sampling adequacy of freshwater mussel surveys …
TRANSCRIPT
SAMPLING ADEQUACY OF FRESHWATER MUSSEL SURVEYS AND VARIATION OF
MUSSEL SPECIES RICHNESS IN ILLINOIS WADEABLE STREAMS
BY
JIAN HUANG
THESIS
Submitted in partial fulfillment of the requirements
for the degree of Master of Science in Natural Resources and Environmental Sciences
in the Graduate College of the
University of Illinois at Urbana-Champaign, 2011
Urbana, Illinois
Advisor:
Dr. Yong Cao
ii
ABSTRACT
Freshwater mussels are one of the most imperiled groups of animals in North America.
Effective conservation strategies and resource management of freshwater mussels require
adequately characterizing local mussel assemblages. However, sampling protocols for mussel
surveys, including sampling efforts, have not been well established and tested. Furthermore, the
percentage of all species captured with a standard sampling effort (e.g., search of man-hours)
may vary greatly among sites, introducing biases into our understanding of species-diversity
patterns and temporal trends. In addressing both questions, I focused on time-based search, one
commonly used sampling technique in stream mussel surveys in the present study. I sampled 18
wadeable-stream sites mainly in east-central Illinois, selected based on watershed size,
dominant-substrate type, and historic species diversity. With 16 man-hour search per site, my
sampling crew collected 27-942 individuals and 5-18 species per site. I estimated the total
species richness at a site with Chao-1 method that accounted for imperfect species detectability. I
measured sampling adequacy at a given effort as the % of all estimated species recorded. A
frequently used effort, 4 man-hour search, captured 15-100% of all species with an average of
61%. Observed species richness was not significantly correlated with the estimated total richness
until sampling effort reached 8 man-hours (Pearson’s r = 0.59, p < 0.05), which captured 70%
of all species at 2/3 of sites. A 10 man-hour search yielded much stronger correlation (Pearson’s
r = 0.78, p < 0.01) and 70% of all species at 72 % of sites. A Random-Forests (RF) regression
model based on watershed and habitat characteristics accounted for 45% of total variance in
sampling adequacy among sites at 4 man-hours. Sampling adequacy decreased with increasing
stream size and substrate size, but increased with % of forests in the riparian zone and logs in
streams. A second RF model was developed based on the same environmental variables to
predict man-hours required for capturing 70% of all species (pesduo-R2
of 41%) at specific sites.
iii
I also showed that species richness at a site tended to increase with watershed size, stream
size, % of open water in the riparian zone, but decrease with % of agricultural land-use. These
findings should serve as a guide for setting standard sampling efforts (e.g., 10 man-hour search)
for mussel surveys in Illinois and likely other Midwest states, and provide critical information
for setting site-specific efforts in the future studies.
iv
ACKNOWLEDGEMENTS
At the very first, I’m honored to express my deepest gratitude to my dedicated advisor,
Dr. Yong Cao. He has offered me valuable ideas, suggestions and criticisms. His patience and
kindness are greatly appreciated. Additionally, he always puts high priority on my dissertation
writing. I have learnt from him a lot not only about dissertation study, but also general scientific
methods.
I also would like to express my thanks to Ms. Ann Holtrop at Illinois Department of
Natural Resources, who provided general supervision to the state-wide mussel surveys, including
the present study, as well as much of GIS data used in my analysis. I also extend my thanks to
Mr. Kevin Cummings in the Illinois Natural History Survey, who guided me through mussel
identification, offered many insights into sampling design, helped with the field sampling and
my thesis writing, to Ms. Shasteen Diane, Sarah Bales and Alison Price, who assisted me in
many ways, including sampling design, field sampling and species identifications. I cannot
accomplish the mussel sampling and this dissertation without their supports and encouragement.
Many summer technicians helped me with field sampling. I really enjoyed all the field trips with
these guys, and I am also proud that we together made efforts on the conservation and
management of freshwater mussels, the beautiful gifts from nature. My committee member, Dr.
Walter Hill at Illinois Natural History Survey, and Dr. Cory Suski at Department of Natural
Resources and Environmental Sciences, University of Illinois at Urbana-Champaign provided
many supports throughout my dissertation writing.
At last but not least, I would like to thank my family and my friends for their thoughtful
and constructive criticisms and encouragement all the way from the very beginning of my
graduate study.
v
TABLE OF CONTENTS
CHAPTER 1 INTRODUCTION .................................................................................................. 1
1.1 Freshwater Mussel Biology ...................................................................................................... 1
1.2 Freshwater Mussel Ecology ...................................................................................................... 1
1.3 Mussel Diversity and Conservation .......................................................................................... 3
1.4 Sampling Mussel Assemblages................................................................................................. 4
1.5 Variation in Mussel Species Richness ...................................................................................... 7
1.6 Objectives ................................................................................................................................. 8
CHAPTER 2 METHODS ........................................................................................................... 10
2.1 Sampling Design ..................................................................................................................... 10
2.2 Mussel Sampling and Data Compiling ................................................................................... 11
2.3 Data Analyses ......................................................................................................................... 13
CHAPTER 3 RESULTS ............................................................................................................. 27
3.1 Mussel Assemblages ............................................................................................................... 27
3.2 Sampling Adequacy ................................................................................................................ 27
3.3 Sampling Efforts for Specific Sampling Adequacy ................................................................ 29
3.4 Modeling Mussel Species Richness ........................................................................................ 30
CHAPTER 4 DISCUSSION ....................................................................................................... 47
4.1 Measuring Sampling Adequacy .............................................................................................. 47
4.2 Mussel Richness-Environment Relationships......................................................................... 50
CHAPTER 5 SUMMARY .......................................................................................................... 52
REFERENCES ………………………………………………………………………………….53
APPENDIX A Mussel Demographic Field Data Sheet .............................................................. 63
APPENDIX B Freshwater Mussel Species and Their Abundance ............................................. 64
at each of the 18 Wadeable-Stream Sites ........................................................... 64
APPENDIX C Number of Mussel Species Accumulated over Sampling Effort
at 18 Wadeable-Stream Sites in Illinois. ............................................................ 66
APPENDIX D The Descriptions of Environmental Variables of 18 Wadeable-Stream
Sites in Illinois. .................................................................................................. 67
1
CHAPTER 1
INTRODUCTION
1.1 FRESHWATER MUSSEL BIOLOGY
Freshwater mussels (also called clams) are in the families of Margaritiferidae and
Unionidae of Mollusca. A freshwater mussel has a soft body covered by two-valved shell that
distinguishes it from other animals. Most mussel species are sexually distinct, but
hetmaphroditism is regular or occasional in certain species (Walker et al. 2001). In reproduction
seasons, female mussels catch sperms in waters and take them into their gills to fertilize eggs.
The fertilized eggs hatch into glochidia (i.e., specialized larvae of freshwater mussels) in weeks.
For metamorphosis, glochidia of most species need to host on fish, crayfish or amphibian
(Watters & O'Dee 1998, Strayer 2008). A species may use a series of sophisticated strategies to
identify and attach to suitable hosts. For example, glochidia may await fish in water, female
adults can capture fish and then shed glochidia into fish's gills, and adults or glochidia in many
species also resemble lures to attract fish hosts (Haag et al. 1995). This parasitic stage facilitates
dispersions of freshwater mussels (Watters 1996, Strayer 2008). After several months’ parasitical
life, juveniles release from hosts and settle down in the sediment and become mature after
typically 1-3 years (Haag & Staton 2003).
1.2 FRESHWATER MUSSEL ECOLOGY
Freshwater mussels usually live on the substrate surface or in the sediment, and obtain
their food by filtering bacteria, planktons, and other organic particles. This group of animals can
account for a large proportion of benthos biomass in many streams (Howard & Cuffey 2006) and
increase benthic productivity by transferring nutrients and energy from the water column to the
2
sediment (Baker & Levinton 2003, Gatenby et al. 2003, Vaughn et al. 2008). Mussel shells,
associated with substrata, provide habitats for benthic fish and other macroinvertebrate such as
worms and crayfish (Gutierrez et al. 2003, Schwalb & Pusch 2007). The complex associations of
freshwater mussel with fish are well documented. The dispersal and abundance of freshwater
mussels can be affected by the viability of suitable host fish (Haag et al. 1995). Glochidia in the
parasitical stage may be also fatal to their host fish, but juvenile mussels are preyed upon by
many fish species, e.g., catfish and carp (Baker 1916, Strayer 2008). Freshwater mussels are also
preyed upon by shorebirds (e.g., Draulans 1982) and mammals, such as raccoons (Procyon lotor)
and muskrats (Ondatra zibethicus) (Nerves & Odom 1989, Berrow 1991, Zahner-Meike &
Hanson 2001). Muskrats (Ondatra zibethicus) may consume great proportion of mussel
population within certain stream segment (Watters 1994). Additionally, freshwater mussels are
important resources, and mussel farming for food resource, pearls, and shells are still important,
contributing about $ 50 million in the United States annually (Claassen 1994, Theler 2000).
