema pseudotenellum L-B. Epiphytes 35.3 0.001 86.6
Sofipgfilis placentula var. lineata (Ehr.) Epiphytes 80.8 0.001 27.9
Eunotia arcus Ehr. Epiphytes 13.7 0.012 82.3
Gomphonema affine Kutz. Epiphytes 18.6 0.013 62.1
Cymbella caespitosa (Kutz.) Brun Epiphytes 23.9 0.014 60.8
F ragilaria elliptica Schumann Epilithon 36.4 0.001 100.0
Nitzschia dissipata (Kutz.) Grun. Epilithon 60.6 0.002 50.9
Navicula stroemii Hust. Epilithon 28.7 0.004 78.2
fggfgjrgugfipucmvar-8’06”“ Epilithon 40.9 0.000 47.0
Opephora olsenii Mull. Epilithon 24.5 0.011 80.0
Achnanthes linearis (W. Sm.) Grun. Epilithon 20.2 0.013 88.2
F ragilaria leptostauron (Ehr.) Hust. Epilithon 31 0.015 45.3
2,1,:th dzsszpata var. media (Hantz.) Epilithon 31.3 0.047 29.1
Cymbella prostrata (Berk) Cl. Epilithon 21.2 0.051 65.3
Navicula costulata Grun. Epipelon 61.7 0.001 79.1
Stauroneis smithii Kram. Epipelon 66.4 0.001 74.8
Navicula menisculus Schum. Epipelon 65.8 0.001 59.2
gfiiflaria construens var. venter (Ehr). Epipelon 60.9 0.001 42.0
Nitzschia recta Hantz. ex Rabh. Epipelon 54.9 0.001 30.1
F ragilaria brevistriata Grun. Epipelon 56.4 0.001 29.5
Navicula reinhardtii (Grun.) Grun. Epipelon 33.3 0.002 100.0
Navicula viridula (Kutz.) Ehr. Epipelon 34.7 0.002 88.0
Cocconeis neothumensis Kram. Epipelon 59.9 0.002 59.8
Amphora libyca Ehr. Epipelon 42.4 0.002 52.9
Cocconeis disculus (Schum.) Cl. Epipelon 42.4 0.003 57.0
Achnanthes conspicua A. Mayer Epipelon 54.5 0.003 26.7
Navicula libonensis Schoeman Epipelon 28.1 0.004 80.1

 

45

Table 3.2. Continued.

 

 

Species Habitat IV p * 2nd IV
F ragilaria construens (Ehr.) Grun. Epipelon 48.4 0.004 39.6
Achnanthes laevis Schimanski Epipelon 32.4 0.005 47.3
3:12:13; igula (Km) Epipelon 34.4 0.006 53.6
Navicula capitata Ehr. Epipelon 41.9 0.006 52.6
Navicula decussis Oestr. Epipelon 28.2 0.007 73.4
Amphora thumensis (A. Mayer) Krieger Epipelon 22.2 0.009 102.3
Navicula trivialis Lange-Bert. Epipelon 30.4 0.009 60.0
Navicula subrotundata Hust. Epipelon 31.3 0.01 46.3
Navicula veneta Kutz. Epipelon 35.1 0.01 43.0
ggzlSonezs oblongella (Naeg. ex Kutz.) Epi pel on 37.7 0.01 41.9
Navicula cincta (Ehr.) Ralfs Epipelon 22.2 0.013 102.3
Gomphonema innocens Epipelon 22.5 0.014 75.6
Planothidium peragalli Brun & Herib. Epipelon 34.2 0.014 43.2
Navicula gregaria Donk. Epipelon 41 0.014 26.9
giggchza [means var. subtrlzs (Grun.) Epipelon 20.9 0.015 89.7
{635013111111 :P’OS’WO" var' ‘M’i" Epipelon 46.5 0.015 23.7
Cymbella sp. K Epipelon 35.4 0.016 65.5
Navicula pseudoscutifonnis Hust. Epipelon 23 0.02 66.7
Achnanthes exigua Grun. Epipelon 38.1 0.021 25.2
Navicula tenelloides Hust. Epipelon 19.3 0.024 83.0
Nitzschia heufleriana Grun. Epipelon 20.2 0.026 54.8
Navicula reichardtiana Grun. Epipelon 39.7 0.026 24.3
Meridian circulare (Grev.) Ag. Epipelon 27.1 0.027 37.5
f (1313:??? ““901“ m dubi" Epipelon 34.1 0.029 34.7
Navicula cari Ehr. Epipelon 27.7 0.033 35.8
Surirella angusta W. Smith Epipelon 16.7 0.034 67.9
Planothidium joursacense Heri. Epipelon 19.8 0.042 57.1
Amphora montana Kras. Epipelon 13.2 0.05 81.4

 

46

Schumann, Nitzschia dissipata (Kutz.) Grun., Navicula stroemii Hust., Fragilaria
capucina var. gram (Oestr.) Hust., and Achfinanthes lineari_s (W. Sm.) Grun. Epipelon
had along list of characteristic species and featured Navicula costulata Grun., Stauroneis
sggthrr Kram., Navicula menisculus Schumnann, Fragilaria construens var. Leger (Ehr).
Grun., Nitzschg recta Hantz. ex Rabh.

Other than Achnanthidium minutissimum and Fragilaria pinnata, no overlap
occurred in the 5 most common taxa in each habitat. The top 5 dominant species from
epiphyton were Achnanthidium minutissimum, Cocconeis placentula var. hm,
Cocconeis placentuLa var. euglvpta, Gomphonema pumilum, and flyiculja l_anceola_tz_1.
The top 5 epilithon species were Achnanthidium minutissimum, Achnanthidium
b_iasolettian_a, Amnhora pediculus, Navicula cryptotenella, and Fragilaria pm.
Fragilaria pin_nat_a, Fragilaria construens, Achnanthidium minutissimum, Melosira m,
and Fragilaria construens f. v_e_n_t§r were the dominant epipelon species.

More variation in species composition of epiphyton can be explained by
environmental variables in CCA than in the cases of epipelic and epilithic assemblages.
From 110 CCA runs of simulation, no overlap was present between epiphyton and the
box plots of epiplithon and epipelon in % species variance and % species-environment
variance explained from 4 axes (Figure 3.3). Values of % variance explained for
epiphytes, 52.2% species variance explained and 66.8% species-environment variance
explained, were greater than the highest values of the other 2 habitats for all 4 axes. The
mean % species variance and % species-environment variance explained by all 4 axes

stabilized and had values of 39.8 and 58.6% respectively for epilithon (Table 3.3). The

47

 

 

 

 

 

 

 

 

54
-o 52' —
8
E 50‘
g a
3 48'
Q
g 461 -
44. - -
'8 -
a 42‘
:0
5“ 40+

38 _ _ _

Epilithon Epipelon Epiphyton
B.

B 70
G
'53.
a.
6
8 65'
,5 0
g -
5 60- -
g _
8 —
E
“.9 55' 0
.33.
U
8.
‘2 50
5

Epilithon Epipelon Epiphyton

Figure 3.3. Box plots for comparing CCA results from 3 habitats. Both epilithon and
epipelon had 110 CCA runs, while epiphyton had 1 run. A. % species variance explained;
B. % species-environment variance explained. 0 represents outliers.

48

Table 3.3. Variances explained by CCA axes in different habitats. Fifteen sites were
randomly selected from epipelon and epilithon 110 times for CCA runs. All variances are
cumulative.

 

 

Axis I 4 Axes total
Epiphyte (n = 15)
% Species variance
19.2 52.2
% Species-environment variance
24.6 66.8

Epipelon (n = 15)
% Mean species variance

17.0 45.8
% Mean species-environment variance
22.6 61 .0

Epilithon (n = 15)
% Mean species variance

14.1 39.8
% Mean species-environment variance
20.8 58.6

 

49

means were stabilized after 110 runs for epipelon and yielded means of 45.8 and 61.0 %
for species variance and % species-environment variance explained for all 4 axes.

The lst PCA axis explained 28.7% of the variance in environmental variables,
which was about twice the variance explained by the 2nd axis (Table 3.4). The lst
principal component, as the major environmental gradient (MEG), was positively
correlated with conductivity, TP, TN, Cl', Si, and N03, and negatively correlated with pH,
stream order, and canopy. Sites with epiphytic samples spanned 80 % of the 1st PCA axis
(Figure 3.4), while sites with epipelon or epilithon sampled were found along the whole
MEG.

From beta diversity along the MEG distance, epiphytes had the highest absolute
slope values and smallest beta diversity, which were —0.132 and 1.87 respectively (Table
3.5). Epiphytic assemblages had faster species turnover along the environmental gradient
than the other 2 habitat assemblages. Epilithon had about the same slope value and beta
diversity as epipelon. After 1,000 runs of simulation on 15 randomly selected sets of
samples, the chance that epilithon’s beta diversity < epiphyton’s beta diversity was 0.001,
and the probability that epipelon’s beta diversity < epiphyte’s beta diversity was 0.001
(Figure 35A). The frequency distribution of epilithon’s beta diversity was about the same
as the distribution of epipelon’s beta diversity.

When similarity was calculated as average similarity with reference sites, the
absolute values of slopes increased (Table 3.5). The results were consistent with pair-
wise beta diversity outcomes. Epiphytes had the highest slope —0. 13 and smallest beta
diversity 1.96 among 3 assemblages. Epilithon’s beta diversity was close to epipelon’s

beta diversity. After 1,000 runs of 15 randomly selected samples, the chance that

50

Table 3.4. Correlations of PCA axes with environmental variables and % variance
explained for PCA axes. * indicates statistical significance.

 

 

Axis 1 Axis 2
% Variance explained 28.69 14.93
N03 058* 0.75*
NH3 00] 0.05
Total N 0.63* 0.72*
Soluble Reactive P 0.11 -0. 13
Total P 0.60* 0.16
Si 0.52* -0.22
Cl 0.54* -0.36*
Temperature 0.33 -0.50*
Conductivity 0.80* -0.37*
pH -0.59* 0.13
Canopy -0.53* 0.27
Order 063* -0.03

 

51

 

 

 

 

4
3 o
2 o
N
.W o 00
2
N 1 O
at) 00 o O
o
n. 0 50'. 69 o
o .0 O O
I
-1 . ° 0 O o
‘3 O
—2 g o
—2.0 -1.0 0.0 1.0 2.0
PCA axis 1

Figure 3.4. A plot of PCA axis 1 vs. axis 2. 0 represents for epiphytic sites, 0 is for non-
epiphytic sites.

52

Table 3.5. Regression statistics for similarity changes along the MEG with pair-wise
comparison and similarity with reference sites approaches. Sample sizes (n) were the
numbers of possible pair—wise comparison.

 

 

n [l Intercept Slope SE r p
Pair-wise similarity
Epiphyton 105 1.87 0.493 -0. 132 0.024 0.215 0.0001
Epilithon 595 6.46 0.31 -0.024 0.007 0.018 0.001
Epipelon 666 6.33 0.481 -0.038 0.005 0.072 0.0001
Similarity with reference sites
Epiphyton 15 1.96 0.342 -0.13 0.039 0.431 0.005
Epilithon 35 4.37 0.291 -0.04 0.014 0.166 0.009
Epipelon 37 4.57 0.46 -0.06 0.013 0.352 0.0001

 

53

 

30

N
C?

 

Beta diversity

—s
C.3

 

 

 

 

Epipelon Epilithon Epiphyton

 

15

Beta diversity
<75

LII

 

 

 

 

Epipélon Epilflhon Epiphyton

Figure 3.5. The comparison of epiphyton’s beta diversity and 1,000 re-sampled beta
diversity of epipelon and epilithon with box plots. A. pair-wise beta diversity; B. beta
diversity calculated from average similarity with reference approach. 0 represents

extreme values, and O is for outliers.

54

epilithon’s beta diversity < epiphyton’s beta diversity was 0.001, and the probability that
epipelon’s beta diversity < epiphyte’s beta diversity was 0.001 (Figure 3.5B). The
frequency distribution of epilithon’s beta diversity was about the same as the distribution
of epipelon’s beta diversity.

The rate of change in MAIA trophic index with the change in the MEG was greater
for epiphyton (0.685) than epilithon (0.131) and epipelon (0.130). From 1,000
simulations of epilithon’s and epipelon’s slopes, no case of epilithon and 3 cases of
epipelon had slopes greater than the slope of epiphyton (Figure 3.6). The median slopes
of epipelon and epilithon were about the same.

No evidence indicated that nutrients or human disturbance homogenized
assemblages from multiple habitats in this study. Similarity in species composition
among habitats within each site (inter-habitat diversity) was not significantly correlated
with the MEG for both Jaccard and Bray-Curtis indices. Whittaker’s [3 in each site was
also not significantly correlated with the MEG. The lowest Whittaker’s B was 1.6 and the
average among sites was 1.8.

Epipelon and epilithon had about the same number of species recorded from 600
valve counts from 12 sites where all habitats were sampled, 229 and 221 respectively.
Epiphyton had the lowest number of species recorded, 138. On average, epipelon had

58.2 species recorded for each site, 42 for epilithon, and the 36.9 for epiphyton.

Discussion

55

 

1.0
8 o

 

 

. i

 

 

Epilithon Epipelon Epiphyton

Figure 3.6. A comparison of the slopes of 3 habitat assemblages with box plots. The
slopes were derived from the regression between MAIA trophic index and the MEG.
Epilithon and epipelon were randomly sampled for 15 sites 1,000 times. 0 represents for
outliers.

56

Streams, just like terrestrial ecosystems, consist of a mosaic of different habitats.
Habitats sampled in this study had unique algal assemblages. Epipelon was especially
different from epiphyton and epilithon. The unique algal assemblages from different
habitats contributed to the overall diversity of stream ecosystems. Diatom assemblages in
all 3 habitats were capable of indicating the changes in environmental gradients.
However, the beta diversity was not reduced along the MEG.

Epipelon had a quite distinct set of species, while epiphyton and epilithon had more
similar species composition. However, Winter and Duthie (2000) showed that epilithon
was most distinct from epipelon and epiphyton in a study from southern Ontario. From
another study in the Maple River in Michigan, Stevenson and Hashim (1989) found that
epipelon deviated from the episammon, epilithon, epiphyton, and plankton aggregate.
These different results may reflect stream- and region-specific factors affecting
assemblages.

Species have varying degrees of affinity to habitats (Lim et al. 2001), so the affinity
species to habitats may not be clear-cut. The significance test of indicator species analysis
provides a basis for assessment of habitat affinity. The species in the dominant habitat
may affect the abundance of species in other habitats. In this study, the dominance of
sand substrata in sampled streams may contribute to the dominance of sand-adapted
species in epipelon and epilithon. Cosmopolitan species were also found dominating in
all habitats, such as Achnanthidium minutissimum.

Epiphyton had the smallest beta diversity among 3 habitats, which indicated a fast
species turnover along the MEG. Epipelon and epilithon had similar species turnover

rates in both analyses. The within-site Whittaker’s [3 had a range from 1.62 to 1.95, which

57

indicated that sampling a single habitat would miss at least 62 % of the total number of
species in a site. Therefore, sampling multiple habitats is required to reveal the overall
biodiversity in streams. The pattern of beta diversity has seldom been studied for algae in
aquatic ecosystems.

The inter-habitat diversity was not reduced along the MEG. The low nutrient levels
in sampled streams may not have a significant effect on beta diversity. The ecological
pattern of reduction in inter-habitat diversity of algae in streams has not been studied, so
there is no basis for comparison.

From both beta diversity calculations, assemblages in all habitats strongly responded
to human disturbance as indicated by the MEG, thus all have the potential to indicate
water quality. Epiphytic assemblages had the strongest response to the MEG in both CCA
variance explained and beta diversity. Epipelon were the 2nd most responsive to the
MEG in CCA approach. Winter and Duthie (2000) also found that epipelon and
epiphyton were slightly more responsive to the environmental gradient than epilithon in
southern Ontario. From littoral habitats of the St. Lawrence River, Reavie and Smol
(1997) showed that the inference models from diatom assemblages on Cladophora and
macrophytes out-performed the model of epilithon. Rott et al. (1998) indicated that both
epilithon and epiphyton responded to water quality, however no comparison was made.

In this study, various aquatic plants and over-hanging grass were collected to sample
epiphyton. Despite the effects of different aquatic plants on diatom species composition
and dominance (Blindow 1987, Shamsudin and Sleigh 1995), epiphytes responded
strongly to the environmental gradient, which was also reported by Winter and Duthie

(2000). The small species number in epiphyton did not reduce its sensitivity to the

58

environmental gradient, and may be controlled by the constant movement in the flow,
less refugia on the plant surface, and possible toxicants released by plants (Blindow
1987).

Epipelon grows on deposition zones of rich organic materials and nutrients. The
instability of fine substrata and the different chemical environment deter the use of
epipelon for biomonitoring. However, epipelon was used to assess water quality change
where organic sediments were the dominant substrata in an Argentina study (Gomez
1998). Winter and Duthie (2000) reported that epipelon was more sensitive than epilithon,
which was true in our CCA results.

Rocks and other hard surfaces are commonly recommended for diatom sampling to
monitor water quality (e.g. Kelly et a1. 1998). However, epilithon was not the most
sensitive assemblage in this study. Many potential factors may constrain the response of
epilithon to environmental variables. Crevices on the rock surface provide a good shelter
for live diatoms where the chemical property of the substrata may dominate (Hiebert and
Bennett 1992). Rocks, relatively stable substrata, often support dense algal mats, while
the chemical environment inside thick algal mats may be different from outside the mats
(Stevenson and Glover 1993). In hydrologically stable streams, grazers are capable of
clearing overstory algae from the rock surface, and thereby constrain algal species
composition to grazer resistant taxa (Jacoby 1987, Steinman 1992). In such a situation,
epiphyton or epipelon would deserve more consideration for biomonitoring.

Algae on substrata other than rocks were more sensitive to the environmental change
in this study. For many streams in our study region, hard surfaces are not naturally

available. Under this situation, Kelly et al. (1998) recommended using introduced

59

(artificial) substrata for biomonitoring. Natural assemblages, such as epiphyton and
epipelon, are sensitive, and require no colonization period, and deserve consideration.
When assemblages from multiple habitats were combined for analysis, the sensitivity
increased over any one habitat alone (Winter and Duthie 2000). A similar result was also
found in a study of algae in Maine wetlands (Wang and Stevenson unpublished results).
For assessment purposes, the sensitivities of epiphyton and epipelon need to be
established with more studies. The autecology of species in different habitats need to be
compared, thus habitat effects can be understood and transfer functions may be
developed. Other natural substrata, such as sand, dominant in certain streams, and wood,

a stable dominant substratum in certain streams, are worthy of study for bioassessment.

60

CHAPTER 4
EFFECTS OF SMALL-SCALE NUTRIENT ENRICHMENT ON ALGAL
COMMUNITY COMPOSITION IN THE GRAND TRAVERSE BAY
WATERSHED
Abstract
The goal of ecological experiments is to understand how natural ecosystems work.