Freshwater mussels can significantly affect water quality by filtering a large volume of
water (Pusch et al. 2001, Vaughn 2010). For a live mussel with about 3 g dry weight of soft part,
the filtration rate was 2.1-4.6 liter per hour (Kryger & Riisgard 1988). On the other hand,
freshwater mussels can be strongly affected by various environmental stressors, such as low
dissolved oxygen (Haag & Warren Jr. 2008), heavy metal contamination (Millington & Walker
1983, Balogh 1988, Wang et al. 2007), and toxic organic compounds (Cossu et al. 2000, Gills et
al. 2010) in the water column and sediment. Acute toxicity tests with copper and ammonia
showed that mussel juveniles and glochidia were among the most sensitive aquatic organisms
(Milam et al. 2005, Keller et al. 2006, Wang et al. 2007). Because of their sensitivity to various
types of pollution, freshwater mussels are often used as indicators of biological conditions
3
(Diamond & Serveiss 2001). For example, US EPA (2007) has used freshwater mussel as
biological indicator to monitor a series of pollutants to waterbodies, including heavy metals,
ammonia, chlorine, insecticides and herbicides.
1.3 MUSSEL DIVERSITY AND CONSERVATION
Freshwater mussels occur in various aquatic habitats (e.g., streams, ponds and lakes)
around the world. Approximately 1000 species were recorded, and North America has the
highest diversity with about 300 mussel species. At the continental scale, climate, geological
events, such as glaciers, evolutionary process (e.g., speciation), and physical constraints (e.g.,
mountain barriers, dams) influence the distribution and dispersal of mussel species (Williams et
al. 1993, Hastie et al. 2003). Southeastern US is one of the global ‘hot spots’ of freshwater
mussels with 267 species recorded (Williams et al. 1993). Mussel diversity is also high in the
Midwest, with about 81 species found in Illinois alone (Herkert 1992).
Biodiversity generally means the variation in biological systems (Gaston & Spicer 2004),
and it is important to ecosystem functions and services (Altieri 1999, Hooper et al. 2005,
Gamfeldt et al. 2008). High biodiversity boosts ecosystem productivity and increase the
resistance and resilience of ecosystems against disturbances (e.g., Giller et al. 2004, Lydeard et
al. 2004, Gamfeldt et al. 2008). Most commonly used measure of biodiversity is species richness
(Purvis & Hector 2000, Hobohm 2003, Gaston & Spicer 2004). However, biodiversity have
encountered great decline during the past centuries, and freshwater mussels are considered as one
of the most endangered taxonomic groups (Neves 1999). Approximately 7% of the 300
freshwater species in the North America are extinct and about 65% of the species considered
endangered and threatened (Williams et al. 1993). In Illinois, 8 species have been extirpated and
4
further 30% of the existing species are imperiled (Illinois Endangered Species Protection Board
2010). Habitat degradation such as impoundments, channelization and dredging is among the
major causes for this retrogression (Watters 1999, Strayer 2008). Water-quality degradation
associated with agricultural runoff and industrial effluent also has strongly affected stream
mussel assemblages (Lynch et al. 1977). Recently, invasive species, such as Asian Clam
(Corbicula fluminea) and zebra mussel (Dreissena polymorpha), also started to threaten native
mussel species (Mackie 1991, Schloesser et al. 1996, Strayer & Malcolm 2007).
1.4 SAMPLING MUSSEL ASSEMBLAGES
Effective mussel conservation and management strategies require reliable estimates of
their richness, abundances and distributions. For example, accurately determining conservation
status of species of concern and identifying critical habitats for rich-species assemblages will
require detailed information about species distribution (Smith et al. 2001, Palmer et al. 2002,
Giam et al. 2010). Both qualitative and quantitative sampling is used in mussel assemblage
surveys, and choices of sampling techniques depend on the goals and resources available of a
particular study. Quantitative search (e.g., setting quadrats) is used to examine population density,
recruitment rates, and age structure (Kovalak et al. 1986), but it is costly, time-consuming, and
infeasible in large-scale surveys (Obermeyer 1998, Strayer & Smith 2003). In comparison,
qualitative searches can efficiently inventory species and these searches are used widely in
biodiversity studies and large-scale surveys (Metcalfe-Smith et al. 2000, Wisconsin DNR 2005,
Tiemann et al. 2009, Krebs et al. 2010). Visual, tactile and snorkeling searches or a combination
thereof are common tools in qualitative techniques with sampling efforts typically measured
according to search time (e.g., Hornbach & Deneka 1996, Schloesser et al. 2006).
5
1.4.1 Definition of sampling adequacy
How adequately field samples characterize assemblages is critical to accurately
understand species distributions and changes in species diversity over space and time. Under-
sampling often leads to misunderstanding of ecological patterns and underestimation of species
richness, abundances and ranges (Kodric-Brown & Brown 1993, Hellmann & Fowler 1999,
Villella & Smith 2005, Smith et al. 2010), and thus it is a major data-quality concern (Remsen
1994, Metcalfe-Smith et al. 2000, Tiemann et al. 2009). In community studies, sampling
adequacy can be measured as the ratio between an estimate and its true value (Cao et al. 2001,
Cao et al. 2002). In the case of characterizing species richness at a site, it refers to the percentage
of species captured. The significance of this particular measure goes beyond species richness per
se. First, species captured in a sample are not a random subset of all species, but a subset of
relatively common and then probably more important ones. This measure of sampling adequacy
is also closely related to the similarity in species composition between two replicate samples
when evaluated with the Jaccard Coefficient, an estimate of sampling adequacy in species
composition (Cao et al. 2004). Therefore, this measure of sampling adequacy has been widely
used for assessing fish sampling protocols (e.g., Cao et al. 2001, Reynolds et al. 2003, Holtrop et
al. 2010).
Time-based mussel searches are widely used in mussel assemblage and conservation
studies, but most surveys have used 4 man-hour search per site (e.g., Tiemann 2006, Karatayev
& Burlakova 2007, Krebs et al. 2010). Metcalfe-Smith et al. (2000) showed that 1.5 man-hour
search failed to adequately compile species lists at 5 rivers in the southwestern Ontario. Tiemann
et al. (2009) also reported that 1 man-hour search for freshwater mussels was insufficient at most
6
reaches of a small Illinois river. These studies suggest that inadequate sampling is common in
time-based mussel surveys, but the extent of under-sampling and particularly its relationship to
the habitat characteristics of a site remain unexamined.
1.4.2 Species richness estimation
To estimate % of species captured, one needs to know how many species actually occur
at a site. However, the true species richness at a site is usually unknown. Many statistical
techniques have been developed to account for potentially missing species (Colwell &
Coddington 1994, Chao 2005). Most of these techniques have been thoroughly evaluated and
several non-parametric methods are often recommended (e.g., Walther & Martin 2001,
Magnussen & Boudewyn 2008), including Chao-l (Chao 1984), Chao-2 (Chao1987), and
Jackknife (Burnham & Overton 1979). Most non-parametric methods require replicate samples
and use the number of rare species to infer the number of missing species. As described later, my
time-based sampling protocol did not yield real replicates. However, one of the recommended
methods, Chao-1, only requires the number of rare species recorded in the whole sample and was
therefore chose for my study.
1.4.3 Standard versus adaptive sampling
A standard sampling effort is often desired for long-term monitoring and large-scale
surveys for assuring data comparability (Metcalfe-Smith et al. 2000, Wisconsin DNR 2005,
Tiemann et al. 2009). Standard search time used in qualitative mussel sampling is, however,
often casual or empirical, and varies among projects and regions (Metcalfe-Smith et al. 2000,
Strayer & Smith 2003, Krebs et al. 2010). In Illinois, a team of three members usually searches
7
for 2~4 man-hours within a 100-300 m stream reach (Illinois Mussel Sampling Protocol 2002).
Sampling adequacies at those sampling efforts remain largely unknown, but are most likely low
and differ among sites, as demonstrated in fish assemblage studies (e.g., Cao et al. 2001, Holtrop
et al. 2010). The among-site variation in sampling adequacy may be related to environmental
characteristics at both watershed and reach scales that affect the detectabilities of mussel
individuals and species (Kovalak et al. 1986, Brim Box & Mossa 1999, Smith et al. 2001,
Gangloff & Feminella 2006, Harriger et al. 2009).
Setting site-specific sampling effort is needed to avoid over- or under-sampling at
individual sites for a standard sampling adequacy (e.g., capturing 70% of all species), i.e.,
adaptive sampling (Holtrop et al. 2010). Modeling the sampling adequacy-environment
relationships provides the possibility of using environmental variables (e.g., stream size and
substrate type) to predict needed sampling efforts for targeted sampling adequacy and thus to
improve the performance of time-based searches. Classic statistical techniques such as multiple
linear regression that have been used to examine the relationships between environmental factors
and mussel assemblage attributes (e.g., density, species richness) (e.g., Strayer 1993, Austin &
Tu 2004) may be not sufficient to model the complex effects of multiple interacting
environmental variables and these techniques are also not applicable if there are more variables
than samples (Copas 1983, Roecker 1991, Strobl et al. 2007). Machine-learning techniques, such
as Random Forests (RF) regression (Breiman 2001), provide alternative ways to model complex
relationships, and they are increasingly adopted in ecological studies (Cutler et al. 2007, Dewalt
et al. 2009, He et al. 2010, Holtrop et al. 2010).
1.5 VARIATION IN MUSSEL SPECIES RICHNESS
8
Understanding the mechanisms that underlie biodiversity among habitats is also critical
in the resource management and conservation. Mussel species diversity can be affected by both
natural environment and human activities. The former may include surface geology in a
watershed (Strayer 1983, Arbuckle & Downing 2002, Krebs et al. 2010) and stream hydrology
and geomorphology (Vaughn & Taylor 2000, Hardison & Layzer 2001, Gangloff & Feminella
2006). The latter includes land-use alternation and water pollution (Allan & Flecker 1993,
Watters 1999, Strayer & Malcolm 2007). For example, dams and impoundments limit the
movements of fish so that mussels could not find hosts and complete the life cycle (Watters 1996,
Galbraith & Vaughn 2011). Channelization and dredging directly reduce or degrade aquatic
habitats, and decrease species diversity (Watters 1999). In addition, fish species composition and
abundance may also affect mussel recruitments and then affect species richness (Vaughn &
Taylor 2000). The RF regression should be useful to model the effects of natural environment
and anthropologic disturbances on mussel species diversity.
1.6 OBJECTIVES
In this study, mussel assemblages were sampled at 18 wadeable-stream sites from 7
basins of Illinois during 2009-2010 with a 16 man-hour search at each site. A range of habitat
characteristics (e.g., channel width, and substrate types) were measured on site, and
environmental factors in riparian zone and watershed, as well as fish-community data, were
compiled from IL-DNR basin survey database (Holtrop et al. 2005, Brenden et al. 2006). In this
study, my main objectives include 1) assessing % of mussel species captured (i.e., sampling
adequacy) at varying sampling efforts at each of the 18 sites, 2) examining the effects of
watershed and reach-habitat characteristics on sampling adequacy, 3) modeling the relationships
9
between the sampling effort needed for a given sampling adequacy and environmental variables,
and 4) examining the effects of environmental variables and fish communities on mussel species
richness.
10
CHAPTER 2
METHODS
2.1 SAMPLING DESIGN
To select a set of representative sampling sites for evaluating sampling adequacy, I first
developed a conceptual model describing sampling adequacy – environment relationships
(Figure 2.1). Many factors may affect sampling adequacy. However, the literature indicates that
three factors appear to be most important: dominant substrate types, species richness, and
watershed size. Substrate composition can affect the detectability of mussel individuals (Brim
Box et al. 2002, Brainwood et al. 2008, Zigler et al. 2008, Harriger et al. 2009) and in turn the
detectability of species. For instance, the detectability of individuals in cobble-dominant streams
is often lower than that in muddy or sandy streams because the size of mussels and cobbles are
equivalent (Brim Box & Mossa 1999). Species-rich assemblages typically contain many rare
species (Kovalak et al. 1986, Cao et al. 1998, Novotny & Basset 2000, Magurran & Henderson
2003), which are likely to be missed (Smith 2006, Kanno et al. 2009) leading to low sampling
adequacy. The number of species at a site is normally unknown before sampling. Fortunately,
Illinois Natural History Survey has mussel records for a large number of stream sites in Illinois,
from which I could get an estimate of species richness at each candidate site. Finally, both
species diversity and habitat complexity often increase with watershed size (Vannote et al.1980,
Strayer 1983, Mcrae et al. 2004, Christian & Harris 2005, Fischer & Paukert 2009). Therefore,
watershed size may strongly affect sampling sufficiency, as shown in fish assemblage surveys
(Angermeier & Winston 1998, Flotemersch et al. 2006, Kanno et al. 2009, Holtrop et al. 2010).
Based on these three factors, I first classified over 1200 candidate sites stored in the IL-
DNR basin survey database (Holtrop et al. 2005) into 12 groups (Table 2.1). All these groups
11
were found in the river basins around Champaign County of Illinois so I decided to restrict the
sampling sites to Vermilion-Wabash valley, Little Vermilion, Sangamon and Embarras River
basins. I randomly selected 1~2 sites from each of the 12 groups and 2 alternative sites in case of
logistic problems or unexpected events (e.g. thunderstorm, high water level). In 2009, 14 sites
based on the design were sampled (Table 2.1). To test the model developed based on the 14 sites,
I chose and sampled 4 additional sites in three river basins beyond Champaign county, i.e.,
Mackinaw, Kaskaskia and Saline River basins in 2010 (Figure 2.3). Field reconnaissance was
always conducted in the week before formal sampling at each site to obtain updated site-specific
information, including water level and accessibility.
2.2 MUSSEL SAMPLING AND DATA COMPILING
All 18 sites were sampled with time-based handpicking technique in June-September.
The sampled reach at a site ranged from 100 to 300 meters, depending on the stream size, water
turbidity and water level. At each site, a team of 6 to 9 people led by at least one experienced
field biologist conducted 16 man-hour search (Figure 2.4) or until no new mussels were collected
in three continuous man-hours. The search period at each site was divided into 4 rounds (i.e., 4
man-hours each) and 16 periods (i.e.,1 man-hour each). The crew started with rapid, wide-range
searches to overview the whole sampling area, and then turned to systematic searches,
particularly at ‘hot spots’ (e.g., riffles, near fallen woods). Crew members walked or crept in the
streams to search for mussels on the stream bottom, and touched the substrate and ran fingers in
the sediment to a depth of 5-10 centimeters to collect small or cryptic individuals. Mussels found
in each man-hour of a round by a crew member were kept in four bags of different colors,
respectively. Mussels in bags of the same color from all crew members were pooled after each
12
round was completed, and were identified and measured (Figure 2.5). The individuals of
federally-listed endangered species and threatened species in Illinois were carefully handled, and
time periods when they were encountered were recorded. All individuals, except for specimens
vouchered for identification confirmation, were returned to stream reaches after sampling
completed. In addition, empty mussel shells were collected whenever seen at each site and 1-2
best-condition shells of each species were kept.
A set of stream characteristics were measured at the end of sampling at a site. First, the
latitude and longitude of the sampled reach were recorded with a GPS unit. We then measured
16 variables (Appendix-D), including water temperature, water clarity, and flow-rate rank (as
described in Wisconsin DNR 2005). We divided the sampled reach into 10 subsections and
obtained 1 measurement on channel width, 3 on water depth (i.e., one point on each bank and
one at middle stream). Ten grabs of substrate were collected from each sub-section to determine
proportion of each substrate type in the sampled stream reach (Figure 2.6). The mean and
standard deviation of these three variables were calculated for further analyses.
I used the IL-DNR basin survey database (Holtrop et al. 2005, Brenden et al. 2006) to
compile data for 80 watershed-level variables (e.g., stream order, catchment area, and land cover
in the riparian zone and watershed). The database summarizes variables for each confluence-to-
confluence stream reach, and provides detailed reach-level data (e.g., riparian zone land-cover)
as well as watershed-level data (e.g., geology and climate). The descriptions of those variables
selected together with habitat characteristics measured on site were summarized in Appendix-D.
I also obtained fish sampling data for most of the 18 sites from IL-DNR fish database, including
species richness and abundances of individual species.
13
2.3 DATA ANALYSES
2.3.1 Terminology
ESRi: Estimated mussel species richness given i man-hours (i = 1, 2, 3…n);
ETSR: Estimated total species richness;
OSRi: Observed mussel species richness given i man-hours (i = 1, 2, 3…n);
OTSR: Observed total species richness;
Fj : Number of species that have j individuals in the whole sample of a site;
SAi: Sampling adequacy with i man-hours (i = 1, 2, 3… n);
MSE: Mean squared error.
2.3.2 Assessing sampling adequacy
In this study, sampling adequacy was measured as the percentage of all species sampled
at a specific sampling effort, calculated as follows:
Sampling adequacy (SAi) = OSRi / ETSR×100% Equation 1
where the OSRi is the observed species richness at i man-hours and ETSR denotes estimated total
species richness based on the whole sample.
2.3.3 Estimations of total mussel species richness
Despite of 16 man-hour search at each site, some rare species were likely to be missed.
To reduce or eliminate this bias, I statistically estimated the true species richness at a site.
Among a large number of methods available (Chao 2005, Colwell 2009), I chose Chao-1 for two
14
reasons. First, this estimator has been reported to outperform many others in empirical
evaluations (e.g., Walther & Martin 2001, Scharff et al. 2003, Magnussen & Boudewyn 2008).
Second, it can be applied to a single sample whereas most other recommended methods require
replicates. Although my sampling was divided into multiple time sections, but those sections
were not area-based replicates and earlier search likely captured more mussels for the same time
length. I calculated Chao-1 using the computer software EstimateS 8.2.0 (Colwell 2009) as:
Equation 2
where OTSR is the observed total species richness at a site, F1 is the number of species recorded
with a single individual only, and F2 is the number of species recorded with exactly two
individuals.
I also examined how much the shell specimen collection helped to reduce the number of
missing species for the standard 4 man-hour search. My question was how many shell-based
species were recorded after the first 4 man hours.