Due to problems associated with experiments, such as lack of natural complexity, small
spatial-temporal scales, and insufficient experimental duration, extrapolation is required
for predicting events at the large scale. To study regional patterns in nutrient limitation,
clay pots with nutrient enriched agar were placed in 16 streams for 2 weeks in the Grand
Traverse Bay Watershed. Five treatments (N, P, Si, N+P, N+P+Si) and one control were
applied. No consistent regional pattern in nutrient limitation was observed for periphyton
biomass. However, species composition responded to nutrient enrichment as shown by
the significant differences in detrended correspondence analysis (DCA) scores between
treatments and controls. The plot of the first 2 axes of DCA also revealed that nutrient
enrichments did not change species composition on clay pots similar to those
assemblages in high nutrient streams. Community assemblages from treatment clay pots
in the same site were more similar than assemblages on clay pots from the same
treatments across sites. Comparison between the rankings of species response to nutrient
enrichments and rankings of species environmental optima indicated no consistent
correlation pattern. Also no correlation was found between rankings of species response

to nutrient enrichments and rankings of species environmental optima from another

regional survey or species indicator values in van Dam’s trophic index. Therefore,

61

extrapolation of experimental results from small-scale experiments to large regional

scales is probably constrained by the local species pools in streams.

Introduction

The ultimate goal of ecological experiments is to understand how the natural world
works. Experiments can examine the processes and mechanisms important to the research
topic and infer causal relationships. However, the results of experiments may differ when
the physical size of experiments changes (Petersen et al. 1999). Field experiments or
model ecosystem studies suffer from the problems of scaling-up and artifacts of
experimental systems (Petersen et al. 1999, Wiens 2001). Experimental systems often
have problems such as no replication for whole ecosystem experiments, lack of natural
complexity (Samelle 1997), insufficient experiment duration (England and Cooper 2003),
and wall effects (Bloesch et al. 1988, Lyche et al. 1996). Hence, the validity of
extrapolation of their results to large scales is often in doubt (Carpenter 1996, Schindler
1998).

Maintaining the regional patterns of nutrient limitation may protect ecological
integrity and biodiversity. Within a region, the understanding of nutrient limitation may
help control nuisance algal blooms (Conley 1999). Nutrient-diffusing substrate (NDS)
experiments often are short-term and smaller than 1 m2 in spatial scale, and have been
widely applied to evaluate the limiting factor for algal growth (Fairchild et al. 1985,
Carrick et al. 1988, Wold and Hershey 1999, Scrimgeour and Chambers 2000). A recent

review of NDS experiments showed that both N and P were of the same importance in

62

limiting algal growth in streams (Francoeur 2001). The pattern of regional nutrient
limitation is often controlled by biogeochemistry. Nitrogen limitation is common in the
Western United States (Grimm and Fisher 1986, Hill and Knight 1988). In Michigan,
streams were found P limited (Pringle and Bowers 1984, Fairchild et al. 1985, Burton et
al. 1991). In contrast, two regional studies found that there was no uniform regional
pattern in nutrient limitation (Wold and Hershey 1999, Snyder et al. 2002).

Most studies reported only responses of algal chlorophyll to nutrient enrichment
(Lohman et al. 1991, Chessman et al. 1992, Scrimgeour and Chambers 2000, Mosisch et
al. 2001, Dodds et al. 2002). Algal species have different and sensitive responses to
different nutrients (Fairchild et a1. 1985, Carrick et al. 1988, Keithan et al. 1988, Marks
and Lowe 1993). The increase in total biomass due to a specific nutrient does not indicate
the responses for all algal species. Only few studies described responses of algal species
composition to nutrient amendments (Pringle and Bowers 1984, Fairchild et al. 1985,
Carrick et al. 1988, Marks and Lowe 1993). Shifts in species composition are probably
more sensitive to nutrient enrichment than changes in biomass, especially when biomass
may be regulated by grazers or physical disturbances.

The responses of the algal community to NDS experiments have not been compared
to algal assemblages in nutrient enriched streams at the regional scale. If small-scale
nutrient enrichment experiments can change community structure on clay pots similar to
those in high nutrient streams, no extrapolation is needed for predicting large-scale events
from small-scale experiments. Alternatively, if the algal assemblages on nutrient enriched
clay pots are more similar to assemblages on control clay pots than those in high nutrient

streams, the validity of small-scale experiments for inferring nutrient effects on

63

community structure is in question and the design of experiments may need to be
modified. Species responding to nutrient enrichments at the experimental scale may not
respond to nutrient enrichments at the regional scale because of constraints on
immigration of species adapted to exploit high levels of nutrients. The comparison of
species responses from experimental and regional scales may illuminate linkages between
community structure and function across scales.

This study evaluates the pattern of regional nutrient limitation for the Grand
Traverse Bay Watershed (GTBW) where sandy substrata, high grazing pressure, and low
nutrient levels make this evaluation challenging. NDS were used to determine whether N,
P, or Si limits algal biomass accrual and regulates species composition. I also evaluated
whether nutrient enrichment can change diatom assemblages similar to those in high
nutrient streams. Responses to nutrient enrichment by individual species were assessed,
and then the responses of species to small-scale experimental nutrient enrichment were

compared to their responses to nutrient availability at the regional scale.

Materials and Methods
Study Area
GTBW, with an area of approximately 2,600 kmz, is located in the northwestern part
of the lower Michigan peninsula (Figure 4.1). It has one of the fastest growing
populations in the Midwest United States. Traverse City, with a 500,000 summer
recreation population, is the largest city in the watershed. Land use in GTBW is
composed of 49% forest, 20% agricultural, 15% herbaceous/shrub/grasslands, 10% water,

and 6% urban. The Boardman River drains the majority of the land of GTBW.

Sampling and Processing

Sixteen sites were selected for nutrient enrichment experiments with nutrient-
diffusing clay pots, which had a sampling area of 75.6 cm2 and an internal volume of 100
ml. Clay pots were soaked in deionized water for at least 24 hours to wash off potential
toxic contaminants. A plastic petri dish was stuck to the mouth of each pot with silicon
glue. Five types of enriched agar were made: N (NaNO;, 1 mol/L), P (KzHPO4, lmol/L),
Si ($02, 1 mol/L), N + P (NaNO;;, 1 mol/L; KzHPO4, lmol/L), and N + P + Si (NaNOg,
1 mol/L; K2HP04, lmol/L; $02, 1 mol/L). Chemicals were dissolved in 2 % hot agar,
then poured into clay pots. Each clay pot was sealed with a rubber stopper (size 000).
Control pots were filled with pure agar. For each stream, 2 clay pots were made for each
treatment and control. Four randomly selected clay pots were glued to a tile to anchor the
pots into streams. Clay pots were kept moist before placement in streams.

Clay pots were placed in streams on 28 and 29 September 2000 and collected 14
days later. The material on clay pots from the 2 pots for each treatment was scraped off
with a toothbrush, stored in a whirl-pac plastic bag, and kept on ice. Canopy cover above
clay pots was estimated with a densiometer. Current velocities were measured with an
electronic portable flowmeter (Flo-mate model 2000, Marsh-McBirney Inc.).
Conductivity, water temperature, and pH were also measured in the field (APSO, Denver
Instrument).

Water chemistry was analyzed with a Technicon Auto-analyzer HI by following
standard methods (APHA 1998). NH3-N was analyzed with an automated phenate
method, NO3-N with an automated Cd-Reduction method, and total N (TN) was digested

with persulfate before using the nitrate method. Soluble reactive phosphorus (SRP) was

65

 

 

 

9‘, .

'Jrf

 

 

A“

16 Ill-8

e Shes
Streams

‘1 Wetlands
- Lakes

Figure 4.1. Site locations in the Grand Traverse Bay Watersheds, Michigan.

66

measured with the automated ascorbic acid method, while total P (TP) was measured
using the SRP method after persulfate digestion and the sample was not filtered. Cl’ was
measured with the automated ascorbic acid method, while total P (TP) was measured
using the SRP method after persulfate digestion and the sample was not filtered. Cl' was
measured by the automated ferricyanide method, while SiOz—Si was measured by an
automated method for molybdate-reactive Silica.

Algal samples were subsampled in the lab for chlorophyll analysis, diatom
enumeration, and soft algae enumeration. Chlorophyll a (Chl a) was extracted by
dissolving the filter membrane and algal pigments in 90 % aqueous acetone and
measured following standard methods (APHA 1998) with a spectrophotometer (Spectro
Instrument). Chl a concentration was calculated as mg/cmz. Diatoms were digested with
the boiling nitric acid method, and catalyzed with potassium dichromate. Diatoms were
identified primarily based on Kramrner and Lang-Bertalot (1986, 1988, 1991a, 1991b),
and Patrick and Reimer (1966, 1967) were used as supplements. Diatoms were identified
to the lowest level possible, often species.

The regional nutrient limitation pattern was evaluated for algal biomass, measured
as Chl a, by the relation between a nutrient limitation index and stream nutrient
concentrations, nutrient ratios, surrounding land uses, current velocity, and canopy cover
from 16 sites. The nutrient limitation index was calculated as the ratio of the difference
between the Chl a level of a treatment and the Chl a of the control divided by the Chl a of
the control. To test the hypothesis that streams in the GTBW were generally limited by
one nutrient or another, the mean value of the nutrient limitation index for each treatment

was compared to 0 with a one—sample t test. To test the hypothesis that nutrient limitation

67

was related to nutrients and other environmental factors, the relationships among nutrient
limitation indices for 5 treatments and nutrients, nutrient ratios, flow, and canopy were
evaluated by Pearson correlation analysis. Si/DIN (N03 +NH3) and TN/TP were
calculated as molar ratios. I assumed that N:P > 23 indicates P limitation, while both Si:P
< 16 and Si:N < 1 indicate Si limitation (Redfield 1958).

Detrended correspondence analysis (DCA) was used to determine whether nutrient
enrichment with NDS can change diatom species composition in a direction similar to
assemblages in high nutrient streams. Relative abundances of species were used for
analysis after log-transformation. The correlations between the first 2 DCA axes and
environmental variables were analyzed with the Pearson correlation analysis. To evaluate
the relationship between changes in species composition and treatment effects, the mean
difference of the first 2 DCA axes scores between a treatment and the control was
compared with zero by a 2-tailed one-sample t test.

To evaluate the response of species at the experimental scale, the species response
index was calculated by dividing the difference in species relative abundance between a
treatment and the control with relative abundance in the control. The absolute values of
this indicator were log-transformed before statistical analysis to meet the assumption of
parametric statistics and control for potential overestimation of common species
responses. The relative abundance of species not found in 600 valve-counts were
substituted by 1/1,200, assuming that species were in low abundance and only lvalve
could be found in 1,200 valve-count. In the one-sample t test, the mean response of a
species from 16 clay pots was compared to zero. The mean value of species response

index was used to rank the strength of species response.

68

To assess the responses of species at regional scale, species appearing in > 3 sites on
control clay pots were chosen for weighted average calculation. TN, TP, Si, and Cl
concentrations were selected for analysis. Cl level has been used as an indicator for
human disturbance (Herlihy et a1. 1998) and was used as a surrogate for trophic state in
this study. Species optima for each specific nutrient were then ranked among species.

The patterns of species response at experimental and regional scales were compared.
The ranks of species responses at both scales were evaluated with the Pearson correlation
analysis to see whether species with strong responses at the experimental scale also had
high ranks at the regional scale. The ranks of species responses to N, P, and Si treatments
were compared to the ranks of species TN, TP, and Si optima respectively. The ranks of
species responses to N+P and N +P+Si treatments were compared to the ranks of species
Cl optima. The top 10 ranked species from each treatment were examined for differential
responses between experimental and regional scales. Species responses at the
experimental scale were also compared with the ranks of species Cl optima from another
regional survey, in which epilithon was sampled from 31 sites in GTBW. The correlation
between species ranks in N+P and N+P+Si treatments at experimental scale and species
indicator categories in van Dam’s trophic index (van Dam et a1. 1994) was evaluated. van
Dam’s indices for taxa were developed at a global geographical scale and therefore

represent a categorization of species at a broad scale.

Results

69

Generally, sampling sites had very low phosphorus and higher nitrogen
concentrations (Table 4.1). The maximum TN was 1.46 mg/L, whereas the minimum TN
was 0.08 mg/L. Ammonium levels were relatively low compared to nitrate levels. NO3-N
ranged from 0.01 to 1.18 mg/L, while NH3-N varied from non-detectable to 0.03 mg/L.
The maximum TP was very low (2.4 [lg/L), while the minimum was close 0 rig/L. pH
ranged from 7.7 to 8.3, while conductivity spanned from 253 through 587 us/cm. Stream
order spanned from 1 to 4. Only 3 sites had canopy cover < 50%. Four sites had about
50% agricultural land in the watershed, while urban development in watersheds was
below 15%.

The NDS failed to show evidence for nutrient limitation of algal biomass accrual.
The Chl a on control clay pots had a range of 0 to 524.4 ug/cmz. The means of nutrient
limitation indices from all 5 treatments were not significantly different from 0 (Figure
4.2). No regional pattern in nutrient limitation was observed based on algal biomass.
However, based on water chemistry, all sites should be P limited because all DIN/SRP
ratios were > 30 and TN/T P ratios > 138. The nutrient limitation indices were not
correlated with any nutrient, land use types, canopy, flow, or other environmental
variables.

A plot of the first 2 axes of DCA showed that changes in species composition of low
nutrient streams caused by NDS nutrient enrichment did not resemble assemblages in
high nutrient streams (Figure 4.3). Species composition on all clay pots from the same
site was more similar to each other than to assemblages on clay pots from the same

treatments across sites. The lst DCA axis was significantly correlated with N03, NH3,

70

Table 4.1. The summary statistics for land use, environmental variables, and nutrient

ratios. ND. is short for non-detectable.

 

 

Variables Minimum Maximum Percentiles

25th 50th 75th
% Urban 0 15 0 2 8
% Agriculture 0 64 3 16 45
% Forest 15 93 25 53 79
% Wetlands 0 27 1 2 5
Conductivity (uS/cm) 253.0 587.0 308.3 345.5 467.3
pH 7.7 8.3 8.1 8.2 8.2
Water Temp (°C) 7.9 13.0 9.2 10.5 11.2
Canopy 0.5 73.0 7.1 18.0 48.7
Stream Order 1 4 1 1.5 2
NO3'-N (mg/L) 0.01 1.18 0.06 0.14 0.51
NHi-N (mg/L) N.D. 0.03 0.00 0.01 0.01
TN (mg/L) 0.08 1.46 0.19 0.27 0.91
SRP (ug/L) 0.8 1.7 0.8 1.0 1.2
TP (ug/L) 0.3 2.4 0.9 1.2 1.5
Si (mg/L) 7.7 17.2 8.3 9.0 9.6
Cl (mg/L) 1.2 21.4 3.2 5.1 10.1
TN/TP 138 3592 31 1 674 2305
DIN/SRP 32 3103 100 330 1147
Si/DIN 4 344 10 32 62

 

71

 

 

8
0

>4 6‘
g o
.5 o O
a
g 4 O
E e
' o
H 2
E
:3
z

 

v %%%é

-2 l

v v

N NT? NRSi P Si

 

Figure 4.2. Box plots of nutrient limitation indices. 0 represents outliers, while 0
represents extreme values.

72

and TN (Table 4.2). The 2nd axis was positively correlated with SRP, but negatively
correlated with % forest. Therefore, nutrients affected species composition more on a
regional scale among streams than among experimentally nutrient enriched treatments
within each stream.

Species composition of diatom assemblages did, however, respond consistently to
experimental nutrient enrichment even though they did not correspond well to regional
relationships between biomass, species composition, and nutrient enrichments.
Comparison between species composition in treatments and controls on the lst DCA axis
showed that the differences between N, P, NP, and NPSi versus controls had means
significantly > 0 (Table 4.3). On the 2nd DCA axis, the difference between Si, P, NP, and
NPSi versus controls had means significantly > 0.

One-sample t tests showed that the relative abundances of 13 species increased
compared to controls in N treatments (Table 4.4). However, M. Lneta decreased in N
treatments. Eighteen species increased their relative abundances compared to controls in
P treatments, while 9 species increased their relative abundance compared to controls in
Si treatments. Six species decreased from control clay pots in Si treatments. In NP
treatments, 20 species had increases in relative abundances, while Cyc. distinguenda, Er_a.
delicatissima, and Egg. W decreased in their relative abundances. In NPSi treatments,
13 species increased their relative abundance except Ann). libm, which decreased its
relative abundance compared to controls.

Species responding to nutrient enrichment experiments did not respond to nutrient

enrichment at the regional scale. No correlation was found between ranks of species

73

 

 

 

 

2.5
SRP
2.0 '1 v
v v' +
(\l 1.5 . 33,, D v 'H' +
.2 +
§ X 9 0% Ch a
1.0 I A
< & $3 . I: °
U n I ' ‘ A "
Q 5 ' .x i rAA
,g<x
.‘*'.**
0.0 I it
% forest 5
-.5 00 .5 1.0 1.5 2'0 2'5 3.0
NH3 DCA axis 1 N03-

TN

Figure 4.3. A plot of the first 2 DCA axes scores of all clay pots. Symbols represent
different sites. The first DCA axis was positively correlated with N03' and TN, and
negatively correlated with NH3. The second DCA axis was positively correlated with
SRP and negatively with % forest.

74

Table 4.2. Correlation between DCA axes of clay pot diatom species and environmental
variables. r is Pearson correlation coefficient. DCAl is the short for DCA lst axis; DCA2
for DCA 2nd axis. * indicates 0.01 < p < 0.05; ** indicates 0.001 < p < 0.01.

 

Correlated environmental variables

 

DCA axis 1 N03- NH3 TN
r 0.71** -O.70** 054*
DCA axis 2 ”/0 Forest SRP

r -0.5 l * 050*

 

75

Table 4.3. The comparison of the difference between a treatment and the control to 0 on 2
DCA axes with t test.

 

t Mean Difference p (l-tailed)

 

DCA axis 1

N — C 1.98 0.06 0.04 *
NP — C 2.74 0.14 0.007 *
NPS - C 1.85 0.10 0.03
P - C 2.27 0.09 0.01 *
S — C 0.99 0.05 0.17
DCA axis 2

N — C 1.47 0.04 0.07
NP—C 2.47 0.10 0.01 *
NPS - C 2.82 0.19 0.008 *
P — C 1.79 0.07 0.03 *
S — C 4.03 0.17 0.0001 *

 

76

Table 4.4. The responses of species to nutrient enrichment treatments compared to
control clay pot. T-tests were performed on the relative abundance difference between
treatments and the control. * indicates p < 0.05 and ** represents p < 0.01.

 

Species N

P

Si NP NPSi

 

Achnanthes conspicua A. Mayer *

Ach. exigua Grun.