2.3.4 Modeling sampling adequacy-environment relationships
I used Random-Forests (RF) regression to examine the relation of sampling adequacy and
environmental variables. Random Forests shares the basics with classification and regression tree
models (Breiman 1984), which recursively divides a group of samples based on specific
environmental variables to minimize the within-group variance in the dependent variables (i.e.,
sampling adequacy in my case). By bootstrapping a set of environmental variables many times,
15
RF builds many trees with 2/3 of all samples and average the prediction of the other 1/3 samples
(out-of-bag or OOB samples) from each tree as the final prediction (Breiman 2001).
I used two steps to identify candidate predictor variables for sampling-adequacy
modeling. First, I removed the variables with over half missing values and zero values, and
variables that varied narrowly. After this step, 16 habitat variables (e.g., water depth, channel
width, percentage of each substrate type) and 13 watershed-level variables (e.g., stream order,
catchment area, and land cover) were kept. Second, these 29 environmental variables were
subject to screening for importance to sampling adequacy. I used a trial RF model with 5000
trees to rank these variables by computing the change in accuracy of predicting sampling
adequacy (i.e., % increase of MSE) after randomizing the values of a specific variable in the
OOB dataset. The greater is the % of increase of MSE, the more important the variable is.
Nineteen variables that caused negative and low % increase in MSE were dropped.
After variable selection, I tried different numbers of variables used at each division or
split (Mtry), and selected the RF model with the highest pseudo-R2
(i.e., percent variance
explained by regression). The effects of 10 key predictor variables on sampling adequacy were
further examined via partial dependence plots, which excluded the individual and combined
effects of other variables (Cutler et al. 2007). All RF procedures were implemented within R-
package (R Development Core Team 2010) and all models had 5000 classification trees.
Although RF has internal three-fold cross validation and provides the RMSE (root mean
squared error) of the model, an independent validation remains useful (Cutler et al. 2007),
especially when the sampling size is relatively small. Thus, the 4 sites sampled in summer 2010
were used for further model validation. I predicted SA4 (i.e., the sampling adequacy examined in
16
my RF model) at the 4 validation sites and computed the RMSE. The lower the RMSE, the more
accurate the prediction model is. The procedure of computing RMSE is as follows:
1. Calculating error for each data point: error = observed SA4 - predicted SA4;
2. Calculating MSE: MSE = (sum of square error of all data points) / n;
3. RMSE = square root of MSE.
2.3.5 Setting a standard sampling efforts vs. site-specific sampling effort
Sampling efforts needed vary with the objectives and resources available of a specific
project. In this study, I recommended the sampling effort that yielded a significant correlation
(Pearson’s r) between observed and predicted species richness as a standard effort.
I then estimated the sampling effort required for capturing a given % of the predicted
richness at all sites and modeled the relationships of sampling effort and environmental factors in
RF. I chose 70% of all species in my modeling because it is approximately the highest sampling
adequacy reached with 16 man-hour search at all 18 sites, as I showed later. The modeling
procedures, including selecting variables, setting parameters and identifying key variables, were
similar to that of the RF model for sampling adequacy described in section 2.3.4. In addition, I
examined sampling efforts needed to detect endangered species and rare species.
2.3.6 Analyzing variation in mussel species richness
Finally, I examined how environmental variables as well as fish assemblage attributes
(Holtrop et al. 2005) were related to mussel species richness at a site, following the same
procedure of the RF model for sampling adequacy-environment relationship. Fish assemblage
attributes used included the number of fish species, species richness of feeding or habitat-
17
preference groups (e.g., benthic invertivore species, native sucker species), IBI score, and total
abundance.
18
Figures and Tables
Figure 2.1. The conceptual model for sampling design in this study. In the site selection,
historical mussel species richness, substrate type and watershed size were used to represent for
assemblage, habitat and watershed characteristics, respectively.
19
Figure 2.2. Three major types of substrate in Illinois wadeable streams. ‘A’ stands for muddy
sites, ‘B’ stands for sandy sites and ‘C’ denotes cobble-dominant sites.
A
B
20
Figure 2.2 (Cont.)
C
21
Figure 2.3. Locations of 18 sites in wadeable streams in Illinois. Fourteen sites sampled in 2009
(circle) and used for model calibration are located in 4 river basins: Sangamon River, Little
Vermilion River, Wabash River Valley and Embarras River. The other 4 sites (triangle) were
sampled at Mackinaw, Kaskaskia and Saline River basins in 2010 and used for model validation.
22
Figure 2.4. Visual and tactile searching for mussels in a wadeable stream.
23
Figure 2.5. Mussels collected from a man-hour search at Site 1 on Embarras River.
24
Figure 2.6. Measuring habitat characteristics, including water depth, channel width, and substrate
composition, and water temperature at a stream site.
25
Table 2.1. Locations and environmental characteristics and historic species richness at 18 wadeable-stream sites in Illinois.
Site Code Basin/Stream Latitude Longitude Catchment
area (km2)
Historical
species richness*
Dominant
substrate
Yr-2009
BEF-02 1 Embarras/Embarras River 39.2645 -87.9077 383.2 7 sand
BEP-01 2 Embarras/Little Embarras River 39.5962 -88.0446 307.2 13 sand
BEZZ-02 6 Embarras/Brushy Fork 39.7371 -88.0696 315.9 13 sand + cobble
BERB-01 8 Embarras/Hackett Branch 39.7681 -88.2424 113.7 3 gravel
BE-19 9 Embarras/Embarras River 39.7532 -88.1713 489.6 18 silt
EIDE-01 5 Sangamon/ Sugar Creek 40.4012 -89.2366 169.2 8 sand
EZZH-02 7 Sangamon/Dickerson Slough 40.4059 -88.3385 57.4 8 silt
EIEI-01 11 Sangamon/Little Kickapoo Creek 40.3451 -88.9783 74 10 cobble
EID-04 13 Sangamon/Sugar Creek 40.2222 -89.4028 860.2 13 cobble + gravel
EIE-10 14 Sangamon/Kickapoo Creek 40.1683 -89.3868 792.4 15 gravel
BPKP-05 4 Vermilion-Wabash/Big Four Ditch 40.4726 -88.2097 5.5 8 silt
* based in INHS collection database
26
Table 2.1 (Cont.)
BPK-12 12 Vermilion-Wabash/Middle Fork 40.2989 -87.8917 716.3 22 sand + gravel
BO-09 3
Wabash River Valley/
Little Vermilion River 39.9142 -87.7328 231.1 12 sand
BO-08 10
Wabash River Valley/
Little Vermilion River 39.9211 -87.8228 171.2 13 gravel
Yr-2010
DK-20 15 Mackinaw/Mackinaw River 40.6483 -88.8627 695.7 17 sand
DKK-01 18 Mackinaw/Panther Creek 40.6708 -89.1804 494 14 cobble + gravel
OZC-01 16 Kaskaskia/Plum Creek 38.1466 -89.8432 166.6 7 Silt
ATF-03 17 Saline/ Saline River 37.7437 -88.3299 1153.1 17 silt + cobble
27
CHAPTER 3
RESULTS
3.1 MUSSEL ASSEMBLAGES
Thirty four native mussel species were collected at these 18 sites. Species richness ranged
from 5 (Site 5) to 18 (Site 17), and the number of mussels collected ranged from 27 to 942 with
mean of 242 per site (Table 3.1). Mussels were most abundant in Wabash River basin (e.g., Sites
4 and 10). Lampsilis cardium (plain pocketbook), Fusconaia flava (Wabash pigtoe), Lampsilis
siliquoidea (fatmucket) and Lasmigona complanata (white heelsplitter) were most abundant. A
species listed as endangered at the federal level, Potamilus capax (fat pocketbook), was found at
Site 17 in Saline River. Three species listed as threatened in Illinois (Alasmidonta viridis, Villosa
lienosa and Cyclonaias tuberculata) were found at 7 sites (Table 3.4). An invasive species,
Asian clams (Corbicula fluminea), were found at 11 out of 18 sites and particularly abundant at
Sites 3, 7 and 8, but no zebra mussels were encountered at any site.
3.2 SAMPLING ADEQUACY
3.2.1 Species accumulation and Chao-1 estimates
The pattern of species accumulation differed considerably among the 18 sites. At Sites 1-
7 and 15-18, the majority of mussel species were captured rapidly during the first 4 man-hour
search, for which the richness estimates by Chao-1 (ETSR) was identical to or only slightly
higher than the observed species richness (OTSR). In comparison, accumulations were slow at
the other 7 sites (Sites 8-14), where many individuals were recorded in the last two time
segments of sampling with many singleton species. As a result, ETSR was much higher than the
OTSR at these sites (Table 3.1).
28
3.2.2 Sampling adequacy
Sampling adequacy was strongly affected by sampling effort. When sampling effort
increased from 4 to 16 man-hours, sampling adequacy on average increased from 61% to 89%
(Table 3.2). However, the rate of increase varied substantially among sites. For example,
sampling adequacy increased by 55% at Site 14 but increased by only 14% and 20% at Sites 8
and 10, respectively (Table 3.2). Empty-shell-based species were confirmed at 10 sites, an
average of 2 species were found after 4 man-hour search, indicating that shells could add
information on both species richness and composition to 4 man-hour samples.
Four man-hours, a standard effort in Illinois, captured 15%-100% of all species at the 18
sites (mean = 61%, standard deviation = 25%). SA4 reached 70% at Sites 1-5 and 17-18, which
were mostly small stream sites with fine substrate. In comparison, SA4 was particularly low (<
40%) at Sites 10-14 located in gravel- or cobble-dominant reaches with high mussel diversity
(Table 3.2). Sampling adequacy even varied greatly among adjacent sites. For example, both
Sites 5 and 13 are located in the Sangamon River and are close to each other (4 km), but SA4 at
Site 5 was higher than that at Site 13 by 50%.
3.2.3 Modeling sampling adequacy
The RF model based on 10 variables (Figure 3.1) accounted for 45% of the total variance
of sampling adequacy for the 14 sites sampled in 2009. Parameters of this RF model were set as
number of trees = 5000, node size = 5 and Mtry = 3. The RMSE was 19.7% for the 14 sites
sampled in 2009, compared with 21.1% at the 4 sites sampled in 2010 for model validation.
There was little evidence of model over-fitting.
29
Partial dependence plots showed that sampling adequacy at 4 man-hours decreased with
increasing catchment area and the sum of upstream links, % of open water in the local riparian
zone, and % of points dominated by gravels and cobbles (Figure 3.2 A-E). In contrast, it
increased with % of upland forests in the local riparian zone, % of sand-dominant points, and %
of log-dominant points in the stream (Figure 3.2 G-H). Sampling adequacy at 4 man-hours also
was negatively correlated with predicted species richness (Pearson’s r = - 0.57, p < 0.05) and
historical species richness (Pearson’s r = - 0.37, p = 0.19), an observation that agrees with the
conceptual model (Figure 3.3). Additionally, sampling adequacy was negatively correlated to the
number of singletons (i.e., species with only one individual in all samples) (Pearson’s r = 0.79, p
< 0.05) (Figure 3.4).
3.3 SAMPLING EFFORTS FOR SPECIFIC SAMPLING ADEQUACY
3.3.1 Standardizing sampling efforts
Observed species richness (OSRi) was not significantly correlated with the estimated
species richness (ETSR) until 8 man-hours (Pearson’s r = 0.59, p < 0.05), implying smaller
samples would not only under-estimate the total species richness, but also fail to rank the sites
for species richness. The correlation became strong at 10 man-hours (Pearson’s r 0.78, p <
0.01) (Figure 3.5). The Chao-1 estimate of species richness at 4 man-hours (ESR4) was also
strongly correlated with ETSR (Pearson’s r = 0.80, p < 0.01). These observations implied that
richness based on 10 man-hour search (Figure 3.5) or the statistical estimates of species richness
based on 4 man-hour search would be sufficient to rank mussel habitats for diversity.
3.3.2 Predicting sampling efforts required for a targeted SA
30
The sampling efforts required for sampling adequacy (SA) of 70 % ± 3% varied among
these sites. Another RF was developed to examine the effects of environmental factors on
sampling effort needed at the 14 calibration sites (pesduo-R2
= 41%, and RMSE = 3.15, similar to
that of using RF to predict sampling effort needed at the 4 sites sampled in 2010 (RMSE = 3.23
man-hour) (Figure 3.6). This RF model incorporated 10 predictor variables (Figure 3.7).
Sampling effort required for SA of 70% increased with key variables including catchment area,
water depth, % of gravel-dominant points and % of cobble-dominant points in streams (Figure
3.8).
3.3.3 Detecting species of concern
More than 4 man-hours were needed to detect most endangered species (Table 3.4). For
instance, a federally endangered species (Potamilus capax) was not encountered until the 8th
man-hour at Site 17 in the North Fork Saline River. The detection of two threatened species in
Illinois, Alasmidonta viridis and Cyclonaias tuberculata, also required an effort over 4 man-
hours (Table 3.4). Sspecies with more than 18 individuals were always detected at the first man-
hour. With 4 man-hours, some rare species (1-3 individuals), such as Lasmigona complanata,
Lampsilis siliquoidea and Amblema plicata, could be detected (Figure 3.9). In contrast, some
common species, Toxolasmas parvus, Lasmigona compressa, and Pleurobema sintoxia were
hard to be found with 4 man-hours, probably because of their small sizes (1.9-5.6 cm). These
results indicate that species abundance is not the only factor that affect species detectability.
3.4 MODELING MUSSEL SPECIES RICHNESS
A RF model was developed to relate estimated mussel species richness (ETSR) to stream
reach and watershed environment (Pseduo-R2
= 38% at Mtry = 3). The proportion of open water
31
in the local riparian zone, riparian zone area, length of upstream channels, and % of non-row
croplands in the local riparian zone were the most important predictors, but none of fish
assemblage characteristics were rated as important. The length of upstream channels and total
riparian zone area, were positively correlated to mussel species richness (Figure 3.10-A, B).
More freshwater mussel species appeared to occur when riparian zones contained high
proportion of open water, but less agricultural land (Figure 3.10-C, D).
32
Figures and Tables
Figure 3.1. The importance of 10 environmental variables used to model sampling adequacy at 4
man-hours, measured with % Increase of MSE when the values of a variable in the OOB samples
are randomized (see Appendix-D for variable description).
33
Figure 3.2. Random-Forests Partial dependence plots showing the change in sampling adequacy
as each of 8 key variable increases when the individual and combined effects of all other
variables are excluded at 14 wadeable-stream sites in Illinois.
34
Figure 3.2. (Cont.)
35
Figure 3.3. Correlation between sampling adequacy at 4 man-hours and historical species
richness (black triangles) and predicted total species richness (red circles) across 18 sites in
Illinois.
36
Figure 3.4. Correlation between sampling adequacy at 4 man-hours and number of singletons,
i.e., species with only one individual in all samples.
37
Figure 3.5. Correlations of estimated total mussel species richness and observed richness at 4 and
10 man-hours (Panel A and B respectively). A solid line represents for 1:1 line and a dash line
represents for fitted regression line.
38
Figure 3.6. Observed and predicted sampling efforts needed for sampling adequacy (SA) of 70%
at the 14 modeling sites (circles) sampled in 2009 and 4 validation sites (dots) sampled in 2010.
The predicted values are based the Random-Forests regression that models relationships of
sampling efforts needed for SA of 70% and environmental variables.
39
Figure 3.7. Importance of predictor variables in predicting the number of man-hours required for
capturing 70 3% of all species in Random-Forests regression.
40
Figure 3.8. Random-Forests Partial dependence plots showing the change in the number of man-
hours required to capture 703% of all species at 14 wadeable-stream sites in Illinois as each of
4 key variable increases with the individual and combined effects of all other variables excluded.
41
Figure 3.9. Sampling efforts required to detect a species and the abundance of the species at a
site with all species with > 18 individuals (detectable at 1 man-hours) were excluded. Species
within Area 1 were easy to detect though rare (≤ 3 individuals), whereas species in Area 3 were
hard to detect (e.g., Toxolasmas parvus).
42
Figure 3.10. Random-Forests Partial dependence plots showing the change in the number of
species at 14 wadeable-stream sites in Illinois as each of 4 key variable increases with the
individual and combined effects of all other variables excluded.
43
Table 3.1. Total number of mussels collected with 16 man-hour search (Abundance), total
species richness observed (OTSR), the number of singletons (F1) and doubletons (F2), and Chao-
1 species richness estimate (ETSR) (See Table 2.1 for site descriptions).
Sites Abundance OTSR F1 F2 ETSR
1 206 6 0 1 6
2 76 6 0 1 6
3 111 8 1 0 8
4 873 13 2 3 13
5 425 11 2 0 12
6 44 11 3 2 12
7 135 6 1 0 6
8 34 5 2 1 7
9 161 12 3 2 13
10 942 9 4 0 15
11 153 10 3 2 12
12 178 14 3 1 18
13 53 9 4 2 11
14 27 9 4 2 13
15 168 10 1 0 10
16 45 6 1 2 6
17 573 18 4 0 19
18 128 13 2 2 13
44
Table 3.2. Sampling adequacy measured with % of species captured with 2-16 man-hour search
at the 18 wadeable-stream sites in Illinois (See Table 2.1 for site descriptions).
Site SA2 SA4 SA6 SA8 SA10 SA12 SA14 SA16
1 83 100 100 100 100 100 100 100
2 50 100 100 100 100 100 100 100
3 50 88 88 88 88 100 100 100
4 70 77 85 92 92 92 92 92
5 75 75 75 75 75 92 92 92
6 42 67 75 83 83 92 92 92
7 50 67 67 67 67 83 100 100
8 43 57 57 71 71 71 71 71
9 31 46 70 77 77 85 92 92
10 33 40 53 53 53 60 60 60
11 42 42 42 58 67 75 83 83
12 22 28 33 33 56 67 78 78
13 27 27 27 55 73 82 81 81
14 8 15 23 31 46 54 54 70
15 70 90 90 90 90 90 90 100
16 33 50 83 83 100 100 100 100
17 47 58 74 79 84 89 89 95
18 54 62 92 100 100 100 100 100
Mean 46 61 69 74 79 85 87 89
45
Table 3.3. A summary of the shell data collected at 10 wadeable-stream sites in Illinois (See
Table 2.1 for sites descriptions).