Ach. laevis Schimanski

Ach. lauenburgiana Hust.
Ach. nollii Bock

Amphora libyca Ehr.

Amp. pediculus (Kutz.) Grun.

Cocconeis neothumensis Kram.
Coc. placentula var. euglypta (Ehr.)

Cl.
Coc. placentula var. lineata (Ehr.) **

V. H.

Cyclotella distinguenda Hust.
Cyc. meneghiniana Kutz.
Diploneis oblongella (Naeg. ex
Kutz.) Ross

Eunotia bilunaris (Ehr.) Mills *
F ragilaria capucina var. rumpens

(Kutz.) Lange-Bert.

F ra. capucina var. vaucheriae *
(Kutz.) Lange-Bert.

F ra. construens f. venter (Ehr).

Grun.

F m. delicatissima (W. Sm.) Lange-

Bert.

F ra. leptostauron var. dubia (Grun.)

Hust.

F ra. nanana Lange-Bert.
Gomphonema acuminatum Ehr.
Gom. angustatum (Kutz.) Rabh.
Gom. clavatum Ehr.

Gom. micropus Kutzing

Gom. parvulum Kutz.

Navicula capitatoradiata Germain
Nav. costulata Grun.

Nav. cryptocephala Kutz.

77

a:

a:

** *

**

**

Table 4.4. Continued.

 

Species

NP NPSi

 

Nav. gregaria Donk.

Nav. menisculus var. grunowii
Lange-Bert.

Nav. radiosa Kutz.

Nav. reichardtiana Grun.

Nav. seminulum Grun.

Nav. subrotundata Hust.

Nav. tripunctata (O. F. Mull.) Bory
Nav. veneta Kutz.

Nitzschia dissipata (Kutz.) Grun.
Nit. gracilis Hantz. ex Rabh.

Nit. incognita Krasske
Nit. linearis (Ag. ex W. Sm.) W.
Sm.

Nit. palea (Kutz.) W. Sm.

Nit. palea var. debilis (Kutz.) Grun.

Nit. pura Hust.

Nit. recta Hantz. ex Rabh.
Planothidium lanceolata var. dubia
(Grun.) Lange-Bert.

Pla. peragalli Brun et Herib.
Reimeria sinuata (Greg) Kocio. &
Stoer.

Stauroneis smithii Kram.
Surirella angusta W. Smith
Synedra ulna W. Sm.

,**

**

78

response from treatments and ranks of regional species environmental optima (Figure
4.4). No correlation was observed between ranks of species response from treatments and
the ranks of species Cl optima from another regional survey (Figure 4.5) and van Dam’s
species classification (Figure 4.6).

From those species ranked in the top 10 from each treatment, 3 species ranked about
the same at both experimental and regional scales: Nav. radiosa, Nav. costrLata, and N_i_t_.
sociabilis. Eight species had ranks at the experimental scale close to ranks at the regional

scale (within 5 ranking difference): Ach. conspicua, Ach. lauenburgiana, Ach. exigua,

 

Cvm. microcephala, Nav. cryptocephala, Nit. recta, M. dissipata, and &. lanceolata var.
dubia. All other species ranked much lower at the regional scale than at the experimental

scale.

Discussion

The N :P ratios and low P concentrations in stream water showed a regional pattern
of P limitation in the GTBW, but this was not confirmed with NDS experiments. Our
results showed no correlation between nutrient limitation indices and canopy cover or
flow. Short experimental periods, low colonization rates, and high grazing may have
limited algal accumulation and manifestation of nutrient limitation with NDS. Spatial
heterogeneity of nutrient limitation has been observed in Minnesota streams (Wold and
Hershey 1999), Alberta streams (Scrimgeour and Kendall 2002), and Idaho rivers
(Snyder et al. 2002). Spatial heterogeneity of nutrient limitation may come from

heterogeneity of biogeochemistry, point sources, and non-point sources. Anthropogenic

79

 

70

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I . '

cu ' ° .
E 601 I . . g
.5. ' I . .
a. . . ' I a.
O 50‘ .. ' I O
E . . ' I . O.
W 401 . . F
0 I u:
'8 I . . ' .2
g‘ 30‘ ' . I . 8
g_ I '.. 5?
O 'f u—
m 20' . . . O
x
g . I .' . . '2
I: IO‘I ' a

I ' C2

0 I :1 '
0 10 20 30 40 50 60 70 _
0 10 20 30 40 50 60 70
Ranks of species response from N treatments
Ranks of species response to P treatments
70 . , 70 .
a . ' .I. . . I. . .0
.E w‘ . . . . .. g 60‘ . .% .
o 50. . ' . - . o 50. 0 o. '
E75 . . . I a . . O
3 40‘ '0 o ' g; 40‘ ' 0
H- . I . ‘- p . ..
i I ' .. 8 . I .
U, 30. II . a. 3m 0 I .
‘0- . . W . . .
O I . “- . . .
m 20. ' o ° 9 I
x g . 9 u: 20‘ ’.
g . . . . é ' . . .
M '0‘ ' . I . a? 101 I. .
C I . . ' . . I
0 I . . 0 . .A C
0 '0 20 30 40 50 60 70 o 10 20 3o 40 50 60 70

Ranks of species Cl optima

Ranks of species response to Si treatments

 

 

 

 

 

30 40 50 60 70

Ranks of species response to NPSi treatment

Ranks of species response in NP treatments

Figure 4.4. Relationship between species environmental optima and species responses to
treatments. The lines indicate 1 to 1 ratio.

80

 

p—I
s a s a

Ranks of species Cl optima

N
O

 

 

 

 

Ranks of species response in NP treatments

 

Ranks of species Cl optima

 

 

 

 

0 10 20 30 40 50 60 70
Ranks of species response in NPSi treatments

Figure 4.5. The relationship between ranks of species Cl optima from a regional survey
with A. ranks of species response to the NP treatment, B. ranks of species response to the

NPSi treatment. The lines indicate 1 to 1 ratio.

81

 

7- IIII III IIII

Categories of van Dam's trophic state
‘3
I
I
I
I
I
I
I
I

 

 

 

0 10 20 30 40 50 60 70

 

6
SdI II-IIIII .I I I I
4 I IIII I II I I

Categories of van Dam's trophic index

 

 

 

0 10 20 30 40 50 60 70

Ranks of species response in NPSi treatments

Figure 4.6. The relationship between van Dam’s category and A. ranks of species in NP
treatments, and B. ranks of species in NPSi treatments.

82

inputs of nutrients can change nutrient ratios in streams (Vitousek et al. 1997). Arbuckle
and Downing (2001) found that pasture land led to a low N:P ratio and row-crop
agriculture contributed to a high N:P ratio. Scrimgeour and Chambers (2000) also
showed that point-source inputs could change nutrient limitation patterns. Other factors,
such as light (Hill and Knight 1988, Steinman et a1. 1989) and grazers (Rosemond et al.
1993), can regulate algal growth rates.

Species composition in short-term nutrient enrichments did not resemble
communities in nutrient-enriched streams. Diatom assemblages from treatment clay pots
tended to cluster more by site than by treatment. Synder et al. (2002) found similar results
and showed that assemblages in treatments were closer together within the same
watersheds than within the same treatments. Hence, they suggested that large-scale
factors are more important in determining the potential diatom assemblage than the
small-scale, proximate nutrient enrichment provided by NDS. Kohler and Wiley (1997)
found that the changes of community composition in microcosm experiments, where the
caddisfly Glossosoma was manually removed, were different from the changes in
community structure when Glossosoma was naturally wiped out by parasites. Brown et al.
(2001) also found that short-term observation or experiments did not detect complex
dynamics of rodent populations found by long-term study. Therefore, the results of small-
scale short-term experiments may not be extrapolated to natural ecosystems.

Most species that responded to nutrient enrichments at the experimental scale did
not respond to nutrient enrichment at the regional scale. However, diatom species did
respond consistently across the region to nutrient enrichments with consistent changes in

DCA scores when comparing treatment to control NDS assemblages. Such subtle

83

responses may only have validity at the experimental scale, and not at the regional scale.
This scale effect is apparent in the DCA plot where diatom assemblages from treatment
clay pots tended to cluster by site than by treatment.

I also found little similarity between species responses on NDS and species
responses to nutrients reported in the literature. Autecological profiles provide responses
of species at the regional scale. Therefore, indices, such as the trophic state index, may
serve as references for predicting ecological changes imposed by human disturbance or
climate change. However, disagreement among indices in different literature sources
impeded their applicability to my results. Among those species ranked closely at both
scales, all had disagreement in their trophic-state associations in the literature. Ach.
e_xigfl and Nav. cryptocephala were ranked as widely tolerant by van Dam, eutrophic by
Lange-Bertalot, but sensitive by Bahls. Nav. cryptocephala was also ranked as eutrophic
by Lowe (1974). Ach. conspicg and Nit. ;e_cta were categorized as widely tolerant in
both van Dam’s and Lange-Bertalot’s trophic indices, but were classified as sensitive by
Bahls. Ach. lauenburgiana was classified as oligotrophic by van Dam and Bahls, but
eutrophic by Lange-Bertalot. gag. microcephala was categorized as mesotrophic (less
tolerant) by van Dam and Bahls, but as wide tolerant by Lange-Bertalot; while M-
M was classified as eutrophic by both Lowe and Lange-Bertalot, mesotrophic by
van Dam, and sensitive by Bahls. Among 3 species ranked about the same at both scales,
disagreement on their autecology persisted in the literature. N_it. sociabilis was classified
as a eutrophic species by van Dam et al. (1994), widely tolerant by Lange-Bertalot (1995),

and less tolerant by Bahls (1993). While Nav. Qdiosa was classified as mesotrophic by

84

van Dam and eutrophic by Lange-Bertalot, it was categorized as oligotrophic (sensitive)
by Bahls. Nav. costulata was classified as less tolerant by Bahls.

Though Bahls’ diatom tolerance index was developed in the US. and has an
extensive list of species found in the US, it had little agreement with species autecology
found in this study. Despite accepted applicability of van Dam’s indices, they have been
criticized for not checking the validity of indices when applied in different regions.
Differences in species autecology among regions may arise due to biogeographic
variability among populations. Therefore, species autecology profiles and indices
developed from a region may be required to appropriately compare species responses to
environmental change in experiments and across regions.

The design of experiments and extrapolation of experimental results may need to
consider factors associated with the species pools involved and regional heterogeneity
(Englund and Cooper 2003). The lack of resemblance in species composition at
experimental and regional scales is probably because the local species pool controls the
community structure of short-term small-scale experiments. In microcosm or clay pot
experiments, colonized species are from the local species or supplied species pool.
Without enough time and the release of competition for species colonization from other
species pools, the end community structure is decided by the local species pool or
supplied species pool.

Community structure changes when disturbance occurs. The local community
structure is controlled by species pools from both local and regional scales. The regional
species pool changes the community structure when a press perturbation occurs over a

large spatial-temporal extent. In a stream, the local community is colonized by the local

85

species pool drifting from upstream. Species from a local species pool are adapted to the
local in-stream environment and can out-compete species from the regional species pool
that are colonizing through aerial dispersal or introduction by animals. When a large area
is affected by a press perturbation, invasive species from the regional species pool may
out-compete the local species, change the community structure, and become established
in the local species pool (Byers 2002). For example, phosphorus enrichment due to the
change of land use shifted the Florida Everglades from a sawgrass dominated ecosystem
to a cattail dominated marsh (Newman et al. 1998). Predator addition enabled invasive
zooplankton species to establish in experimental ponds (Shurin 2001).

Only with the knowledge of the species pool involved can we predict the expected
change of the impacted ecosystems (sensu Kolar and Lodge 2002). Through
understanding the autecology of species involved, we may more precisely predict the
magnitude and outcome of impact on ecosystem structure and function from a specific
event. Uncertainty is much higher when inferring the structural and functional changes of
ecosystems without knowing species involved. Therefore, small-scale experiments need
to include the regional species pool, instead of only the local species pool, to extrapolate

to real ecosystems and over large spatial and temporal scales.

86

CHAPTER 5
DEVELOPMENT AND EVALUATION OF A DIATOM-BASED INDEX OF
BIOTIC INTEGRITY FOR THE INTERIOR PLATEAU ECOREGION
Abstract
Environmental assessments should measure valued ecological attributes, stressors

that likely affect valued ecological attributes, and human activities that produce stressors.
Many modern diatom indicators used in bioassessments infer abiotic conditions (such as
nutrient concentration, pH, or salinity), which are usually defined as stressors. In this
research I developed a diatom-based Index of Biotic Integrity (IBI) for the Interior
Plateau Ecoregion (IPE) of the US that more closely assesses effects of human
disturbance on the biotic condition of periphyton, a valued ecological attribute. Benthic
diatoms were collected and characterized by the Kentucky Division of Water. Diatom
attributes at reference and impacted sites were selected for the IBI based on significant
differences between site groups using the Mann Whitney-U test, high separation power,
and low CV. The 7 metrics selected from a list of 59 attributes tested were the Kentucky
(KY) Diatom Tolerance Index, % species characterized as sensitive in KY, number of
distinct reference species, average similarity of species composition with reference sites,
% Achnanthes / (% Achnanthes + % Navicula), % Cymbella species of all species, and %
Navicula species of all species. IBIs were calculated by summing metrics for a site after
transforming metrics to a discrete 1, 3, 5 scale and a continuous 0-10 scale. Both scaling
systems successfully separated reference and impaired sites and IBI scores were
significantly related to measures of human disturbance in watersheds. The diatom IBI
was tested using a second dataset collected from IPE streams and processed using slightly

different methods and taxonomy. Sites were again classified as reference or impaired

87

sites based on the same criteria for the development dataset. Diatom IBI scores differed
significantly between reference and impaired streams in the second dataset and they
classified 80 % of these sites correctly. Compared to other diatom IBIs, this IBI had more
separation power than the others. Thus diatom indicators of ecological condition can
provide valuable characterizations of biotic condition and human disturbance in stream

ecosystems.

Introduction

According to a report by the United States Environmental Protection Agency (USEPA
1998), 40% of streams and rivers surveyed by states in 1996 were threatened or polluted.
Agriculture, modification of flow regimes, urban runoff, and storm sewers were
identified as primary sources of pollution (USEPA 1998), and all of these are known to
have significant effects on stream ecosystem structure and function (Karr and Chu 1999,
Carpenter et a1. 1998, Mac et al. 1998). Pressure on aquatic resources through water use
and habitat alteration is expected to increase due to increasing population (Mac et al.
1998). To maintain sustainable ecosystems, effective methods of assessing the health of
our stream ecosystems are urgently needed.

Indices of biotic integrity (IBI) have been developed to detect structural and
functional differences in valued ecological attributes between a test community and the
reference condition (sensu Cairns 1977, Karr and Dudley 1981, Barbour et al. 1999,
Stevenson and Hauer 2002). These valued ecological attributes are important for

establishing criteria to protect and restore ecosystems. IBIs usually are composed of

88

structural attributes that respond to human disturbance (i.e., metrics), such as diversity
indices, relative abundance of species, and species in taxonomic or functional groups that
are sensitive to pollution (Barbour et a1. 1999). This multimetric approach is statistically
effective for evaluating human disturbance (Fore et al. 1994) and has been used widely
for assessing and communicating biotic integrity of streams across the US (Barbour et al.
1996, Barbour et al. 1999).

Diatoms have a long history of use for environmental monitoring in Europe, North
America, and to some extent worldwide (Stevenson and Smol 2002), but development
and use of diatom-based IBIs are relatively new and restricted to the US (Bahls 1993,
KDOW 1993, Hill et al. 2000). Bahls (1993) used the sum or minimum of 4 metrics to
evaluate sites. The Kentucky Division of Water used 5 metrics to assess biotic integrity
(KDOW 1993). Hill et al. (2000) selected 10 periphyton metrics and demonstrated their
relationship to environmental variables.

Identifying and solving environmental problems requires assessing valued ecological
attributes such as biotic condition as 111 as the stressors and human activities causing
stressors. Most diatom indicators, used both individually and in an IBI, have been
developed to indicate stressors in aquatic ecosystems, such as trophic status (Pan et al.
1996), salinity (Blinn 1993), pH (Dixit et al. 1999), and climate change (Pienitz and
Vincent 2000). Bioindicators of stressor conditions are valuable complements to
measuring abiotic stressors (Stevenson and Smol 2002), however emphasis should also
be placed on developing diatom indicators of biotic condition to help assess, protect, and

restore valued ecological attributes.

89

The objectives of this study were to develop and test diatom indicators of biotic
condition and a multimetric diatom IBI. I tested the hypothesis that diatom attributes
reflecting elements of biotic condition were capable of separating streams with and
without extensive human activity in watersheds using a dataset from the Interior Plateau
Ecoregion (IPE) collected by the Kentucky Division of Water (KDOW). I selected a set
of diatom metrics for a multimetric IBI and tested the capability of the IBI to detect
effects of human disturbance with the KDOW dataset, as is usually done in IBI
assessments. To test the diatom IBI more rigorously, I then compared diatom IBI values
from reference and disturbed streams with a different dataset from the same ecoregion,
but collected and analyzed for another project using slightly different methods. Finally, I

compared the performance of existing periphyton IBIs with the test dataset.

Materials and Methods

IBI development dataset

The dataset used for developing the diatom-based IBI was generated by the biological
sampling program of the KDOW. Data from 19 impaired and 7 reference sites were
selected from the IPE (Figure 5.1). Land use in the IPE was dominated by cropland and
pasture with some woodland and forest (USEPA 1997). The geology of the IPE is mostly
composed of limestone with some sandstone and shale distributed in the south and West
part of the region (McDoIll 1986). Most soils in the region have formed from limestones

(Bailey and Winsor 1964). Mean annual rainfall ranges from 127 cm in the

90

 

u \
\
\
I . V -\
o t
O a 0... 0’0 If
0 s
a f‘ \dtbQ I ' K
(‘4!) I \‘l’. I: 6; .
I I
t ‘7
.— 3 O /
I” I
I
l‘ \ . I
‘I ‘\~’ ~ I . . I (If
‘ I ,2
’ r
s’ I _,o-
0’" ru—.———°—.—v—I. ____________ r
—————— .1
N lnterlor Plateau Ecoreglon
/\/ Kentucky boundary
I Training altee
0 Test altea
100 0 100 200 300 400 lOIomatera

——:_—_:1

Figure 5.1. Study area and locations of sampling sites in Kentucky (USA) and the Interior
Plateau Ecoregion. e indicates sites sampled in development dataset, while 0 represents
sites sampled in test dataset.

91

south to 102 cm in the north; whereas mean annual average temperature is 14 °C (USDA
1938).

KDOW data collected between 1992 and 1995 and between 20 April and 15 June
were chosen for analysis because of data availability and because spring is one of the
most biologically diverse seasons in Kentucky streams. Diatoms were collected
qualitatively by scraping algae off rocks with a toothbrush. The diatoms were acid-
cleaned, identified based mainly on Patrick and Reimer (1966, 1975), and 500 valves
were counted for each sample. Since multiple samples were collected from the same sites
during this period, one sample from each site was selected randomly for the dataset used
to develop the IBI.

Streams were designated as reference and impacted sites based on criteria in Mills et
al. (1997). Reference sites were initially selected based upon the past records of an
undisturbed riparian zone, absence of towns or communities along the stream bank,
absence of resource extraction activities in the watershed, absence of hydrologic
modification in the watershed, and absence of major sewage treatment dischargers. The
second phase of site selection was based on cross-reference with past state biological and
chemical data. The reference sites were confirmed by field reconnaissance by KDOW
staff. Additional criteria based on land use were added: > 50 % forest in the watershed
and within 100-meter buffer, and < 20 % of agriculture, urban and suburban combined in
the watershed. All other sites were classified as impacted sites.

Attributes Tested
The development of the diatom IBI was based on the recommended methods in

USEPA rapid bioassessment protocols (Barbour et a1. 1999) with minor modifications.

92

First, a large pool of diatom attributes (potential metrics) was compiled because relatively
few attributes have been evaluated in literature. Attributes were classified into 7
categories of indicators (Table 5.1): biotic indices, diversity indices, growth form,
sensitive species, tolerant species, similarity, and community structure. As described in
detail later, potential metrics were selected from a pool of biotic community attributes
when they were related to human disturbance (Karr and Chu 1999), as indicated by
Mann-Whitney U tests (Mann and Whitney 1947). Final metrics were selected for high
separation power and low coefficient of variation (CV).

The first class of attributes evaluated, biotic indices, was calculated based on the
relative abundances and autecological values of species. Species autecological
characteristics were determined by review of literature documenting habitat
characteristics in which taxa were most abundant (LoI 1974, KDOW 1993, Bahls 1993).
The Kentucky Diatom Pollution Tolerance Index (KYDPTI) (KDOW 1993) and the
Montana Diatom Pollution Index (MTDPI) (Bahls 1993), 2 indices commonly used in the
US (Stevenson and Pan 1999), are expected to decrease with impairment. They are the
sum of products of species pollution tolerance values (PTVs) (03) and their proportional
relative abundance (p,) for all species (XpiGi). These values are available from the authors
upon request.

In the diversity index category, I tested the Shannon diversity index (Shannon and
Weaver 1949), the evenness index (Pielou 1966), species richness, and the number of
diatom genera. Species diversity and evenness was hypothesized to decrease with human

disturbance (Sheehan 1984, Odum 1985).

93

Table 5.1. Algal population and community attributes, their descriptions, and their
expected response to impairment. + represents attribute increasing with impairment, —
represents attribute decreasifi with imairment, and V represents variable.

 

 

Metric Tafionom'c Description R
evel

Biotic Index

Kentucky Diatom Pollution Species Weighted average of average abundance ——

Tolerance Index (KYDPTI) and tolerance value

Montana Diatom Pollution Species Weighted average of relative abundance —

Index (MTDPI) and tolerance value

Diversity Index

Shannon Diversity Index Species A measure of evenness and richness —-

Evenness Index Species A measure of species composition evenness —

Species Richness Species Number of species in the count —

No. of Genera Genus Number of Genera —

Growth Form

% Prostrate individuals Genus Relative abundance of prostrate genera +

% Erect individuals Genus Relative abundance of erect genera —

% Stalked individuals Genus Relative abundance of stalked genera V

% Unattached individuals Genus Relative abundance of unattached genera +

% Motile individuals Genus Relative abundance of motile genera +

Sensitive Species

KY % sensitive species Species Relative abundance of KYDPTI sensitive —
species

KY no. of sensitive species Species Number of KYDPTI sensitive species —

KY % no. of sensitive species Species Percent of number of sensitive species to —
the total number of species in the count

MO % sensitive individuals Species Relative abundance of MODPI sensitive -
species

MO no. of sensitive species Species Number of MODPI sensitive species -

MO % no. of sensitive species Species Percent of number of sensitive species to —
the total number of species in the count

No. of distinct reference species Species No. of species found primarily in reference ——
sites

Tolerant Species

KY % tolerant individuals Species Relative abundance of KYDPTI tolerant +
species

KY no. of tolerant species Species Number of KYDPTI tolerant species +

KY % no. of tolerant species Species Percent of number of tolerant species to +
the total number of species in the count

MO % tolerant individuals Species Relative abundance of MODPI tolerant +
species

MO no. of tolerant species Species Number of MODPI tolerant species +

MO % no. of tolerant species Species Percent of number of tolerant species to +

the total number of species in the count

94

Table 5.1. Continued.

 

 

Metric Taxonom'c Description R
level
Similarity
Average similarity with Species Average of Jaccard similarity to all —

reference sites
% reference species

No. of distinct reference species

Community Composition

% Achnanthes minutissima
individuals

% Achnanthes/
(Achnanthes+Navicula)

% Cymbella/
(Cymbella-l-Navicula)

% Dominant species

% Achnanthes individuals

% Amphora individuals

% Cocconeis individuals

% Cyclotella individuals

% Cymbella individuals

% F ragilaria individuals

% F rustulia individuals

% Gomphonema individuals

% Navicula individuals

% Rhoicosphenia individuals

% Surirella individuals

% Synedra individuals

% No. of Achnanthes species

% No. of Amphora species

% No. of Cocconeis species

% No. of Cyclotella species

% No. of Cymbella species

% No. of F ragilaria species

% No. of F rustulia species

% No. of Gomphonema species

% No. of Navicula species

% No. of Rhoicosphenia species

% No. of Surirella species
% No. of Synedra species

Species
Species

Species
Genus
Genus

Species
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus
Genus

reference sites

Percent of species found in reference sites
No. of species found primarily in reference

sites, not in impaired sites

Relative abundance of Achnanthes
minutissima

The ratio of Achnantes to Achnanthes and

Navicula combined

The ratio of Cymbella to Navicula

and Cymbella combined

Relative abundance of dominant species
Relative abundance of Achnanthes
Relative abundance of Amphora
Relative abundance of Cocconeis
Relative abundance of C yclotella
Relative abundance of Cymbella
Relative abundance of F ragilaria
Relative abundance of F rustulia
Relative abundance of Gomphonema
Relative abundance of Navicula
Relative abundance of Rhoicosphenia
Relative abundance of Surirella
Relative abundance of Synedra

Percent of No. of Achnanthes species
Percent of No. of Amphora species
Percent of No. of Cocconeis species
Percent of No. of Cyclotella species
Percent of No. of Cymbella species
Percent of No. of F ragilaria species
Percent of No. of F rustulia species
Percent of No. of Gomphonema species
Percent of No. of Navicula species
Percent of No. of Rhoicosphenia species
Percent of No. of Surirella species
Percent of No. of Synedra species

95

Diatom growth forms represent morphological adaptations to environmental
conditions and community successional stages (Peterson and Stevenson 1992, Peterson
1996) (Table 5.2). Metrics in this category were all based on genus-level identification
because species in a genus usually have the same morphological adaptations. Motile
diatoms have a raphe, can move around in silt particles, and may show increased
abundances in streams with high siltation. Erect and stalked diatoms are susceptible to
severe hydraulic disturbance. Conversely, prostrate diatoms grow flat within the
boundary layer and may indicate heavy grazing, hydraulic disturbance, or early
developmental status of a diatom community (Stevenson 1996). Stalked individuals are
believed to be adapted to overgrow and outcompete prostrate diatoms for nutrients and
light (Hudon and Legendre 1987). Finally, high abundance of unattached growth forms
can indicate stagnant water.

Sensitive species occur more frequently in reference than polluted streams and may
represent native species. Pollution tolerant taxa occur in relatively low numbers in
reference streams and thrive in polluted streams. I hypothesized that sensitive species
would decrease and tolerant species would increase with human disturbance (sensu
Odum 1985, Rapport and Whitford 1999). Species sensitivies and tolerances to pollution
were characterized based on assigned PTVs in species tables of the KYDPTI and MTDPI.
Species with the highest PTVs were assigned as sensitive species. Similarly, tolerant
species were those with the loIst PTVs. Percent relative abundance of individuals,
number of species, and % of species with specific environmental preferences are 3 ways

to calculate metrics relating species composition of assemblages

96

Table 5.2. Diatom genera growth form and motility. P = prostrate, E = erect, S = stalked,
U = unattached, V = variable, M = highly motile diatoms.

 

 

Genera Growth form Genera Growth form
Achnanthes P Gyrosigma P, M
Amphipleura P Hannaea E
Amphora P Hantzschia P, M
Anomoeoneis P Melosira U
Asterionella U Meridian E
Biddulphia E Navicula P
Caloneis P Neidium P
Cocconeis P Nitzschia P, M
Cyclostephanos U Opephora E
Cyclotella U Pinnularia P
Cymatopleura P, M Rhoicosphenia S
Cymbella S Rhopalodia P
Denticula P Simonsenia P
Diatoma U Stauroneis P
Diploneis P Stenopterobia P, M
Epithemia P Stephanodiscus U
Eunotia E Surirella P, M
F ragilaria V Synedra E
F rustulia P Tabellaria U
Gomphoneis S Thalassiosira U
Gomphonema S

 

97

and species autecologies (Barbour et al. 1999). All 3 were calculated for both sensitive
and tolerant species indices.

I hypothesized that similarity in species composition between reference and impaired
sites would decrease with increasing human impairment (Sheehan 1984, Odum 1985,
Rapport and Whitford 1999). Similarity with reference sites was calculated using the
average of Bray-Curtis dissimilarity index (Bray and Curtis 1957) between a test site and
all reference sites. The % reference species index was the relative abundance of all
species observed in all reference sites. In contrast, the number of distinct reference
species was the number of species that appeared primarily in reference sites and not in
impaired sites as identified by indicator species analysis (Dufrene and Legendre 1997,
McCune and Grace 2002). In indicator species analysis, indicator values (IV) were
calculated for each species based on the exclusiveness of species occurrences in a group
(the relative abundance of species in that group versus all sites) and the probability that
the species would be observed at sites in that group. The maximum IV (IVmax) was then
determined for each species to assign them to a group. A Monte Carlo method was used
to determine the statistical significance of IVmax and designation of a species as an
indicator for that specific group.

Species and genus-level community structure were hypothesized to change with
human disturbance (Sheehan 1984, Rapport and Whitford 1999). Environmental
responses of taxonomic groups may be regulated by genetic and phenotypic similarities
among species within genera, families, and higher-level taxa (e. g. % Ephemeroptera,
Plecoptera, Trichoptera). Achnanthes minutissima (Ktltzing) is a very common species in

samples and is expected to decrease with human disturbance (Stevenson and Bahls 1999).

98

Genus-based attributes summarize the average response of con-generic species to human
disturbance. The proportion of all individuals that are in one genus can be used as a
genus-level attribute of biotic condition. For genera with many species in a sample, the
percentage of species in a sample from that one genus can also be used as an attribute.
Achnanthes (Kutzing), Cocconeis (Ehrenberg), Cymbella (Agardh), Fragilaria (Lyngbye),
Frustulia (Rabenhorst), Gomphonema (Agardh), Rhoicosphenia (Grunow), and Syn—edg
(Ehrenberg) were considered to be clean water genera (Bahls 1993). Amphora

(Ehrenberg). Navicula (Bory), Nitzchia (Hassall), and Surirella (Turpin) were thought to

 

be neutral genera (Bahls 1993). Achnanthes and Navicula classification was based on
Achnanthes and Navicula sensu lato (i.e., as defined before recent separation of some
species into other genera (Round 1990)). Percent Achnanthes was divided by the sum of
percentage of Achnanthes and Navicula in the denominator to reduce the variance of the
attribute. Percent Cymbella was similarly standardized by the sum of percent Cymbella
and Navicula.
Metric selection

Attributes were selected and excluded based on the following sequence of analyses.
Attributes with medians of 0 for both reference and impaired sites were eliminated first
because low values would prevent the identification of differences. The ability of
remaining attributes to separate reference and impaired sites was evaluated with the
Mann-Whitney U test (Mann and Whitney 1947). When tests showed statistically

significant differences (a < 0.05) between site groups, potential metrics were identified.

Separation power of these potential metrics was evaluated with box plots, where

separation power was defined by the overlap of box plots for reference and impaired sites

99

(Barbour et al. 1996, Barbour et al. 1999). The box marks the range between the 25th and
the 75th quartiles of an attribute. The highest power of 3 was assigned when there was no
overlap between boxes of reference and impaired sites. A separation power of 2, 1, and 0
were assigned respectively when interquartile ranges overlapped but did not reach the
medians, when only one median was within the interquartile range of the other box, and
when both medians were within the range of the other boxes.

The final metrics were selected from the set of potential metrics based on separation
power and CV (Table 5.3). Metrics with separation power < 3 and CV > 1 were excluded.
Similar metrics, such as % sensitive species and % number of sensitive species, were
evaluated to select one based on highest separation power and loIst CV.

Scaling systems and IBI calculation

Two scaling systems were used to normalize the range of metrics. The l, 3, 5 scaling
system has been commonly used with fish and macroinvertebrate IBIs (Karr 1981, Keran
and Karr 1994, Barbour et al. 1999). Metrics decreasing with impairment were scored 5 if
values exceeded the 25th percentile of reference sites, 3 if values were between the 25th
percentile of reference and the 50th percentile of impaired sites, and 1 if the metric value
was less than the 50th percentile of impaired sites. Metrics increasing with impairment
were assigned 5 if the metric value was < 75th percentile of reference sites, 3 if the value
was between the 75th percentile of reference and the 50th percentile of impaired sites,
and 1 if the value was above 50th percentile of impaired sites.

A 0-10 system (Hill et al. 2000) was included to compare with the 1, 3, 5 system. For
metrics decreasing with impairment, the score was calculated by dividing the metric

value for each site by the 90th percentile from reference sites, then multiplied by 10. The

100

90th percentile was used to avoid the distortion of scores by the possible extreme
maximum value. For metrics increasing with impairment, the metric value was divided
by the 90th percentile of impaired sites, subtracted from 1, and then multiplied by 10. IBI
values were the sum of metric scores.

The relationship between the IBI and land use in watersheds was examined by linear
regression. Land use coverage was obtained from the US Geological Survey with a
resolution of l:250,000. The coverage was classified according to Level II categories of
the land use and land cover classification system (Anderson et al. 1976). The watershed
was defined as the hydrologically contributing land surface upstream of the sampling site.
Watershed boundaries were derived from 30 by 30 meter digital elevation grids through
digital hydrological modeling with Arc/Info software (Version 7; ESRI, Inc., Redlands,
CA). Percent land use was square-root transformed to normalize the data.

Independent testing of metrics and IBI

The IBI was tested with a dataset from a project designed to evaluate relations among
stream algae, invertebrates, and nutrients (SAIN Project). Thirty SAIN sites from the IPE
were sampled during spring 1996. Diatoms were cleaned with nitric acid. Six hundred
valves were counted and identified following the taxonomy of Kramer and Lange-
Bertalot (1986, 1988, 1991a, 1991b). SAIN reference sites were defined using the same
criteria as used for KDOW reference sites.

Tests were conducted to determine if the developed IBI could distinguish reference
and impaired sites in the SAIN dataset. From the development dataset, both the 25th

percentile of IBI scores at reference site and the average of IBI scores for the 75th

101

percentile of the impaired sites and the 25th percentile of reference site were subjectively
set as the minimum values for reference sites.

I also tested diatom indices of Hill et al. (2000), KDOW (Kentucky Division of Water
1993), and Bahls (1993) with the SAIN test dataset to determine the relative performance
of the IBI developed in this study. Tests were performed using box separation ratios,
which were the distances between boxes divided by the medians of IBI scores from
reference sites. Five diatom metrics used by Hill at al. were included in our metric
selection process (diatom taxa richness, % dominant taxa, % acidophilic genera, % motile
genera, and % eutrophentic genera). All 5 metrics of KDOW (diatom taxa richness,
Shannon diversity, KYDPTI, percent similarity with reference sites, and % sensitive
species) and all of Bahl’s metrics (Shannon diversity index, diatom pollution index,

siltation index, and similarity index) were included in our tests.

Results

After initial evaluation, 29 of 55 attributes were classified as potential metrics (Table
5.3). Percent Cyclotella and % Frustulia attributes were eliminated because both had
medians of 0 for reference and impaired sites. Twenty-four metrics were rejected due to
non-significant differences between reference and impaired sites based on Mann-Whitney
U tests. For example, species richness and number of diatom genera were unable to
separate reference from impaired sites. Additionally, Shannon diversity and evenness
indices were eliminated because both increased with impairment, which contradicted the

commonly predicted responses.

102

Table 5.3. Selected algal metrics and their separation power, coefficient of variation,
response to impairment, and reasons for exclusion. + represents metric increasing with
impairment caused by human disturbance, — represents metric decreasing with impairment.

 

 

Meme 32%?“ CV $5.933? 53:23.3“
KYDPTI 3 0.15 -

MTDPI 3 0.16 __ High correlation
KY % sensitive species 2 1.32 — Low power

KY % tolerant species 2 0.9 + Low power

MT % tolerant species 2 1.08 + High CV

KY no. sensitive species 3 0.87 — High correlation
KY % no. sensitive species 3 0.81 —

MT no. tolerant species 2 0.43 + LOW power

MT % no. tolerant species 1 0.41 + LOW power

No. distinct reference species 3 0.6 —

Shannon Diversity index 3 0.15 + Opposite response
Evenness index 3 0.1 + Opposite response
% Reference species 2 0.1 — Low power
Average similarity to reference 3 0.26 -

% Achnanthes minutissima 3 1.32 - High CV

% Achnanthes /(Achnanthes+NavicuIa) 3 0.9 —

% Cymbella /(Cymbella+Navicula) 3 1.4 — High CV

% Achnanthes 2 1.0 - Low power

% Amphora 2 1.0 + Low power

% Cocconeis 3 1.56 + High CV

% Cymbella 3 1.64 — High CV

% Gomphonema 2 1.03 + Low power

% Navicula 3 0.79 + High correlation
% Synedra 3 1.29 — High CV

% No. Cocconeis species 3 0.63 + Few species

% No. Cymbella species 3 0.81 -

% No. Navicula species 3 0.79 +

% No. Synedra species 3 1.06 - High CV

% Erect individuals 3 1.29 — High CV

 

103

Potential metrics were screened based on separation power, CV, and redundancy.
Kentucky (KY) % sensitive species, KY % tolerant species, Montana (MT) number of
tolerant species, MT % number of tolerant species, % reference species, % Achnanthes,
% Amphora, and % Gomphonema were removed because of low separation power. MT
% tolerant species, % Achnanthes minutissim_a, % Cymbella / (% Cymbella + %
Navicula), % Cocconeis, % Cymbella, % Sm, % number Sm species and % F
erect individuals were eliminated due to CV > 1. Both MTDPI and KYDPTI performed
equally 111, but MTDPI was removed because of its redundancy with KYDPI and because

the KYDPI was based on the accumulated knowledge of regional flora. KY % number of

 

sensitive species was selected over KY number of sensitive species because of its loIr CV
and redundancy. Percent number of Cocconeis species was excluded because only a few
species were present in the area. Percent number of Navicula species was selected over %
Navicula because I expected that % number of Navicula is more transferable among
regions.