Sites Total number of
shell-species
Number of shell-species recorded
after the first 4 man-hours
1 5 0
2 1 0
4 2 2
6 8 2
7 5 2
12 5 3
14 6 5
15 3 1
16 4 0
17 11 4
18 6 2
46
Table 3.4. Time-sections when threatened and endangered species were recorded at 18 wadeable-
stream sites in Illinois.
Species Site Abundance Time-section
encountered
Missed if only 4
man-hours
Alasmidonta viridis 7 1 11 Yes
Alasmidonta viridis 10 1 2 No
Alasmidonta viridis 12 1 14 Yes
Cyclonaias tuberculata 12 3 9, 10, 14 Yes
Potamilus capax *
17 3 8, 10, 12 Yes
Villosa lienosa 3 4 2, 3, 5 No
Villosa lienosa 4 73 1-15 No
Villosa lienosa 8 1 3 No
Villosa lienosa 10 39 1-12, 15-16 No
Villosa lienosa 12 2 10, 16 Yes
Note: *denotes federally-listed endangered species, while others are considered
threatened in Illinois according to Illinois Endangered Species Protection Board (2010).
47
CHAPTER 4
DISCUSSION
Evaluating the adequacy of time-based mussel sampling and understanding the
environmental gradients that underlie mussel diversity are critical for assuring data quality in
mussel surveys and protecting mussel biodiversity. In the present study, mussel assemblages
were sampled at 18 sites in 7 basins that differed in species diversity, watershed and habitat
characteristics. Intensive sampling, combined with the use of Chao-1 estimator, can be expected
to yield accurate estimates of species richness, something essential to address both questions
above. More important is that key environmental factors can be associated sampling adequacy
and species richness respectively. The former would serve as a framework for setting site-
specific efforts in adaptive sampling, and the latter could help locate current ‘hot spots’ of
mussel diversity in the study area.
4.1 MEASURING SAMPLING ADEQUACY
In this study, I focused on species richness for assessing sampling adequacy. Species
richness is a central concept of biodiversity conservation (e.g., Williams et al. 1993) and a
widely used indicator of biological integrity (Karr & Chu 1999). The implications of % of
species richness sampled as the measure of sampling adequacy go beyond species richness per se.
As demonstrated in Cao et al. (2002, 2004), % of species recorded at a site also indicates how
well a sample characterizes species composition in an assemblage. Holtrop et al. (2010) further
reported that the sampling efforts required for a given sampling adequacy in species richness was
also sufficient to achieve the same adequacy in estimating site-to-site similarity in species
composition. Therefore, it is a simple, but highly informative measure of sample adequacy. My
48
findings in the present study can be used to guide stream mussel surveys in two different ways, 1)
setting a standard sampling effort for wadable streams in Illinois and potentially other part of the
Midwest, and 2) setting site-specific sampling effort among different habitats, as discussed in
Section 4.1.1 and Section 4.1.2, respectively.
4.1.1 Setting a standard sampling effort
If standardizing sampling effort is desired, as it is in most large-scale biological
monitoring programs, 10 man-hours per site appear adequate in stream mussel surveys in Illinois.
Ten man-hour searches captured over 70% of mussel species at most sites in this study, and
yielded strong correlations between observed richness and the predicted total species richness
(Pearson’s r = 0.78, p < 0.01). In comparison, the standard efforts used in previous studies are
much lower (e.g., 4 man-hours in Illinois, 1.5 man-hours in Ontario). I recognize that this
recommended effort is not always affordable for large-scale surveys and propose two options for
reducing the effect of under-sampling. First, Chao-1 method could be used to improve the
estimate of richness. As shown earlier, Chao-1 estimates at 4 man hours were strongly correlated
with the estimates from all 16 man-hour search (Pearson’s r = 0.84, p < 0.01). Second, empty
mussel shells may also help better estimate species richness. In this study, nearly one third of
shell-based species on average were recorded after 4 man-hours, implying that shells were useful
to detect the presence of mussel species missed.
4.1.2 Effects of environmental variables on sampling adequacy and adaptive sampling
The present study identified a set of environmental variables strongly associated with
sampling adequacy. Several variables were negatively related to sampling adequacy. First,
49
sampling adequacy decreased with increasing catchment area (Figure 3.2-B), as assumed in the
study design. Larger watershed generally support higher mussel diversity and contain more
heterogeneous habitats (e.g., Magurran & Henderson 2003, Gangloff & Feminella 2006), which
would make it hard to reach high sampling adequacy. Second, sampling adequacy also decreased
with % of water in the riparian zone (e.g., wetlands, ponds) (Figure 3.2-C), possibly because
adjacent waterbodies may add more species to the species pool available for the stream site and
then reduce sampling adequacy. Third, sampling adequacy decreased with increasing substrate
size (Figure 3.2-D, E, G). Substrate composition is known to affect the detectability of individual
species (Brim Box & Mossa 1999, Smith et al. 2000, Smith 2006). Based on my field experience,
sampling crew often have to spend more time to distinguish mussels from cobbles, and their
fingers were easily getting numb when running into coarse substrates for several hours, resulting
in lower sampling adequacy.
In contrast, sampling adequacy increased with % of log in substrates. My crew members
and I often found more mussels near fallen woods (from the riparian forests). This may be
because logs can stabilize substrates (Benke & Wallace 2003, Golladay 2004, Harriger et al.
2009) and provide complex micro-habitats that may reduce the risk of mussels being predated by
mammals (e.g., raccoons and muskrats) and birds (ducks) (Strayer 2008).
Once key predictor variables were identified, one can use a statistical model to estimate
the site-specific sampling efforts needed for a specific sampling adequacy. The model I
developed for sampling adequacy of 70% accounted for > 40% of total variance, providing some
solid base for implementing adaptive sampling in mussel surveys. The performance of the RF
model could be improved in three ways. First, more sampling sites would help, covering more
types of habitats and assemblages and in turn performing better across the state. Second, more
50
environmental variables should be incorporated into the model, such as ratio of riffle/pool and
flow characteristics predicted from hydrology model (Carlisle et al. 2010). Third, several key
predictor variables in the RF model, such as dominant-substrate types are available for those
streams sites used by Illinois DNR –EPA monitoring programs, but not available for other sites.
However, such variables may be modeled based on watershed-level variables that are widely
available, including geology, soil, and land-cover (Woodcock et al. 2006, Newton et al. 2008).
Advanced technologies may also allow one to quickly gather reliable local-habitat data. For
example, side-scan sonar and fine-resolution remote sensing can be used to estimate substrate
types or riffle/pool ratios (e.g., Feurer et al. 2008, Kaeser & Litts 2010).
Additionally, one also can apply model-based adaptive scheme to other types of mussel
sampling techniques (e.g., snorkeling in non-wadeable rivers and lakes) as well as other
taxonomic groups, such as time-based electrofishing or seine netting for crayfish (Westman et al.
1978, Price & Welch 2009).
4.2 MUSSEL RICHNESS-ENVIRONMENT RELATIONSHIPS
Understanding diversity-environment relationships is critical to make strategies in the
conservation initiatives and resource management. In this study, the 16 man-hour search per site
coupled with Chao-1 estimator provided a solid basis to examine the relationships, which helped
to identify critical habitats or ‘hot spot’ of freshwater mussels. The positive correlation of mussel
species richness with watershed size (Figure 3.10) observed in the present study is supported by
several previous studies (e.g., Strayer 1983, Gangloff & Feminella 2006). Additionally, mussel
diversity increased with riparian zone area, likely because larger riparian buffers can mitigate
disturbances from the watershed on mussel assemblages (Morris & Corkum 1996, Poole et al.
51
2004). The positive effects of open water in the riparian zone on mussel diversity may be
because the presence of the water bodies is potentially associated with stable and high base-flow,
something important for mussels to survive (Golladay et al. 2004, Haag & Warren 2008).
Interestingly, fish assemblage attributes did not affect mussel richness in my analysis, despite the
close relationship between fish and mussel that is well documented (Watters 1996, Vaughn &
Taylor 2000). The lack of fish data for 4 of the 18 sites in my analysis may partly account for the
weak relationship, but the mobility of fish species may have larger effects. Average fish richness
and density over multiple years may be more relevant to mussel diversity and should be
examined in the future studies.
Furthermore, the RF model identified some key stressors of mussel assemblages at the
sampling sites. The negative effects of agricultural land use on richness observed in the RF
model may be related to siltation and channelization (McMahon 1991, McGregor & Garner
2003). Best-management practices (Weigel et al. 2000, Broadmeadow & Nisbet 2004), including
reduction of soil erosion and restoration of riparian vegetation and wetlands, should help to
maintain or restore stream mussel biodiversity.