Seven metrics were selected for the IBI. The KYDPTI, KY % number of sensitive
species, number of distinct reference species, average similarity with reference sites, %
Achnanthes / (Achnanthes + Navicula), and % number of Cymbella species decreased
with impairment. Percent number of Navicula species increased with impairment (Table
5.4). Average similarity with reference sites had the smallest distribution range, while %
Achnanthes / (% Achnanthes + % Navicula) had the widest range. Box plots for each
metric showed good separation between reference and impaired sites (Figure 5.2).

IBI, based on the seven metrics with both the 1, 3, 5 and 0-10 scaling systems, clearly

separated quartiles of reference sites from quartiles of impaired sites (Figure 5.3). The

104

Table 5.4. Scoring range and 90th percentile and minimum values for 10 selected metrics
of IBI.

 

 

 

Score Statistics

Metrics 5 3 l min 90th
KYDPTI >28 2.8-2.3 <2.3 2.1 3.5
KY % No. sensitive species >16.2 l6.2-S.7 <57 0 21.2
No. distinct reference species >5 5-3 <3 0 8
Average similarity with reference sites >0.46 0.46-0.35 (0,35 0.24 0,54
% Achnanthes /(Achnanthes+Navicula) >279 27.9-11,7 <1 1_7 0,4 85
% No. Cymbella species >150 15.0-30 <30 0 18.2
% No. Navicula Species <25.0 25.0-32.0 >320 9.1 38.5

 

105

 

 

4.0 100

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

9 80‘
... 30 2‘
E ' § 3 60‘
Q o
s E s, I .
>4 2.0 g -
. < 200 !
1.0 ' O
,6 30 8 20
U —i
O U
é
é _.
.5 T) I
g g 10«
e' 10' 6‘
t:
s 2
>1 0
>4 0 s O
10 e 50
.9 8
8 8 a 40‘
90‘ cc: ;
E 6‘ E 301 %
H e;
‘5 ' 10.
0' 2J 2
Z 50 0 T
0 Ref Imp
0.6
b
0 Q
5 0'4 -
330 0.3
e
“>’ 0.2 _J_
< 0.1

 

 

 

Ref Imp

Figure 5.2. Box plots of diatom metrics for reference (Ref) and impaired (Imp) sites of
development dataset. Boxes are interquartile ranges (between 25th percentile and 75th
percentile); middle lines are medians; and whiskers are 1.5 interquartile ranges beyond
the boxes.

106

 

 

 

 

 

 

 

 

 

 

 

40
30 I
8
o
O
2 201 l
N
‘5
E-4
10‘ | ~
0 - -
Reference Impaired
B.
80
05 60‘ I
§ 1%
T‘s 40‘ T
o
[-4 |_,,,_, “4
20‘ 1
0

 

 

 

Reference Impaired

Figure 5.3. Box plots of total IBI scores of reference sites versus impaired sites from
development dataset. A. 1-3-5 scaling system. B. 0-10 scaling system.

107

averages of the 75th percentile of the impaired sites and the 25th percentile of reference
site were 39.5 and 25.0 for the 0-10 system and 1,3,5 system, respectively. The 25th
percentiles of reference sites were 47.5 and 29 for the 0-10 system and 1,3,5 system,
respectively. The IBI increased significantly with % forest cover in watersheds in the
development data set (linear regression, r2 = 0.67 and p < 0.001) (Figure 5.4). The
correlation between IBI and % forest cover was higher than all component metrics except
% number of Cymbella species (Table 5.5). KY % number of sensitive taxa was the
second highest metric correlated with % forest cover and % number of Navicula species
was the third (Table 5 .5).

After application of the selected IBI to the SAIN dataset, IBIS using both scaling
systems were significantly different between reference and impaired sites with a high
separation power of 3 (Figures 5.5, 5.6). The average of the IBI scores for the 75th
percentile of the impaired sites and the 25‘h percentile of reference site was used as a
criterion to distinguish reference and impaired sites because it correctly classified more
sites as reference and impaired than a criterion based on the 25th percentile of IBI scores
at reference sites. Many reference sites were misclassified as impaired when the 25th
percentile of the reference condition was used as a criterion. Only 18 of 30 sites were
correctly classified under the l, 3, 5 scoring system. With the 0-10 system, 20 of 30 sites
were correctly categorized. With the average of the 75th percentile of the impaired sites
and the 25th percentile of reference site as the criterion, the 1, 3, 5 system correctly
classified 25 of 30 sites. Three reference sites were classified as impaired sites. The 0-10

system correctly classified 24 of 30 sites with 4 impaired sites classified as reference sites.

108

 

 

IBI score

 

0 20 40 60 80 1 00

% Forest

Figure 5.4. Relation between index values of biotic integrity from the KDOW data set
and % forest cover in watersheds (adjusted r2 = 0.616, p < 0.001). Percent forest was
arcsine squared-root transformed.

109

Table 5.5. Pearson correlation coefficients between IBI metrics and % forest cover in
watersheds. * denotes statistical significance, 0 < 0.05. Ach. is short for Achnanthes; Nav.
for Navicula.

 

 

 

 

Metrics Pearson r p

IBI total score 0.82 0000*

KYDPTI 0.69 0000*

KY % No. sensitive species 0.8 0.000*

No. distinct reference species 0.64 0.000*

gleesrage srnulanty w1th reference 052 0.00 6* p—
Ach. / (Ach. + Nav.) 0.55 0004*

% No. of Cymbella species 0.84 0000*

% No. of Navicula species 0.69 0.000*

110

 

 

4.0 100

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3.5 , ’5 80 T .
E
E 3.0 .3" 60
M 2.5 E 40
2.0 59 20 ‘ I
1.5 0
,8 40 8 40
a e
3 30 _~_ ii 30 °
.2 g
g 20‘ ._ . .3 20.
' U
8 10‘ I- o- 10‘
ee — z .
g 0 T es 0
0. 7 a 40
‘2 61 . 8
i 5 8‘ 30
: 4 2 20 -
8 3 E a 1 -
03 2
“U 2 —‘ O 10‘
2% (1) —* g 0 -
Ref Imp
0.6
29
'5 ”-5 é I
E 0.4 '—
80 0.3 ' ~
‘3 I
8 0.2
< 0.1 f _
Ref Imp

Figure 5.5. Box plots of diatom metrics for reference (Ref) and impaired (Imp) sites of
test dataset. Boxes are interquartile ranges (between 25th percentile and 75th percentile);
middle lines are medians. e = outliers.

lll

 

 

 

 

 

 

 

 

 

 

 

 

40
o 30' '
‘65 ' '
o _L_
Z.” 201 z -
Cd
‘5
.. L—l
10 I
O 1 .
Reference Impaired
B.
60
50 I l o
93 40' ~
0 I
O
3 30' if i
S H
o
H 20' *
10 _J_
O

 

 

 

Reference Impaired

Figure 5.6. Box plots of total IBI scores of reference sites versus impaired sites from test
dataset. A. 1-3-5 scaling system. B. 0-10 scaling system.

112

 

All 3 tested IBIs from the literature were significantly different between reference and
impaired sites, however, box separation ratios of these 3 IBIs were lower than the IBI
developed in this project (Table 5.6). The Shannon diversity index and species richness,
which were included in KDOW’s (1993) IBI, gave the opposite of expected responses to
human disturbance and showed higher values for impaired sites. The diatom pollution
tolerance index and similarity with reference sites both had separation power of 3. Except
for the Shannon diversity index, the metrics in Bahls’s (1993) IBI (similarity to reference
sites, siltation index, and MTDTI) were able to separate reference from impaired sites.
Two metrics from the 5 Hill’s IBI metrics (Hill et al. 2000) (taxa richness and acidophilic
genera) had separation power of 1. Percent dominant taxa had higher values for reference
sites than for impaired sites. Eutraphentic genera and motile genera had high separation

power (3).

Discussion

The diatom IBI developed for the IPE of Kentucky was able to separate reference
from impaired sites in both the dataset in which it was developed and in a test dataset
generated with slightly different sampling and analytical methods. IBI scores were a good
indicator of human disturbance regime in watersheds as measured by % forest cover. The
diatom IBI developed for the IPE was also more powerful than 3 pre-existing indices
when applied to streams in the IPE.

Several misclassified cases during tests of our IBI with a new dataset were under-

appraisals of biotic integrity. The metric for number of distinct reference species failed to

113

 

Table 5.6. Box separation ratios of developed IBIs, Bahls IBI, KY diatom IBI, and Hill’s
IBI on test dataset.

 

 

IBI Box separation ratio
O-lO system 0.14

l, 3, 5 system 0.15

Balhs IBI 0.03

KY diatom IBI O

Hill's IBI 0

 

114

separate reference and impaired sites in the test dataset using both 1, 3, S and 0—10
systems. Samples in the KYDOW and SAIN projects were analyzed by different
technicians, who followed different taxonomic references. Rapid changes in
nomenclature present a challenge to ecologists (Stoermer 2001, Kociolek and Stoermer
2001). Taxonomic consistency should improve metrics that use a few species that are
sensitive to different identification. Cross-referenced taxonomic databases with names,
pictures, and literature citations can improve taxonomic consistency and can be
established and regularly updated through venues like the taxonomy workshops
sponsored by USGS National Water Quality Monitoring Program. Despite potential
taxonomic discrepancies between projects, our IBI separated impaired from reference
sites with higher statistical significance and consistency than other algal IBIS.

Our misclassified cases included several impaired sites classified as reference sites,
which reflects a type II statistical error. Type II errors with multimetric IBIS can be
checked by 2 methods. First, they can be checked if a certain metric has a value below a
critical value or percentile. For example, with the l, 3, 5 system I could correct 2
misclassified sites if I checked for KYDP’I‘ I values below 3. Second, I could tally the
number of selected metrics below certain values or percentiles for each site. For example,
I could correct classification of 2 sites if I checked for sites with 3 or more metrics with
low scores (e.g., score 1 in 1, 3, 5 system).

Another type of error was characterizing unimpaired sites as impaired. Some highly
correlated metrics for the test dataset may be the cause of this failure. This type of error is

difficult to check, but may be discovered with diagnostic metrics if no specific stressors

115

exceed criteria. Multiple lines of evidence, such as fish and invertebrate data, could also
help provide correct classification.

I included both species and genus-based attributes in our metrics. The advantage of
genus-based metrics is their ease of use because of easier identification than species. The
limitation of genus-based metrics is the masking of individual species responses.
Congeneric species may have different environmental optima, sensitivity, and tolerance
levels. Hill et al. (2001) found no significant loss of power for several genus-based
metrics of environmental conditions when compared to species-based metrics, but found
fewer environmental conditions can be inferred by genera than by species. Genus-based

metrics such as % number of Cymbella species were good indicators of biotic condition

1.

in the IPE. Genus-based metrics may be most suitable for specific types of aquatic
habitats and small regions, rather than for national or international use because the
number of species per genus is likely positively related to the size and habitat diversity of
the region. The likelihood of all taxa in a genus responding similarly to human
disturbance should be inversely related to the sizes of regions studied. Thus, genus-based
metrics may not be as transferable among regions as species-based metrics. For example,
% Achnanthes decreased with impairment in our study; hoIver, Kelly (1998) classified
Achnanthes as a median nutrient genus. Genus based metrics do have potential for
application, particularly for some stressors and within regions where sets of species
within genera respond similarly to human disturbance (Hill et al. 2001).

Species diversity is commonly used as a metric in diatom assessments, even though it
is poorly correlated to environmental change. Archibald (1972) and van Dam (1982)

found that Shannon diversity index increased with impairment as in our study.

116

Conversely, Stevenson (1984) found diversity indices changed unpredictably with
increasing pollution. However, diversity indices are still used in several IBIs (Bahl 1993,
KDOW 1993, Hill et al. 2001). Response of diversity indices to stressors depends on the
type and severity of pollution and may be too unpredictable for general assessment.
Although high diversity is commonly considered a valued ecological attribute, the use of
diversity in algal bioassessment is cautioned. Species numbers in a 600-va1ve count
poorly reflect the number of taxa in a habitat (Stevenson and L01 1986). In addition, high
numbers of species may not be characteristic of natural conditions, such as those in low
nutrient habitats (McCormick and O’Dell 1996).

I recommend using a 0—10 scaling system as proposed by Hill et al. (2000). No
distinct difference was observed in correct assessment of streams for the 1, 3, 5 and 0-10
scaling systems, however the 0—10 system was analytically simpler and easier to
communicate with public. The choice of categorical scaling systems (1, 2, 3, and 2, 4, 6)
is often considered arbitrary.

The advantage of a multimetric system over a univariate assessment is the assumed
greater certainty of detecting impairment and transferability among habitats in a region
and potentially between regions. The multimetric diatom IBI had a higher separation
power and lower CV in our test dataset than all metrics except the KYDPTI, which had
similar power and CV with the IBI. Multimetric systems can provide more thorough
integration of the overall system condition (Gerritsen 1995) because no one metric can
respond to all types of human impact on ecological systems (Karr 1999). They can

account for erroneous responses of a few metrics in assessment projects and may include

117

metrics related to multiple ecological attributes valued by both decision makers and
stakeholders.

In the future, I recommend developing an approach in which greater weight is given
to one or more metrics, in addition to the IBI scores. Examining the response of IBI to
human disturbance can mask the response of individual metrics because of the averaging
feature of a multimetric system (Suter 1993). Fewer streams would have been
misclassified if I had used more than the multimetric IBI in the assessment. Individual

criteria could be set based on a metric indicating a valued ecological attribute that has

 

sufficient public or ecological concern that it merits specific management attention. For

. 5'.

example, the minimum scores of 3 or the 75th percentile for metrics highly correlated
with % forest cover, such as KYDPT I and % number of sensitive taxa, can be used as
criteria for protection. The number of metrics not meeting pre-defined benchmarks could
also be used as another criterion. Different management actions could be triggered by
different kinds of violations, ranging from more thorough study to implementing
restoration or protection plans. Scores of multimetric indices should be used primarily to
summarize the overall ecological conditions and communicate results.

In conclusion, the diatom-based IBI for the [PE derived from a development dataset
classified reference and impaired sites in an independent dataset with higher accuracy
than other multimetric indices. The IBI values highly reflected the degree of human
disturbance in watersheds. Ecologically significant metrics should be examined
individually and should have minimum acceptable criterion. More extensive study of
periphyton assemblages with all algae and algal functional groups could provide indices

of periphyton function. I need to more directly address the valued ecological structures

118

and functions of assemblages and develop means of assessing those attributes.
Distinguishing between metrics of valued ecological attributes and metrics diagnosing
stressors provides management with specific information, targeting reasons, and

strategies for action.

119

 

CHAPTER 6
CONCLUSION

Various ecological attributes respond to pollutants imposed by human activities as
summarized by Sheehan (1984), Odum (1985), Rapport et al. (1985), Schindler (1987),
and Rapport and Whitford (1999). These attributes have wide literature support and
provide biological and ecological basis for bioassessment. Therefore, they should be used
to develop bioassessment tools with different kinds of organisms and ecosystems.

In chapter 2, 6 hypotheses of community structural response to stress were tested.

These 6 attributes are valued ecological attributes. From correlation analyses, land use

 

fl
idling

types had significant relationships with environmental variables. Overall, these algal
attributes responded to land use and environmental variables under low-level human
disturbance except for species diversity. Algal species composition and dominant species
changed with the human disturbance gradient (HDG). Average diatom biovolume was
positively correlated with the HDG. Relative abundance of trophic sensitive species
significantly decreased with the HDG, while % trophic tolerant species significantly
increased with the HDG. Percent native species was negatively correlated with the HDG,
while % trophic sensitive and native species had a higher correlation with the HDG than
% native species. Both trophic sensitive and tolerant species were significantly correlated
with HDG, Cl', and conductivity, but not with nutrient levels. Nutrient inputs, which are
often correlated with elevated C1" and conductivity, were reduced by biotic uptake. Hence,
diatom species were a better indicator for human disturbance than nutrient levels in

streams.

120

Elli

em

Rock is often the main sampling substratum for algal study and bioassessment. In
chapter 3, the sensitivities of epilithic, epiphytic, and epipelic diatom assemblages to an
environmental gradient were compared. Species composition of epipelon was most
different from epilithon and epiphyton. Each assemblage has its own characteristic
species list as explored by indicator species analysis. Epiphyton had the shortest species
list from counts, while epilithon and epipelon had about 220 species in the counts.

Epiphytes had the most % species variance explained and % species-environment relation

 

variance explained from CCA, while epipelon was ranked 2nd. Epiphytes also had the

smallest beta diversity and fastest species turnover along the MEG, while epipelon and

 

epilithon had about the same beta diversity and species turnover rate. Further tests
showed that epiphyton was more sensitive than epilithon and epipelon when MAIA
trophic index applied to the 3 assemblages. Inter-habitat diversity was not homogenized
along the MEG. Natural habitats, besides rock, deserve more study for biodiversity and
bioassessment research.

Understanding regional nutrient limitation is another kind of assessment that can
contribute to stream management. Small-scale nutrient enrichment experiments
conducted in 16 streams showed no consistent regional pattern in nutrient limitation.
However, species responded to nutrient enrichment as shown by the significant
differences between treatments and controls on their detrended correspondence analysis
(DCA) scores. The plot of first 2 axes of DCA also revealed that nutrient enrichments did
not change species composition on clay pots similar to those assemblages in high nutrient
streams. Community assemblages from treatment clay pots in the same site were more

similar than assemblages on clay pots from the same treatments across sites. Comparison

121

between the rankings of species response to nutrient enrichments and rankings of species
environmental optima indicated no consistent correlation pattern. Also no correlation was
found between rankings of species response to nutrient enrichments and rankings of
species environmental optima from another regional survey and between rankings of
species response to nutrient enrichments and species indicator values in van Dam’s
trophic index. Therefore, extrapolation of experimental results from small-scale
experiments to large scales needs to consider regional species pools and regional species
autecology.

In chapter 5, a diatom index of biotic integrity (IBI) was developed for the Interior
Plateau Ecoregion in Kentucky. Based on IBI development protocols (US EPA 1999) and
ecological attributes tested, 7 metrics were selected: the Kentucky (KY) Diatom
Tolerance Index, % species characterized as sensitive in KY, number of distinct reference
species, average similarity of species composition with reference sites, % Achnanthes/
(% Achnanthes + % Navicula), % number of Cymbella species, and % number of
Navicula species. IBIs were calculated by summing metrics for a site after transforming
metrics to a discrete 1, 3, 5 scale and continuous 0—10 scale. Both scaling systems
successfully separated reference and impaired sites and IBI scores were significantly
related to measures of human disturbance in watersheds. The diatom IBI was tested using
a second dataset collected from IPE streams and processed using slightly different
methods and taxonomy. Sites were again classified as reference or impaired sites based
on the same criteria for the development dataset. Diatom IBI scores differed significantly
between reference and impaired streams in the second dataset and they classified 80 % of

these sites correctly. Compared to other diatom IBIs, this IBI had more separation power

122

than the others. Thus diatom indicators of ecological condition can provide valuable
characterizations of biotic condition and human disturbance in stream ecosystems.
Future directions

Community structural attributes have been supported as sensitive in this study.
However, the temporal variability and uncertainty of these attributes to human
disturbance have not been explored. The understanding of uncertainty of these attributes
can provide a foundation for risk analysis in decision-making processes. Hence, the risks
associated with different levels of biocriteria can be communicated to managers, stake-

holders, and the public.

 

Despite the sensitivity of community structure to human disturbance, ecosystem L9,
functions are also valued ecological attributes and related to ecosystem services directly.
Structural and functional ecological attributes represent different ecological and societal
values and can be used as different endpoints for assessment and management. Therefore,
both need to be studied together to have direct comparison for future research.
Changes of community structure can be the product of the interactions between
local and regional species pools. Experiments for evaluating community structural
changes need to consider the impacts of species from the regional species pool. The
knowledge of species pool involved and species autecology can also facilitate the realism
of ecological forecasting and prediction. Understanding of roles of species pool may

provide linkages to current knowledge in ecology.

123

APPENDICES

 

124

Appendix A. Species optima and ranks for TN (mg/L), TP (pg/L), Si (mg/L), and Cl
(mg/L). Species optima were calculated from species on control clay pots and appeared in
> 3 sites. opt is the short for optima.

 

Species TN opt TN rank TP opt TP rank Si opt Si rank C1 opt Cl rank

ACbiasol 0.51 13 1.02 13 2.94 1 1 2.48 38
ACclevei 0.65 38 1.07 22 2.99 23 2.27 23
ACconspi 0.73 50 1.19 58 3.21 59 2.70 52
ACexigua 0.58 26 1.22 64 3.30 63 2.32 25
AClauenb 0.75 55 1.20 61 2.92 6 2.00 8

ACminuti 0.69 45 1.09 30 3.08 45 2.47 34
ALpelluc 0.68 43 1.19 57 3.41 65 2.81 53
AMlibyca 0.70 47 1.17 54 3.19 56 2.36 28
AMpedcls 0.68 44 1.10 31 3.04 36 2.65 48
CCneothu 0.61 32 1.07 24 3.02 31 2.48 37
CCplaeug 0.61 30 1.08 27 3.04 37 2.48 36
CCplalin 0.65 37 1.17 52 3.12 47 2.39 31
CCpsethu 0.55 23 1.06 20 2.99 21 2.17 17
CMaffins 0.46 3 0.94 3 2.95 12 2.31 24
CMcesati 0.49 6 1.00 1 1 3.05 40 2.50 39
CMcistul 0.44 2 0.94 4 2.97 19 2.01 9

CMmicroc 0.52 16 1.13 39 3.18 55 2.47 35
CMminuta 0.51 14 1.08 26 2.97 18 2.33 26
Cmsilesi 0.65 39 1.25 65 3.62 68 2.70 51
Cydistin 0.49 10 1.04 16 3.05 39 2.81 54
Datenue 0.41 1 1.06 18 2.83 1 1.83 3

Frbrevis 0.55 24 1.07 23 2.94 10 2.10 14
FRcapgra 0.59 28 1.04 15 2.90 4 1.96 7

FRcaprum 0.50 12 1.15 45 2.92 9 2.08 1 1
FRcapvau 0.53 19 1.00 12 2.89 3 2.02 10
FRconstr 0.52 15 1.10 32 2.92 8 1.92 4

FRconven 0.64 35 1.07 25 2.99 24 2.17 19
FRdelcat 0.52 17 0.97 9 2.92 5 1.81 1

FRlepdub 0.55 25 1.04 17 2.96 16 2.17 18
FRnanana 0.53 20 1.22 63 3.29 62 2.15 16
FRparasi 0.96 66 1.18 56 3.00 26 2.54 44
FRpinnat 0.77 56 1.12 37 2.96 17 2.19 20
FRtenera 0.53 18 0.95 6 3.07 43 2.26 22

 

125

Appendix A. Continued.

 

Species TN opt TN rank TP opt TP rank Si opt Si rank Cl opt Cl rank

GOgracil 0.64 34 1.27 67 3.60 67 2.52 40
GOkobaya 0.73 51 1.21 62 3.04 35 3.24 62
GOmicrop 0.48 5 0.86 1 3.01 30 2.54 45
GOolivcm 0.61 31 1.13 41 2.98 20 2.62 47
GOparvul 0.53 22 0.97 10 3.08 44 2.53 43
GOpumilu 0.46 4 0.96 7 2.96 14 2.13 15
GOtrunca 0.49 9 0.95 5 2.92 7 1.96 5

MDcircul 0.95 65 0.94 2 3.16 51 2.92 58
NAcapita 0.78 59 1.17 55 3.16 53 3.41 64
NAcaprad 0.74 53 1.16 49 3.02 32 2.68 50
NAcostul 0.74 52 1.19 60 3.01 29 2.52 41
NAcrypto 0.91 64 1.16 48 3.12 46 3.16 61
NAcryten 0.64 36 1.17 53 3.23 61 2.60 46
N Agregar 0.67 41 1.10 35 3.07 41 3.27 63
NAignacc 0.65 40 1.13 38 2.99 22 2.23 21
NAlanceo 0.69 46 1.08 29 3.20 58 3.66 67
NAmengru 0.99 67 1.19 59 3.07 42 2.84 55
N Aradios 0.59 27 1.26 66 3.40 64 2.37 29
NAreicha 0.75 54 1.15 46 3.00 25 2.53 42
NAtripun 0.63 33 1.08 28 2.95 13 2.33 27
N Aveneta 0.68 42 1.16 50 3.20 57 3.79 68
NIamphib 0.70 48 1.36 68 3 .44 66 2.38 30
NIdissip 0.77 58 1.14 42 3.04 34 3.04 59
NIlinear 0.88 63 1.02 14 3.18 54 3.41 65
NIpaldeb 0.79 60 1.14 44 3.05 38 2.85 56
NIpalea 0.61 29 1.10 33 3.01 28 2.43 32
NIrecta 0.71 49 1.06 19 3.14 49 3.07 60
NIsociab 0.81 61 1.13 40 3.16 52 3.44 66
NIsupral 0.49 7 1.12 36 3.01 27 2.09 12
PTlandub 1.00 68 1.14 43 3.14 48 2.90 57
PT lanfre 0.82 62 1.15 47 3.15 50 2.46 33
PTlanros 0.77 57 1.17 51 3.22 60 2.68 49
PT peraga 0.50 11 1.10 34 3.02 33 2.10 13
REsinuat 0.53 21 1.06 21 2.88 2 1.82 2

SYulna 0.49 8 0.97 8 2.96 15 1.96 6

126

Appendix B. Species Cl optima, Cl ranks, and indicator categories from van Dam’s
trophic index. Cl optima were from an epilithon survey in the GTBW. Species listed
appeared in > 3 sites in both control clay pots and the epilithon survey.

 

Species code Cl Opt Cl rank VD trophic

 

ACbiasol 5 .75 76 3
ACclevei 5 .97 70 4
ACconspi 7.1 1 43 7
ACexigua 5.56 82 7
AClauenb 9.36 22 2
ACminuti 7.23 42 7
ALpelluc 8.14 3 1 2
AMlibyca 6.65 57

Ampedcls 7.32 39 5
CCplaeug 8.15 30 5
CCplalin 5 .26 89 5
CCpsethu 5.36 87

CMaffins 5 .60 80 5
CMcistul 5.48 84 5
CMmicroc 6.58 60 4
CMminuta 6.78 53

CMsilesi 4.53 94 7
DAtenue 5.21 90 5
FRbrevis 3.75 96 7
FRcapgra 5.57 81 2
FRcaprum 7.09 44 2
FRcapvau 5.39 86 5
FRconstr 5.93 72 4
FRconven 5.76 75 4
FRlepdub 6.81 5 1

FRparasi 6.47 63 4
FRpinnat 7.43 37 7
GOgracil 7.32 40 3
GOkobaya 12.84 8

GOolivcm 14.65 3 5
GOparvul 9.77 18 5

GOpumilu 8.72 27 7

 

127

Appendix B. Continued.

 

Species code Cl Opt C1 rank VD trophic

 

MDcircul 8.88 25 7
NAcapita 12.84 7 4
NAcaprad 5 .94 7 1 5
NAcrypto 6.14 67 7
N Acryten 10.18 15 7
NA gregar 12.68 10 5
NAignacc 6.74 54

NAlanceo 19.22 1 5
NAmengru 6.92 48

NAradios 6.80 52 4
NAreicha 8.50 28

NAtripun 9.65 20 5
NAveneta 6.26 65 5
NIdissip 8.85 26 4
NIlinear 7.08 45 4
NIpalea 1 1.24 14 6
NIrecta 5 .80 73 7
NIsociab 1 1.66 12 5
PT landub 7.60 35

PTlanfre 6.56 61 7
PT lanros 6.86 50 5
PT peraga 7.66 34 1
REsinuat 7.39 38 3
SYulna 6.59 59 7

 

128

Appendix C. The correspondence between species codes and species names.

 

 

Species code Species name
ACbiasol Achnanthes biasolettiana (Kutz.) Grun.
ACclevei Achnanthes clevei Grun.

ACconspi Achnanthes conspicua A. Mayer

ACexigua Achnanthes exigua Grun.

AClauenb Achnanthes lauenburgiana Hust.

ACminuti Achnanthes minutissimum Kutz.

ALpelluc Amphipleura pellucida Kutz.

AMlibyca Amphora libyca Ehr.

Ampedcls Amphora pediculus (Kutz.) Grunow
CCneothu Cocconeis neothumensis Kram.

CCplaeug Cocconeis placentula var. euglypta (Ehr.) Cl.
CCplalin Cocconeis placentula var. lineata (Ehr.) V. H.
CCpsethu Cocconeis pseudothumensis Reich.

CMaffins Cymbella aflinis Kutz.

CMcesati Cymbella cesatii (Rabh.) Grun. ex A. S.
CMcistul Cymbella cistula (Ehr.) Kirchn.

CMmicroc Cymbella microcephala Grun.

CMminuta Cymbella minuta Hilse ex Rabh.

CMsilesi Cymbella silesiaca Bleisch

CYdistin Cyclotella distinguenda Hust.

DAtenue Diatoma tenue Ag.

FRbrevis F ragilaria brevistriata Grun.

FRcapgra F ragilaria capucina var. gracilis Hust.
FRcaprum F ragilaria capucina var. rumpens (Kutz.) Lange-Bert.
FRcapvau F ragilaria capucina var. vaucheriae (Kutz.) Lange-Bert.
FRconstr F ragilaria construens (Ehr.) Grun.

FRconven F ragilaria construens f. venter (Ehr). Grun.
FRdelcat F ragilaria delicatissima

FRlepdub F ragilaria leptostauron var. dubia (Grun.) Hust.
FRnanana F ragilaria nanana Lange-Bert.

FRparasi F ragilaria parasitica (W. Sm.) Grun.
FRpinnat F ragilaria pinnata Ehr.

FRtenera F ragilaria tenera (W. Sm.) Lange-Bert.
GOgracil Gomphonema gracile Ehr. emend. V. H.

 

 

129

Appendix C. Continued.

 

 

Species code Species name

GOkobaya Gomphonema kobayasii Kocio. & Kings.

GOmicrop Gomphonema micropus Kutzing

GOolivcm Gomphonema olivaceum (Lyngb) Kutz.

GOparvul Gomphonema parvulum Kutz.

GOpumilu Gomphonema pumilum (Grun.) Reich. & Lange-Bert.
GOtrunca Gomphonema truncatum Ehr.

MDcircul Meridian circulare (Grev.) Ag.

NAcapita Navicula capitata Ehr.

NAcaprad Navicula capitatoradiata Germain

NAcostul Navicula costulata Grun.

NAcrypto Navicula cryptocephala Kutz.

NAcryten Navicula cryptotenella Lange-Bert.

NAgregar Navicula gregaria Donk.

NAignacc Navicula ignota var. acceptata (Hust.) Lange-Bert.
NAlanceo Navicula lanceolata (Ag) Kutz.

NAmengru Navicula menisculus var. grunowii Lange-Bert.
NAradios Navicula radiosa Kutz.

NAreicha Navicula reichardtiana Grunow

NAtripun Navicula tripunctata (O. F. Mull.) Bory

NAveneta Navicula veneta Kutz.

NIamphib Nitzschia amphibia Grun.

NIdissip Nitzschia dissipata (Kutz.) Grun.

NIlinear Nitzschia linearis (Ag. ex W. Sm.) W. Sm.

NIpaldeb Nitzschia palea var. debilis (Kutz.) Grun.

NIpalea Nitzschia palea (Kutz.) W. Sm.

NIrecta Nitzschia recta Hantz. ex Rabh.

NIsociab Nitzschia sociabilis Hust.

N Isupral Nitzschia supralitorea Lange-Bert.

PTlandub Planothidium lanceolata var. dubia (Grun.) Lange-Bert.
PTlanfre Planothidium lanceolata subsp. frequentissima Lange-Bert.
P’Tlanros Planothidium lanceolata subsp. rostrata (Oestrup) Lange-Bert.
PTperaga Planothidium peragalli Brun & Herib.

REsinuat Reimeria sinuata (Greg) Kocio. & Stoer.

SYulna Synedra ulna W. Sm.

 

130

 

 

BIBLIOGRAPHY

131

Allen , J. D., D. L. Erickson, and J. Fay. 1997. The influence of catchment land use on
stream integrity across multiple spatial scales. Freshwater Biology 37: 149-161.

American Public Health Association (APHA). 1998. Standard methods for the
examination of water and wastewater, 20th edition. American Public Health
Association, Washington, DC., US.

Anderson, J. R., E. E. Hardy, J. T. Roach, and R. E. Witmer. 1976. A Land Use and Land
Cover Classification System for Use with Remote Sensor Data. US. Geological
Survey, Professional Paper 964. Reston, Virginia, US.

Arbuckle, K. E., and J. A. Downing. 2001. The influence of watershed land use on lake
N : P in a predominantly agricultural landscape. Limnology and Oceanography 46:
970-975.

Archibald, R. E. M. 1972. Diversity in some South Africa diatom assemblages and its
relation to water quality. Water Research 6: 1229-1238.

Bahls, L. L. 1993. Periphyton bioassessment methods for Montana streams. Montana
Department of Health and Environmental Sciences, Helena, Montana, US.

Bailey, H. H., and J. H. Winsor. 1964. Kentucky soils. University of Kentucky,
Agricultural Experiment Station, Lexington, US.

Barbour, M. T., J. Gerritsen, B. D. Snyder, and J. B. Stribling. 1999. Rapid bioassessment
protocols for use in wadeable streams and rivers. EPA 841-B-99-002. Office of
Water, US Environmental Protection Agency, Washington, DC., US.

Barbour, M. T., J. Gerritsen, G. E. Griffith, R. Frydenborg, E. McCarron, J. S. White, and
M. L. Bastian. 1996. A framework for biological criteria for Florida streams using

benthic macroinvertebrates. Journal of the North American Benthological Society
15: 185-21 1.

Baron, J. S., N. L. Poff, P. L. Angermeier, C. N. Dahm, P. H. Gleick, N. G. Hairston, R.
B. Jackson, C. A. Johnston, B. D. Richter, and A. D. Steinman. 2002. Meeting

ecological and societal needs for freshwater. Ecological Applications 12: 1247-1260.

Bertness, M. D., and A. M. Ellison. 1987. Determinants of pattern in a New England salt
marsh plant community. Ecological Monographs 57: 129-147.

Biggs, B. J. 2000. Eutrophication of streams and rivers: dissolved nutrient-chlorophyll

relationships for benthic algae. Journal of the North American Benthological Society
19: 17-31.

132

 

'F

Biggs, B. J. F., and R. A. Smith. 2002. Taxonomic richness of stream benthic algae:
effects of flood disturbance and nutrients. Limnology and Oceanography 47: 1175-
1186.

Blair, R. B. 1996. Land use and avian species diversity along an urban gradient.
Ecological applications 6: 506-519.

Blindow, I. 1987. The composition and density of epiphyton on several species of
submerged macrophytes — the neutral substrate hypothesis tested. Aquatic Botany 29:
257-268.

Blinn, D.W. 1993. Diatom community structure along physicochemical gradients in
saline lakes. Ecology 74: 1246-1263.

Bloesch, J ., P. Bossard, H. Buhrer, H. R. Burgi, and U. Uehlinger. 1988. Can results from
limnocorral experiments be transferred to in situ condition? Hydrobiologia 159: 297-
308.

Boutt, D. F., D. W. Hyndman, B. C. Pijanowski, and D. T. Long. 2001. Identifying
potential land use-derived solute sources to stream base-flow using ground water
models and GIS. Ground Water 39: 24-34.

Bray, J. R., and J. T. Curtis. 1957. An ordination of the upland forest communities of
southern Wisconsin. Ecological Monographs 26: 325-349.

Brezonik, P. L. et al. 1993. Experimental acidification of Little Rock Lake, Wisconsin:
chemical and biological changes over the pH range 6.1 to 4.7. Canadian Journal of
Fisheries and Aquatic Science 50: 1101-1121.

Brown, J. H. 1999. Macroecology: progress and prospect. Oikos 87: 3-14.

Brown, J. H., T. G. Whitman, S. K. M. Ernest, and C. A. Gehring. 2001. Complex
species interactions and the dynamics of ecological systems: long-term experiments.
Science 293: 643-650.

Burton, T. M., M. P. Oemke, and J. M. Molloy. 1991. Contrasting effects of nitrogen and
phosphorus additions on epilithic algae in a headwater and a softwater stream in

northern Michigan. Verh. International Verein Limnology 24: 1644-1653.

Byers, J. E. 2002. Impact of non-indigenous species on natives enhanced by
anthropogenic alteration of selection regimes. Oikos 97: 449-485.

Cairns, J. Jr. 1974. Indicator species vs. the concept of community structure as an index
of pollution. Water Resources Bulletin 10: 338-347.

133

 

Cairns, J ., Jr. 1977. Quantification of biological integrity. Pages 171-187 in R. K.
Ballentine and L. J. Guarria (editors). Integrity of Water. US Environmental
Protection Agency, Washington, DC., US.

Cairns, J, Jr., and P. V. McCormick. 1992. Developing an ecosystem-based capability for
ecological risk assessments. The Environmental Professional 14: 186-192.