52
CHAPTER 5
SUMMARY
Effective management and conservation of freshwater mussels require not only reliable
estimates of richness and distribution, but also clear understanding of the sampling adequacy-
environment relationships. Key findings in this study include:
1) The number of species recorded at a site varied between 5 and 18 with totally 27-942
live individuals. To account for possible missing species, I estimated the total richness using
Chao1 method, which yielded 0~3 more mussel species than the observed total species richness.
2) The commonly used sampling effort, 4 man-hours per site, captured ≥70% of species
at 36% of the sites. An eight man-hour search is recommended because the total species richness
became significantly correlated with the observed species richness at this effort.
3) A Random-Forests (RF) model based on watershed and habitat characteristics (e.g.,
stream size and dominant-substrate types) accounted for 45 % of the total variance in sampling
adequacy at 4 man-hours. Sampling adequacy decreased with increasing stream size and
substrate size, but increased with the proportion of upland forests in the riparian zone and fallen
woods in streams.
4) Random-Forests regression indicated that more freshwater mussel species were present
at sites located in relatively larger watersheds with higher proportion of open water. In
comparison, the proportion of agricultural area in the local riparian zone exerted the opposite
effect on mussel species richness.
5) Conclusively, findings in this study should serve as a guide for setting standard
sampling efforts for mussel surveys in Illinois and likely other Midwest states, and provide
critical information for setting site-specific efforts toward adaptive sampling.
53
REFERENCES
Allan, J.D. & A.S. Flecker. 1993. Biodiversity conservation in running waters: Identifying the
major factors that threaten destruction of riverine species and ecosystems. BioScience
43:32-43.
Altieri, M.A.1999. The ecological role of biodiversity in agroecosystems. Agriculture,
Ecosystems and Environment 74:19-31.
Angermeier, P.L. & M.R. Winston. 1998. Local vs. regional influences on local diversity in
stream fish communities of Virginia. Ecology 79:911-927.
Arbuckle, K.E. & J.A. Downing. 2002. Freshwater mussel abundance and species richness: GIS
relationships with watershed land use and geology. Canadian Journal of Fisheries and
Aquatic Sciences 59:310-316.
Austin, P.C. & J.V. Tu. 2004. Bootstrap methods for developing predictive models. The
American statistician 58:131-137.
Baker, F.C. 1916. The relation of mollusks to fish in Oneida Lake. New York State College of
Foresty and Syracuse University Technical Publication I 366, Syracuse, NY.
Baker, S.M. & J.S. Levinton. 2003. Selective feeding by three native North American freshwater
mussels implies food competition with zebra mussels. Hydrobiologia 505:97-105.
Balogh, K.V. 1988. Comparison of mussels and crustacean plankton to monitor heavy metal
pollution. Water, Air, and Soil Pollution 37:281-292.
Benke, A.C. & B.J. Wallace. 2003. Influence of wood on invertebrate communities in streams
and rivers. In: S.V. Gregory, K.L. Boyer and A.M. Gurnell, Editors, The Ecology and
Management of Wood in World Rivers. American Fisheries Society, Symposium 37
Bethesda, MD.
Berrow, S.D., T.C. Kelly & A.A. Myers. 1991. Crows on estuaries: distribution and feeding
behaviour of the Corvidae on four estuaries in southwest Ireland. Irish Birds 4:393-412.
Brainwood, M., S. Burgin & M. Byrne. 2008. The role of geomorphology in substratum patch
selection by freshwater mussels in the Hawkesbury-Nepean River (New South Wales)
Australia. Aquatic Conservation: Marine and Freshwater Ecosystems 18:1285-1301.
Brenden, T.O., R.D. Clark, Jr., A.R. Cooper, P.W. Seelbach, L. Wang, S.S. Aichele, E.G. Bissell
& J.S. Stewart. 2006. A GIS framework for collecting, managing, and analyzing multi-
scale landscape variables across large regions for river conservation and management.
Pages 49-74 in R.M. Hughes, L. Wang, and P.W. Seelbach, editors. Influences of
54
landscapes on stream habitats and biological assemblages. American Fisheries Society,
Symposium 48, Bethesda, Maryland.
Breiman, L., J. Friedman, C. Stone & R. Olshen. 1984. Classification and Regression Trees.
Chapman and Hall, New York.
Breiman, L. 2001. Random forests. Machine Learning 45:15–32.
Brim Box, J. & J. Mossa. 1999. Sediment, Land Use, and Freshwater Mussels: Prospects and
Problems Journal of the North American Benthological Society 18:99-117.
Brim Box, J., R.M. Dorazio & W.D. Liddell. 2002. Relationships between streambed substrate
characteristics and freshwater mussels (Bivalvia: Unionidae) in Coastal Plain streams.
Journal of the North American Benthological Society 21:253-260.
Broadmeadow, S & T.R. Nisbet. 2004. The effects of riparian forest management on the
freshwater environment: A literature review of best management practice. Hydrology and
Earth System Sciences 8:286-305.
Burnham, K.P. & W.S. Overton. 1979. Robust estimation of population size when capture
probabilities vary among animals. Ecology 60:927-936.
Carlisle, D.M., J. Falcone, D.M. Wolock, M.R. Meador & R.H. Norris. 2006. Predicting the
natural flow regime: Models for assessing hydrological alteration in streams. River
Research and Applications 6:118-136.
Cao, Y, D.D. Williams & N.E. Williams. 1998. How important are rare species in aquatic
community ecology and bioassessment? Limnology and Oceanography 43:1403-1998.
Cao, Y., D.P. Larsen & R.M. Hughes. 2001. Evaluating sampling sufficiency in fish assemblage
surveys: a similarity-based approach. Can. J. Fish. Aquat. Sci. 58:1782–1793.
Cao, Y., D.D. Williams & D.P. Larsen. 2002. Comparison of ecological communities: the
problem of sample representativeness. Ecological Monographs 72:41-56.
Cao, Y., D.P. Larsen & D. White. 2004. Estimating regional species richness using a limited
number of survey units. Ecoscience 11:23-35.
Chao, A. 1984. Non-parametric estimation of the number of classes in a population.
Scandinavian Journal of Statistics 11:265-270.
Chao, A. 1987. Estimating the population size for capture-recapture data with unequal
catchability. Biometrics 43:783-791.
Chao, A. 2005. Species richness estimation. In Encyclopedia of Statistical Sciences,
N.Balakrishnan, C. B.Read, and B.Vidakovic (eds), 2nd edition. Wiley, New York.
Christian, A.D. & J.L. Harris. 2005. Development and assessment of a sampling design for
mussel assemblages in large streams. American Midland Naturalist 153:284-292.
Claassen, C. 1994. Washboards, pigtoes and muckets: historic musseling in the Mississippi
watershed. Historical Archaeology 28:1-145.
55
Colwell, R.K. & J.A. Coddington. 1994. Estimating terrestrial biodiversity through extrapolation.
Philosophical Transactions of the Royal Society 345:101-118.
Colwell, R.K. 2009. EstimateS: Statistical estimation of species richness and shared species from
samples. Version 8.2. User's Guide and application published at:
http://purl.oclc.org/estimates.
Copas, J.B. 1983. Regression, prediction and shrinkage. J. Roy. Statist. Soc. Series B. 45:311-
354.
Cossu, C., A. Doyotte, M. Babut, A. Exinger & P. Vasseur. 2000. Antioxidant biomarkers in
freshwater bivalves, Unio tumidus, in response to different contamination profiles of
aquatic sediments. Ecotoxicology and Environmental Safety 45:106-121.
Cutler, D.R., T.C. Edwards Jr., K.H. Beard, A. Cutler, K.T. Hess, J. Gibson & J.J. Lawler. 2007.
Random Forests for classification in Ecology. Ecology 88:2783-2792.
Dewalt, R.E., Y. Cao, L. Hinz & T. Tweddale. 2009. Modelling of historical stonefly
distributions using museum specimens. Aquatic Insects 31:253-267.
Diamond, J.M. & 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.
Draulans, D. 1982. Foraging and Size Selection of Mussels by the Tufted Duck, Aythya fuligula.
Journal of Animal Ecology 51:943-956.
Feurer, D, J.S. Bailly, C. Puech, Y.L. Coarer & A.A. Viau. 2008. Very-high-resolution mapping
of river-immersed topography by remote sensing. Process in Physical Geography 32:403-
419.
Fischer, J.R. & C.P. Paukert. 2009. Effects of sampling effort, assemblage similarity, and habitat
heterogeneity on estimates of species richness and relative abundance of stream fishes.
Canadian Journal of Fisheries and Aquatic Sciences 66:277-290.
Flotemersch, J.E., J.B. Stribling & M.J. Paul. 2006. Concepts and approaches for the
bioassessment of nonwadeable streams and rivers. U.S. Environmental Protection Agency,
Report 600/R-06/127, Washington, D.C.
Galbraith, H.S. & C.C. Vaughn. 2011. Effects of reservoir management on abundance, condition,
parasitism and reproductive traits of downstream mussels. River Research and
Applications 27:193-201.
Gamfeldt, L., H. Hillebrand & P.R. Jonsson.2008. Multiple functions increase the importance of
biodiversity for overall ecosystem functioning. Ecology 89:1223-1231.