Campbell, D. E. 2001. Proposal for including what is valuable to ecosystems in
environmental assessments. Environmental Science and Technology 35: 2867-2873.

Canham, C. D. 1988. Growth and canopy architecture of shade-tolerant trees: response to
canopy gaps. Ecology 69: 786-795.

Carpenter, S. R. 1996. Microcosm experiments have limited relevance for community
and ecosystem ecology. Ecology 77: 667-680.

Carpenter, S. R., J. J. Cole, T. E. Essington, J. R. Hodgson, J. N. Houser, J. F. Kitchell,
and M. L. Pace. 1998. Evaluating Alternative Explanations in Ecosystem
Experiments. Ecosystems 1: 335-344. 5'

 

Carpenter, S. R., N. F. Caraco, D. L. Correll, R. W. Howarth, A. N. Sharpley, and V. H.
Smith. 1998. Nonpoint pollution of surface waters with phosphorus and nitrogen.
Ecological Applications 8:559-568.

Carrick, H. J ., R. L. Lowe, J. T. Rotenberry. 1988. Guilds of benthic algae along nutrient
gradients: relationship to algal community diversity. Journal of North American
Benthological Society 7: 117-128.

Cattaneo, A. 1987. Periphyton in lakes of different trophy. Canadian Journal of Fisheries
and Aquatic Sciences 44: 296-303.

Cattaneo, A., Méthot, G., B. Pinel-Alloul, T. Niyonsenga, and L. Lapierre. 1995.
Epiphyte size and taxonomy as biological indicators of ecological and toxicological
factors in Lake Saint-Francois (Quebec). Environmental Pollution 87: 357-372.

Charles, D. F. 1996. Use of algae for monitoring rivers in the United States: some
examples. Pages 109-118 in B. A. Whitton and E. Rott (editors). Use of algae for
monitoring rivers, II. Institut fur Botanik, Universitat Innsbruck.

Charles, D. F., C. Knowles, and R. S. Davis. ed. 2002. Protocols for the analysis of algal
samples collected as part of the US. Geological Survey national water-quality

assessment program. Report No. 02-06. The Academy of Natural Science,
Philadelphia, US.

134

Chessman, B. C., P. E. Hutton, and J. M. Burch. 1992. Limiting nutrients for periphyton
growth in sub-alpine, forest, agriculture and urban streams. Freshwater Biology 28:
349-361.

Chetelat, J ., F., R. Pick, A. Morin, and P. B. Hamilton. 1999. Periphyton biomass and
community composition in rivers of different nutrient status. Canadian Journal of
Fisheries and Aquatic Science 56: 560-569.

Conley, D. J. 1999. Biogeochemical nutrient cycles and nutrient management strategies.
Hydrobiologia 410: 87-96.

Cooper, C. M. 1993. Biological effects of agriculturally derived surface water pollutants
on aquatic systems — a review. Journal of Environmental Quality 22: 402-408.

Cumming, B. F., and J. P. Smol. 1993. Development of diatom-based salinity models for
paleoclimatic research from lakes in British Columbia (Canada). Hydrobiologia
269/270: 575-586.

Dale, V. H., S. Brown, R. A. Haeuber, N. T. Hobbs, N. Huntly, R. J. Naiman, W. E.
Riebsame, M. G. Turner, and T. J. Valone. 2000. Ecological principles and
guidelines for managing the use of land. Ecological Applications 10: 639-670.

Descy, J. P. 1979. A new approach to water quality estimation using diatoms. Nova
Hedwigia 64: 305-323.

Diamond, J. M., and V. B. Serveiss. 2001. Identifying sources of stress to native aquatic
fauna using a watershed ecological risk assessment framework. Environmental
Science and Technology 35: 4711-4718.

Dixit, S.S., J. P. Smol, D. F. Charles, R. M. Hughes, S. G. Paulsen, and G. B. Collins.
1999. Assessing water quality changes in the lakes of the Northeastern United States

using sediment diatoms. Canadian Journal of Fisheries and Aquatic Sciences
56: 13 l- 152.

Dodds, W. K., and D. A. Gudder. 1992. The ecology of Cladophora in freshwaters.
Journal of Phycology 28: 415-427.

Dodds, W. K., M. A. Evans-White, N. M. Gerlanc, L. Gray, D. A. Gudder, M. J. Kemp,
L. Lopez, D. Stagliano, E. A. Strauss, J. L. Tank, M. R.Whiles, and W. M.

Wollheim. 2000. Quantification of the nitrogen cycle in a prairie stream. Ecosystems
3: 574-589.

Dodds,W. K., V. H. Smith, and K. Lohman. 2002. Nitrogen and phosphorus

relationships to benthic algal biomass in temperate streams. Canadian Journal of
Fisheries and Aquatic Sciences 59: 865-874.

135

 

Dodson, S. 1., and S. E. Amott, and K. L. Cottingham. 2000. The relationship in lake
communities between primary productivity and species richness. Ecology 81: 2662-
2679.

Douglas, B. 1958. The ecology of the attached diatoms and other algae in a small stony
stream. Journal of Ecology 46: 295-322.

Dufrene, M., and P. Legendre. 1997. Species assemblages and indicator species: the need
for a flexible asymmetrical approach. Ecological Monographs 67: 345-366.

Elser, J. J ., E. R. Marzolf, C. R. Goldman. 1990. Phosphorus and nitrogen limitation of If-
phytoplankton growth in the freshwaters of North America: A review and critique of
experimental enrichments. Canadian Journal of Fisheries and Aquatic Sciences 47:
1468- 1477.

England, G., and S. D. Cooper. 2003. Scale effects and extrapolation in ecological
experiments. Advances in Ecological Research 33: 161-213.

 

:1-

Fairchild, G. W., R. L. Lowe, and W. B. Richardson. 1985. Algal preiphyton growth on
nutrient-diffusion substrates: an in-situ bioassay. Ecology 66: 465-472.

Finlay, J. C. 2001. Stable-carbon-isotope ratios of river biota: Implications for energy
flow in lotic food webs. Ecology 82: 1052-1064.

Fischer, A. G. 1960. Latitudinal variation in organic diversity. Evolution 14: 64-81.

Fischer, J. M., J. L. Klug, A. R. Ives, and T. M. Frost. 2001. Ecological history affects
zooplankton community responses to acidification. Ecology 82: 2984-3000.

Fore, L. S., J. R. Karr, and L. L. Conquest. 1994. Statistical properties of an index of
biological integrity used to evaluate water resources. Canadian Journal of Fisheries
and Aquatic Sciences 51:1077-1087.

Francoeur, S. N. 2001. Meta-analysis of lotic nutrient amendment experiments: detecting
and quantifying subtle responses. Journal of North American Benthological Society
20: 358-368.

Gerritsen, J. 1995. Additive biological indices for resource management. Journal of the
North American Benthological Society 14: 451-457.

Gomez, N. 1998. Use of epipelic diatoms for evaluation of water quality in the Matanza-
Riachuelo (Argentina), a Pampean plain river. Water Research 32: 2029-2034.

Grand Traverse County. 1996. Focus 2020: Grand Traverse County master plan.
Williams & Works, Traverse City, Michigan, US.

136

Gregory, S. V., F. J. Swanson, W. A. McKee, and K.W. Cumrrrins. 1991. An ecosystem
perspective of riparian zones. BioScience 41: 540-551.

Griffith M. B., P. R. Kaufmann, A. T. Herlihy, and B. H. Hill. 2001. Analysis of
macroinvertebrate assemblages in relation to environmental gradients in Rocky
Mountain streams. Ecological Applications 11: 489-505.

Grimm, N. B., and S. G. Fisher. 1986. Nitrogen limitation potential of Arizona streams
and rivers. Journal of the Arizona-Neveda Academy of Science 21: 31-43.

Gulis, V, and K. Suberkropp. 2003. Leaf litter decomposition and microbial activity in
nutrient-enriched and unaltered reaches of a headwater stream. Freshwater Biology
48: 123-134.

Gullberg, K. R., W. Goedkoop, and R. K. Johnson. 1997. The fate of diatom carbon
within a freshwater benthic community - a microcosm study. Limnology and
Oceanography 42: 452-460.

Gunderson, L. H. 2001. Managing surprising ecosystems in southern Florida. Ecological
Economics 37: 371-378.

Havens, K. E. 1994. Experimental perturbation of a freshwater plankton community: a
test of hypotheses regarding the effects of stress. Oikos 69: 147-153.

Havens, K. E., and R. E. Carlson. 1998. Functional complementarity in plankton
communities along a gradient of acid stress. Environmental Pollution 101: 427-436.

Herlihy, A. T., J. L. Stoddard, and C. B. Johnson. 1998. The relationship between stream
chemistry and watershed land cover data in the rrrid-Atlantic region, US. Water, Air,
and Soil Pollution 105: 377-386.

Hiebert, F. K., and P. C. Bennett. 1992. Microbial control of silicate weathering in
organic-rich ground water. Science 258: 278-281.

Hill, B. H., R. J. Stevenson, F. H. McCormick, and C. B. Johnson. 2000. Use of
periphyton assemblage data as an index of biotic integrity. Journal of the North
American Benthological Society 19: 50-67.

Hill, B. H., R. J. Stevenson, Y. Pan, A.H. Herlihy, P. R. Kaufmann, and C. B. Johnson.
2001. Correlations of stream diatoms with their environment: a comparison of

genus-level and species-level identifications. Journal of the North American
Benthological Society 20: 299-310.

Hill, W. R., and A. W. Knight. 1988. Nutrient and light limitation of algae in two northen
California streams. Journal of Phycology 24: 125-132.

137

 

Hudon, C., and P. Legendre. 1987. The ecological implications of growth forms in
epibenthic diatoms. Journal of Phycology 23: 434-441.

Hurlbert, S. J. 1984. Pseudoreplication and design of ecological field experiments.
Ecological Monographs 54: 187-211.

Hutchinson, G. E. 1957. Concluding remarks. Cold Spring Harbour Symposium on
Quantitative Biology 22: 415-427.

Hynes, H. B. N. 1975. The stream and its valley. Internationale Vereinigung fur
Theoretische und Angewandte Limnologie Verhabdlungen 19: 1-15.

J acoby, J. M. 1987. Alterations in periphyton characteristics due to grazing in a Cascade
foothill stream. Freshwater Biology 18: 495-508.

Jtlttner, I., H. Rothfritz, and S. J. Ormerod. 1996. Diatoms as indicators of river quality in
the Nepalese Middle Hills with consideration of the effects of habitat-specific
sampling. Freshwater Biology 36: 475-486.

Karr, J. R. 1991. Biological integrity: a long-neglected aspect of water resource
management. Ecological Applications 1: 66-84.

Karr, J. R. 1999. Defining and measuring river health. Freshwater Biology 41: 221-234.

Karr, J. R., and D. R. Dudley. 1981. Ecological perspective on water quality goal.
Environmental Management 5: 55-68.

Karr, J. R., and E. W. Chu. 1999. Sustaining living rivers. Hydrobiologia 422/423: 1-14.

KDOW (Kentucky Division of Water). 1993. Methods for assessing biological integrity
of surface waters. Kentucky Department of Environmental Protection, Frankfort,
Kentucky, US.

Keithan, E. D., R. L. Lowe, and H. R. DeYoe. 1988. Benthic diatom distribution in a
Pennsylvania stream: the role of pH and nutrients. Journal of Phycology 24: 581-585.

Kelly, M. G. 1998. Use of the trophic diatom index to monitor eutrophication in rivers.
Water Research 32: 236-242.

Kelly, M. G., and B. A. Whitton. 1995. The trophic diatom index: a new index for
monitoring eutrophication in rivers. Journal of Applied Phycology 7: 433-444.

Kelly, M. G., A. Cazaubon, E. Cting, A. Dell’Uomo, L. Ector, B. Goldsmith, H. Guasch,
J. Hilrlimann, A. Jarlman, B. Kawecka, J. Kwandrans, R. Laugastle, E.-A.
Lindstrom, M. Leito, P. Marvan, J. Padisak, E. Pipp, J. Prygiel, E. Rott, S. Sabater,
H. van Dam, and J. Vizinet. 1998. Recommendations for the routine sampling of

138

 

diatoms for water quality assessments in Europe. Journal of Applied Phycology 10:
215-224.

Kentucky Division of Water (KY DOW). 1993. Methods for assessing biological
integrity of surface waters. Kentucky Department of Environmental Protection,
Frankfort, Kentucky.

Kingston, J. C., R. L. Lowe, E. F. Stoermer, and T. B. Ladewski. 1983. Spatial and
temporal distribution of benthic diatoms in northern Lake Michigan. Ecology 64:
1566-1580.

Kociolek, J. P., and E. F. Stoermer. 2001. Taxonomy and ecology: A marriage of
necessity. Diatom Research 16: 433-442.

Kohler, S.L., and M.J. Wiley. 1997. Pathogen outbreaks reveal large-scale effects of
competition in stream communities. Ecology. 87: 2164-2176.

Kolar, C. S., and D. M. Lodge. 2002. Ecological predictions and risk assessment for alien
fishes in North America. Science 298: 1233-1236.

Kolwitz, R., and M. Marsson. 1908. Okologie der pflanzliche saprobien. Berichte
Deutsche Botanische Gesellschaften 26: 505-19.

Kramrner, K., and H. Lange-Bertalot. 1986. Bacillariophyceae, Teil 1: Naviculaceae.
Gustav Fischer Verlag, Stuttgart, Germany.

Kramrner, K., and H. Lange-Bertalot. 1988. Bacillariophyceae, Teil 2: Bacillariaceae,
Epithemiaceae, and Surirellaceae. Gustav Fischer Verlag, Stuttgart, Germany.

Kramrner, K., and H. Lange-Bertalot. 1991a. Bacillariophyceae, Teil 3: Centrales,
Fragilariaceae, and Eunotiaceae. Gustav Fischer Verlag, Stuttgart, Germany.

Kramrner, K., and H. Lange-Bertalot. 1991b. Bacillariophyceae, Teil 4: Achnanthaceae.
Gustav Fischer Verlag, Stuttgart, Germany.

Kremen, C. 1994. Biological inventory using target taxa: a case study of the butterflies of
Madagascar. Ecological Applications 4: 407-422.

La Peyre, M. K. G., J. B. Grace, E. Hahn, and I. A. Mendelssohn. 2001. The importance
of competition in regulating plant species abundance along a salinity gradient.

Ecology 82: 62-69.

Lange-Bertalot, H. 1995. Rote Liste der Kieselalgen (Bacillariophyceae) Deutschlands.
28: 1-45.

139

 

Leach, M. K., and T. J. Givnish. 1999. Gradients in the composition, structure, and

diversity of remnant oak savannas in southern Wisconsin. Ecological Monographs
69: 353-374.

Legendre, P., and L. Legendre. 1998. Numerical ecology. Elsevier Science, Amsterdam,
Netherlands.

Leland, H. V. 1995. Distribution of phytobenthos in the Yakima River basin, Washington,
in relation to geology, land use, and other environmental factors. Canadian Journal
of Fisheries and Aquatic Science 52: 1108-1129.

Leland, H. V., and S. D. Porter. 2000. Distribution of benthic algae in the upper Illinois
River basin in relation to geology and land use. Freshwater Biology 44: 279-301.

Leung, B., D. M. Lodge, D. Finnoff, J. F. Shogren, M. A. Lewis, and G. Lamberti. 2002.
An ounce of prevention or a pound of cure: bioeconomic risk analysis of invasive

species. Proceedings of the Royal Society of London Series B — Biological Sciences
269: 2407-2413.

Likens, G. E., and RH. Bormann. 1974. Linkages between terrestrial and aquatic
ecosystems”, BioScience 24: 447-456.

Lim, D. S. S., C. Kwan, and M. S. V. Douglas. 2001. Periphytic diatom assemblages
from Bathurst Island, Nunavut, Canadian High Arctic: An examination of
community relationships and habitat preferences. Journal of Phycology 37: 379-392.

Locke, A., and W. G. Sprules. 1993. Effects of acidification on zooplankton population
and community dynamics. Canadian Journal of Fisheries and Aquatic Science 50:
1238-1247.

Lohman, K., J. R. Jones, and C. Baysinger-Daniel. 1991. Experimental evidence for
nitrogen limitation in a northern Ozark stream. Journal of North American
Benthological Society 5: 14-23.

Lowe, R. L. 1974. Environmental requirement and pollution tolerance of freshwater
diatoms. EPA-670/4-74-005. US Environmental Protection Agency, Cincinnati,
Ohio, US.

Loreau, M. 2000. Are communities saturated? On the relationship between alpha, beta
and gamma diversity. Ecology Letters 3: 73-76.

Lowe, R. L. 1974. Environmental requirements and pollution tolerance of freshwater

diatoms. Office of Research and development, US. Environmental Protection
Agency, Cincinnati, Ohio, US.

140

[Jim

 

Lozano, R. B., and J. R. Pratt. 1994. Interaction of toxicants and communities: the role of
nutrient. Environmental Toxicology and Chemistry 13: 361-368.

Lyche, A., T. Anderson, K. Christofferesen, D. O. Hessen, P. H. B. Hansen, and A.
Klysner. 1996. Mesocosm tracer studies, 2, The fate of primary production and the

role of consumers in the pelagic carbon cycle of a mesotrophic lake. Limnology and
Oceanography 41: 475-487.

Mac, M. J ., P. A. Opler, C. E. Puckett Haecker, and P. D. Doran. 1998. Status and trends
of the nation’s biological resources. Vol. 1. US. Department of the Interior, US
Geological Survey, Reston, Virginia, US.

MacArthur, R. H. 1957. On the relative abundance of bird species. Proceedings of the
National Academy of Sciences of the United States of America 43: 293-295.

Malmqvist, B., and S. Rundle. 2002. Threats to the running water ecosystems of the
world. Environmental Conservation 29: 134—153.

Mann, H. B., and D. R. Whitney. 1947. On a test of whether one of two random variables
is stochastically larger than the other. Annals of Mathematical Statistics 18: 50-60.

Marks, J. C., and R. L. Lowe. 1993. Interactive effects of nutrient availability and light
levels on the periphyton composition of a large oligotrophic lake. Canadian Journal
of Fisheries and Aquatic Science 50: 1270-1278.

McCormick, P. V., and M. B. O’Dell. 1996. Quantifying periphyton responses to
phosphorus in the Florida Everglades: a synoptic-experimental approach. Journal of
North American Benthological Society 15: 450-468.

McCune, B., and J. B. Grace. 2002. Analysis of ecological communities. MjM software
design, Gleneden Beach, Oregon, US.

McDoIll, R. C. 1986. The Geology of Kentucky: a text to accompany the geologic map
of Kentucky. Department of the Interior, US Geological Survey, Denver, Colorado,
USA.

McKinney, M. L., and J. L. Lockwood. 1999. Biotic homogenization: a few winners
replacing many losers in the next mass extinction. Trends in Ecology and Evolution
14: 450-453.

Miller, A. R., R. L. Lowe, and J. T. Rotenberry. 1987. Succession of diatom communities
on sand grains. Journal of Ecology 75: 693-709.

Mills, M. R., R. E. Houp, J. F. Brumley, and K. L. Smathers. 1997. Reference reach fish

community. Technical Report No. 52. Kentucky Department for Environmental
Protection, Division of Water. Frankfort, Kentucky, US.

141

Mitsch, W. J ., J. W. Day, J. W. Gilliam, P. M. Groffman, D. L. Hey, G. W. Randall, and
N. W. Wang. 2001. Reducing nitrogen loading to the Gulf of Mexico from the

Mississippi River Basin: Strategies to counter a persistent ecological problem.
BioScience 51: 373-388.

Mittelbach, G. G., C. F. Steiner, S. M. Scheiner, K. L. Gross, H. L. Reynolds, R. B.
Waide, M. R. Willig, S. I. Dodson, and L. Gough. 2001. What is the observed
relationship between species richness and productivity? Ecology 82: 2381-2396.

Mosisch, T. D., S. E. Bunn, and P. M. Davis. 2001. The relative importance of shading ..
and nutrients on algal production in subtropical streams. Freshwater Biology 46: F
1269-1278.

Munn, M. D., R. W. Black, and S. J. Gruber. 2002. Response of benthic algae to .‘
environmental gradients in an agriculturally dominated landscape. Journal of the k

 

North American Benthological Society 21: 221-237.

Naeem, S. 2001. Experimental validity and ecological scale as criteria for evaluating
research program. Pages 223-250 in Scaling relations in experimental ecology. R. H.
Gardner, W. M. Kemp, V. S. Kennedy, J. E. Peterson eds. Columbia University
Press, New York, US.

Newman, S., J. Schuette, J. B. Grace, K. Rutchey, T. Fontaine, K. R. Reddy, and M.
Pietrucha. 1998. Factors influencing cattail abundance in the northern Everglades.
Aquatic Botany 60: 265-280.

Niyogi, D. K., W. M. Lewis, Jr, and D. M. McKnight. 2002. Effects of stress from mine
drainage on diversity, biomass, and function of primary producers in mountain
streams. Ecosystems 5: 554-567.

Noss, R. F. 1990. Indicators for monitoring biodiversity: a hierarchical approach.
Conservation Biology 4: 355-364.

Odum, E. P. 1985. Trends expected in stressed ecosystems. BioScience 35: 419-422.

Ostrander, G. K., K. M. Armstrong, E. T. Knobbe, D. Gerace, and E. P. Scully. 2000.
Rapid transition in the structure of a coral reef community: The effects of coral
bleaching and physical disturbance. Proceedings of the National Academy of
Sciences of the United States of America 97: 5297-5302.

Pan, Y. D., and R. J .Stevenson. 1996. Gradient analysis of diatom assemblages in
western Kentucky wetlands. Journal of Phycology 32: 222-232.

Pan, Y., R. J. Stevenson, B. Hill, A. Herlihy, and G. Collins. 1996. Using diatoms as

indicators of ecological conditions in lotic systems: a regional assessment. Journal of
North American Benthological Society 15: 481-495.

142

Patrick, R. 1973. Use of algae, especially diatoms, in the assessment of water quality.
Pages 76-95 in Biological Methods for the Assessment of Water Quality, ASTM
STP528. American Society for Testing and Materials. Philadelphia, Pennsylvania,
US.

Patrick, R., and C. W. Reimer. 1966. The diatoms of the United States, Volume 1. The
Academy of Natural Sciences of Philadelphia, Philadelphia, Pennsylvania, US.

Patrick, R., and C. W. Reimer. 1975. The diatoms of the United States, Volume 2. The
Academy of Natural Sciences of Philadelphia, Philadelphia, Pennsylvania, US.

Paul, M. J ., and J. L. Meyer. 2001. Streams in the urban landscape. Annual Review in
Ecology and Systematics 32: 333-365.

Perrings, C., M. Williamson, E. B. Barbier, D. Delfino, S. Dalmazzone, J. Shogren, P.
Simmons, and A. Watkinson. 2002. Biological invasion risks and the public good:
an economic perspective. Conservation Ecology 6: art. no. 1.

Peters, R. H. 1983. The ecological implications of body size. Cambridge University Press,
Cambridge, UK.

Peterson, C. G. 1996. Response of benthic algal communities to natural physical
disturbance. Pages 375-403 in R. J. Stevenson, M. L. Bothwell, and R. L. Lowe
(editors). Algal ecology: freshwater benthic ecosystems. Academic Press, New York,
US.

Peterson, C. G., and R. J. Stevenson. 1992. Resistance and resilience of lotic algal
communities — importance of disturbance timing and current. Ecology 73: 1445-
1461.

Peterson, J. E., J. C. Comwell, and M. W. Kemp. 1999. Implicit scaling in the design of
experimental aquatic ecosystems. Oikos 85: 3-18.

Petit, M., G. P. Alves, and P. Lavandier. 1999. Phytoplanktonic exudation, bacterial
reassimilation and production for three diel cycles in different trophic conditions.
Archiv ftlr Hydrobiologie 146: 285-309.

Pielou, E. C. 1966. The measurement of diversity in different types of biological
collections. Journal of Theoretical Biology 13: 131-144.

Pielou, E. C. 1975. Ecological diversity. Wiley-Interscience publication, New York, US.

Pienitz, R., and W. F. Vincent. 2000. Effect of climate change relative to ozone depletion
on UV exposure in subarctic lakes. Nature 404: 484-487.

143

 

Prescott, G. W. 1951. Algae of the western Great Lakes area. Cranbrook Institute of
Science, Bloomfield Hills, Michigan, US.

Pringle, C. M., and J. A. Bowers. 1984. An in situ substratum fertilization technique:
diatom colonization on nutrient enriched, sand substrata. Canadian Journal of
Fisheries and Aquatic Sciences 41: 1247-1251.

Progar, R. A., and T. D. Schowalter. 2002. Canopy arthropod assemblages along a
precipitation and latitudinal gradient among Douglas-fir Pseudotsuga menziesii
forests in the Pacific Northwest of the United States. Ecography 25: 129-138.

Prygiel, J ., and M. Coste. 1993. The assessment of water quality in the Artois-Picardie
water basin (France) by the use of diatom indices. Hydrobiologia 269/270: 343-349.

Rahel, F. J. 2002. Homogenization of freshwater faunas. Annual Review of Ecology and
Systematics. 33: 291-315.

Rapport, D. J ., H. A. Regier, and T. C. Hutchinson. 1985. Ecosystem behavior under
stress. American Naturalist 125: 617-640.

Rapport, D. J ., and W. G. Whitford. 1999. How ecosystem respond to stress. BioScience
49: 193-203.

Reavie, E. D., and J. P. Smol. 1997. Diatom-based model to infer past littoral habitat
characteristics in the St. Lawrence River. Journal of Great Lakes Research 23: 339-
348.

Richards, C., L. Johnson, and G. E. Host. 1996. Landscape influences on stream habitats
and biota. Canadian Journal of Fisheries and Aquatic Science 53 (Suppl. 1): 295-311.

Richter, B. D., R. Mathews, and R. Wigington. 2003. Ecologically sustainable water

management: Managing river flows for ecological integrity. Ecological Applications
13: 206-224.

Rodewald, A. D., and R. H. Yahner. 2001. Influence of landscape composition on avian
community structure and associated mechanisms. Ecology 82: 3493-3504.

Rodriguez, M. A., and W. M. Lewis. 1997. Structure of fish assemblages along

environmental gradients in floodplain lakes of the Orinoco River. Ecological
Monographs 67: 109-128.

Romani, A. M., and S. Sabater. 1999. Effect of primary producers on the heterotrophic
metabolism of a stream biofilm. Freshwater Biology 41: 729-736.

144

 

Rosemond, A. D., P. J. Mulholland, and J. W. Elwood. 1993. Top-down and bottom-up
control of stream periphyton: effects of nutrients and herbivores. Ecology 74: 1264-
1280.

Rosemond, A. D., C. M. Pringle, A. Ramirez, M. J. Paul, and J. L. Meyer. 2002.
Landscape variation in phosphorus concentration and effects on detritus-based
tropical streams. Limnology and Oceanography 47: 278-289.

Round, E. F. 1981. The ecology of algae. Cambridge University Press, Cambridge, UK.

Round, E. F. 1990. The diatoms: biology and morphology of the genera. Cambridge F“
University Press, Cambridge, UK.

Round, F. E. 1991. Diatoms in river water-monitoring studies. Journal of Applied
Phycology 3: 129-145.

 

Rozdilsky, I. D., J. Chave, S. A. Levin, and D. Tilman. 2001. Towards a theoretical basis d
for ecosystem conservation. Ecological Research 16: 983-995. ‘

Sarnelle, O. 1997. Daphnia effects on microzooplankton: comparisons of enclosure and
whole lake responses. Ecology 78: 913-928.

Schindler, D. W. 1977. Evolution of phosphorus limitation in lakes. Science 195: 260-
262.

Schindler, D. W. 1987. Detecting ecosystem responses to anthropogenic stress. Canadian
Journal of Fisheries and Aquatic Sciences 44 (Suppl. 1): 6-25.

Schindler, D. W. 1998. Replication versus realism: the need for ecosystem scale
experiments. Ecosystems 1: 323-334.

Schlosser, I. J. 1995. Critical landscape attributes that influence fish population dynamics
in headwater streams. Hydrobiologia 303: 71-81.

Scrimgeour, G. J ., and RA. Chambers. 2000. Cumulative effects of pulp mill and
municipal effluents on epilithic biomass and nutrient limitation in a large northern
river ecosystem. Canadian Journal of Fisheries and Aquatic Science 57: 1342-1354.

Scrimgeour, G. J ., and S. Kendall. 2002. Consequences of livestock grazing on water
quality and benthic alagal biomass in a Canadian natural grassland plateau.
Environmental Management 29: 824-844.

Shamsudin, L., and M. A. Sleigh. 1995. Seasonal changes in composition and biomass of
epiphytic algae on the macrophyte Ranunculus penicillatus in a chalk stream, with
estimates of production, and observations on the epiphytes of Cladophora glomerata.
Hydrobiologia 306: 85-95.

145

Shannon, C. E., and W. Weaver. 1949. The mathematical theory of communication.
University of Illinois Press, Urbana, Illinois, US.

Sheehan, P. J. 1984. Effects on community and ecosystem structure and dynamics. Pages
23-145 in P. J. Sheehan (editor). Effects of pollutants at the ecosystem level. Wiley,
Chichester, UK.

Shurin, J. B. 2001. Interactive effects of predation and dispersal on zooplankton
communities. Ecology 82: 3404-3416.

Simpson, E. H. 1949. Measurement of diversity. Nature 163: 688. n...

Sladecek, V. 1973. System of water quality from the biological point of view. Archiv fur
Hydrobioloie und Ergebnisse Limnologie 7: 1-218.

Smith, V. H. 2003. Eutrophication of freshwater and coastal marine ecosystems.
Environmental Science and Pollution Research 10: 126—139.

 

Snyder, E. 13., and C. T. Robinson, G. w. Minshall, and s. R. Rushforth. 2002. Regional E”

patterns in periphyton accrual and diatom assemblage structure in a heterogeneous
nutrient landscape. Canadian Journal of Fisheries and Aquatic Science 59: 564-577.

Southwood, T. R. E. 1977. Habitats, the templet for ecological strategies? Journal of
Animal Ecology 46: 337-365.

Steinman, A. D. 1992. Does an increase in irradiance influence periphyton in a heavily-
grazed woodland stream? Oecologia 91: 163-170.

Steinman, A. D., C. D. McIntire, S.V. Gregory, and G. A. Lamberti. 1989. Effects of
irradiance and grazing on lotic algal assemblages. Journal of Phycology 25: 478-485.

Stevenson, R. J. 1984. Epilithic and epipelic diatoms in the Sandusky River, with
emphasis on species diversity and water quality. Hydrobiologia 114: 161-175.

Stevenson, R. J. 1996. An introduction to algal ecology in freshwater benthic habitats.
Pages 3-30 in Stevenson, M. L. Bothwell, and R. L. Lowe (editors). Algal ecology:
freshwater benthic ecosystems. Academic Press, New York, US.

Stevenson, R. J ., and R. L. Lowe. 1986. Sampling and interpretation of algal patterns for
water quality assessments. Pages 118-149 in B. G. Isom, Ed. Rationale for sampling
and interpretation of ecological data in the assessment of freshwater ecosystems.
ASTM STP 894, American Society Testing and Materials, Philadelphia, US.

Stevenson, R. J ., and S. Hashim. 1989. Variation in diatom community structure among
habitats in sandy streams. Journal of Phycology 25: 678-686.

146

Stevenson, R. J ., and R. Glover. 1993. Effects of algal density and current on ion

transport through periphyton communities. Limnology and Oceanography 38: 1276—
1281.

Stevenson, R. J ., and L. L. Bahls. 1999. Periphyton protocols. Pages 6-1-6-22 in M. T.
Barobour, J. Gerritsen, B. D. Snyder, and J. B. Stribling. Rapid bioassessment
protocols for use in streams and wadeable rivers: Periphyton, benthic
macroinvertebrates, and fish, second edition. EPA 841-B-99-002. US Environmental
Protection Agency, Office of Water, Washington, DC, US.

Stevenson, R. J., and Y. Pan. 1999. Assessing environmental conditions in rivers and F‘
streams with diatoms. Pages 11-40 in E. F. Stoermer and J. P. Smol (editors). The
diatoms: applications for the environmental and earth science. Cambridge University
Press, Cambridge, UK.

Stevenson, R. J ., and F. R. Hauer. 2002. Integrating Hydrogeomorphic and Index of
Biotic Integrity approaches for environmental assessment of wetlands. Journal of the
North American Benthological Society 21: 502-513. 1

 

Stevenson, R. J. and J. P. Smol. 2002. Use of algae in environmental assessments. Pages
775-804 in J. D. [hr and R. G. Sheath (editors). Freshwater algae of North America:
ecology and classification. Academic Press, New York, US.

Stoermer, E. F. 2001. Diatom taxonomy for paleolimnologists. Journal of Paleolimnology
25: 393-398.

Sullivan, M. J. 1982. Similarity of an epiphytic and edaphic diatom community
associated with Spartina altemiflora. Transactions of the American Microscopic
Society 101: 84-90.

Suter, G. W. 1993. A critique of ecosystem health concepts and indices. Environmental
Toxicology and Chemistry 12:1533-1539.

Swift, L. W., and W. T. Swank. 1981. Long-term responses of stream flow following
clearcutting and regrowth. Hydrological Science B, 26: 245-256.

Taft, C. E., and C. W. Taft. 1990. The algae of western Lake Erie. Bulletin of the Ohio
Biological Survey New Series 4: 1-189.

ter Braak, C. J. F., and H. van Dam. 1979. Inferring pH from diatoms: a comparison of
old and new calibration methods. Hydrobiologia 178: 209-223.

ter Braak, C. J. F., and I. C. Prentice. 1988. A theory of gradient analysis. Advances in
Ecological Research 18: 271-317.

147

ter Braak, C. J. F., and P. F. M. Verdonschot. 1995. Canonical correspondence analysis
and related multivariate methods in aquatic ecology. Aquatic Science 57: 255-264.

USDA (US Department of Agriculture). 1938. Soils and Man: 1938 Yearbook of
Agriculture. US Department of Agriculture, Washington DC., US.

USEPA (US Environmental Protection Agency). 1997. Level III ecoregions of the
continental United States (revision of Omemik 1987). National Health and
Environmnetal Effects Research Laboratory, US Environmental Protection Agency,
Corvalis, Oregon, US.

 

USEPA (US Environmental Protection Agency). 1998. 1998 National water quality
inventory report to congress. EPA841-F-00-006. Office of Water, US Environmental
Protection Agency, Washington, DC., US.

 

Vallentyne, J. R. 1972. Phosphorus and the control of eutrophication. Canadian
Resources Development (May-June): 36—43.

van Dam, H. 1982. On the use of measures of structure and diversity in applied diatom
ecology. Nova Hedwigia 73: 97-115.

van Dam, H., A. Mertens, and J. Sinkeldam. 1994. A coded checklist and ecological
indicator values of freshwater diatoms from the Netherlands. Netherlands Journal of
Aquatic Ecology 28: 117-133.

Vitousek, P. M., J. D. Aber, R. H. Howarth, G. E. Likens, P. A. Matson, D. W. Schindler,
W. H. Schlesinger, and D. G. Tilman. 1997. Human alteration of the global nitrogen
cycle: source and consequences. Ecological Applications 3: 737-750.

Vollenweider, R. A. 1976. Advances in defining critical loading levels for phosphorus in
lake eutrophicaiton. Memorial Institution Italian Hydrobiology 33:53-83.

Wang, Y-K., R. J. Stevenson. 2002. An ecological assessment framework for watershed
management. Journal of the Chinese Institute of Environmental Engineering 12:
209-215.

Watanabe, T., A. Kazumi, and A. Houki. 1990. Numerical simulation of organic
pollution in flow waters. Pages 251-281 in P. N. Cherernisinoff (editor).
Encyclopedia of environmental control technology, Volume 4, Hazardous waste
containment and treatment. Gulf Publishing Company, Houston, Texas.

Whitford, L. A. and G. J. Schumacher. 1984. A manual of freshwater algae, revised
edition. Sparks Press, Raleigh, North Carolina, US.

Whittaker, R. H. 1956. Vegetation of the Great Smoky Mountains. Ecological
Monographs 26: 1-80.

148

Whittaker, R. H. 1972. Evolution and measurement of species diversity. Taxon 21: 213-
25 1 .

Wiens, J. A. 2001. Understanding the problem of scale in experimental ecology. Pages
61-88 in R. H. Gardner, W. M. Kemp, V. S. Kennedy, and J. E. Petersen (editors).
Scaling relations in experimental ecology. Columbia University Press, New York,
US.

Winter, J. G., and H. C. Duthie. 2000. Stream epilithic, epipelic, and epiphytic diatoms:
habitat fidelity and use in biomonitoring. Aquatic Ecology 34: 345-353.

Wold, A. P., and A. E. Hershey. 1999. Spatial and temporal variability of nutrient
limitation in 6 North Shore tributaries to Lake Superior. Journal of North American
Benthological Society 18: 2- 14.

Yoder, C. O., and E. T. Rankin. 1998. The role of biological indicators in a state water _
quality management process. Environmental Monitoring and Assessment 51: 61-88. J

 

Zampella, R. A., and J. F. Bunnell. 1998. Use of reference-site fish assemblages to assess
aquatic degradation in pinelands streams. Ecological Applications 8: 645-658.

149

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

ll‘illflllllllflljlflllzllfllIgllljfllltl