Gangloff, M.M. & J.W. Feminella. 2006. Stream channel geomorphology influences mussel
abundance in southern Appalachian streams, U.S.A. Freshwater Biology 52:64-74.
56
Gaston, K.J. & J.I. Spicer. 2004. Biodiversity: An Introduction, Second Edition. Wiley-
Blackwell. Hoboken.
Gatenby, C.M., D.M. Orcutt, D.A. Kreeger, B.C. Parker, V.A. Jones & R.J. Neves. 2003.
Biochemical composition of three algal species proposed as food for captive freshwater
mussels. Journal of Applied Phycology 15:1-11.
Giam, X., T.H. Ng, V.B. Yap & H.T.W. Tan. 2010. The extent of undiscovered species in
Southeast Asia. Biodiversity and Conservation 19:943-954.
Giller, P.S., H. Hillebrand, U.G. Berninger, M.O. Gessner, S. Hawkins, P. Inchausti, C. Inglis,
(...), G. O'Mullan. 2004. Biodiversity effects on ecosystem functioning: Emerging issues
and their experimental test in aquatic environments. Oikos 104:423-436.
Golladay, S.W., P. Gagnon, M. Kearns, J.M. Battle & D.W. Hicks. 2004. Response of freshwater
mussel assemblages (Bivalvia: Unionidae) to a record drought in the Gulf Coastal Plain of
southwestern Georgia. Journal of the North American Benthological Society 23:494-506.
Gutierrez, J.L., C.G. Jones, D.L. Strayer & O.O. Iribarne. 2003. Mollusks as ecosystem
engineers: the role of shell production in aquatic habitats. Oikos 101:79–90.
Haag, W.R., R.S. Butler & P.D. Hartfeld. 1995. An extraordinary reproductive strategy in
freshwater bivalves: prey mimicry to facilitate larval dispersal. Freshwater Biology 34:471-
476.
Haag, W.R. & L.J. Staton. 2003. Variation in fecundity and other reproductive traits in
freshwater mussels. Freshwater Biology 48:2118–2130.
Haag, W.R. & M.L. Warren, Jr. 2008. Effects of severe drought on freshwater mussel
assemblages. Transactions of the American Fisheries Society 137:1165-1178.
Hardison, B.S. & J.B. Layzer. 2001. Relations between complex hydraulics and the localized
distribution of mussels in three regulated rivers. River Research and Applications 17:77-
84.
Harriger, K., A. Moerke & P. Badra. 2009. Freshwater mussel (Unionidae) distribution and
demographics in relation to microhabitat in a first-order Michigan stream. The Free
Library. Retrieved March 15, 2010 from http://www.thefreelibrary.com/Freshwater mussel
(Unionidae) distribution and demographics in...-a0218112262.
Hastie, L.C., P.J. Cosgrove, N. Ellis & M.J. Gaywood.2003. The Threat of Climate Change to
Freshwater Pearl Mussel Populations. Ambio 32:40-43.
He, Y., J. Wang, S. Lek-Ang & S. Lek. 2010. Predicting assemblages and species richness of
endemic fish in the upper Yangtze River. Science of the Total Environment 408:4211-
4220.
Hellmann, J.J. & G.W. Fowler. 1999. Bias, precision & accuracy of four measures of species
richness. Ecological Applications 9:824-834.
57
Herkert, J.R., ed. 1992. Endangered and threatened species of Illinois: status and distribution.
Volume 2 animals. Illinois Endangered Species Protection Board, Springfield.
Hobohm, C. 2003. Characterization and ranking of biodiversity hotspots: centres of species
richness and endemism. Biodiversity and Conservation 12:279-287.
Holtrop, A., D. Day, C. Dolan & J. Epifanio. 2005. Ecological classification of rivers for
environmental assessment and management: stream attribution and model preparation.
Illinois Natural History Survey Technical Report 2005/04, Springfield, Illinois.
Holtrop, A.M., Y. Cao & C.R. Dolan. 2010. Estimating Sampling Effort Required for
Characterizing Species Richness and Site-to-Site Similarity in Fish Assemblage Surveys of
Wadeable Illinois Streams. Transactions of the American Fisheries Society 139:1421-1435.
Hooper, D.U., F.S. Chapin III, J.J. Ewel, A. Hector, P. Inchausti, S. Lavorel, J.H. Lawton, (...),
D.A. Wardle. 2005. Effects of biodiversity on ecosystem functioning: A consensus of
current knowledge. Ecological Monographs 75:3-35.
Hornbach, D. J. & T. Deneka. 1996. A comparison of a qualitative and a quantitative collection
method for examining freshwater mussel assemblages. Journal of the North American
Benthological Society 15:587-596.
Howard, J.K. & K.M. Cuffey. 2006. The functional role of native freshwater mussels in the
fluvial benthic environment. Freshwater Biology 51:460-474.
Illinois Endangered Species Protection Board. 2010. Checklist of Endangered and Threatened
Animals and Plants of Illinois. Illinois Endangered Species Protection Board, Springfield,
Illinois.
Kaeser, A.J. & T.L. Litts. 2010. A novel technique for mapping habitat in navigable streams
using low-cost side scan sonar. Fisheries 35:163-174.
Kanno, Y., J.C. Vokoun & D.C. Dauwalter. 2009. Influence of Rare Species on Electrofishing
Distance When Estimating Species Richness of Stream and River Reaches. Transactions
Of The American Fisheries Society 138:1240-1251.
Karatayev, A.Y & Burlakova L.E. 2007. East Texas Mussel Survey Final Report.
Karr, J.R. & E.W. Chu. 1999. Restoring Life in Running Waters - Better Biological Monitoring.
Island Press, Covelo, CA.
Keller, A, M. Lydy, D.S. Ruessler. 2006. Unionid mussel sensitivity to environmental
contaminants. In Farris JL, Van Hassel JH, eds, Freshwater Bivalve Ecotoxicology.
SETAC, Pensacola, FL, USA.
Kodric-Brown, A. & J.H. Brown. 1993. Incomplete data sets in community ecology and
biogeography: a cautionary tale. Ecological Applications 3:736-742.
Kovalak, W.P., S.D. Dennis & J.M. Bates. 1986. Sampling effort required to find rare species of
freshwater mussels, p. 34–45. Nib. G. Isom (ed.). Rationale for sampling and interpretation
58
of ecological data in the assessment of freshwater ecosystems. ASTM STP 894. American
Society for Testing and Materials, Philadelphia, Pennsylvania.
Krebs, R.A., W.C. Borden, E.R. Steiner, M.S. Lyons, W. Zawiski & B.M. Walton. 2010.
Determinants of mussel diversity in Lake Erie tributaries. Journal of the North American
Benthological Society 29:506-520.
Kryger, J. & H.U. Riisgard. 1988. Filtration rate capacities in 6 species of European freshwater
bivalves. Oecologia (Berlin) 77:34–38.
Lydeard, C., R.H. Cowie, W.F. Ponder, A.E. Bogan, P. Bouchet, S.A. Clark S.A., K.S.
Cummings, (...), F.G. Thompson. 2004. The global decline of nonmarine mollusks.
BioScience 54:321-330.
Lynch, J.A., J.H. Patric & G.M. Aubertin. 1977. Forests and water quantity. In: Sopper, W.E., et
al., eds. Water resources at the forest-urban interface. U.S. Department of Agriculture,
Forest Service, Pinchot Institute Environmental Forest Study.
Mackie, G.L. 1991. Biology of the exotic zebra mussel, Dreissena polymorpha, in relation to
native bivalves and its potential impact in Lake St. Clair. Hydrobiologia 219:251-268.
Magnussen, S. & P. Boudewyn. 2008. Estimators of tree species richness: an assessment for
Central and Eastern Canada. J. Sust. For. 26:77-96.
Magurran, A.E. & P.A. Henderson. 2003. Explaining the excess of rare species in natural species
abundance distributions. Nature (London) 422:714–716.
McGregor, S.W. & J.T. Garner. 2003. Changes in the freshwater mussel (Bivalvia: Unionidae)
fauna of the Bear Creek system of Northwest Alabama and Northeast Mississippi.
American Malacological Bulletin 18:61-70.
McMahon, R.F. 1991. Mollusca: Bivalvia. In Ecology and classification of North American
freshwater invertebrates. Edited by J.H. Thorp and A.P. Covich. Academic Press, Inc. San
Diego, CA.
Mcrae, S.E., J.D. Allan & J.B. Burch. 2004. Reach- and catchment-scale determinants of the
distribution of freshwater mussels (Bivalvia: Unionidae) in south-eastern Michigan, U.S.A.
Freshwater Biology 49:127-142.
Metcalfe-Smith, J.L., J. Di Maio, S.K. Staton & G.L. Mackie. 2000. Effect of sampling effort on
the efficiency of the timed search method for sampling freshwater mussel communities.
The Journal of North American Benthological Society 19:725-732.
Milam, C.D., J.L. Farris, F.J. Dwyer, D.K. Hardesty. 2005. Acute toxicity of six freshwater
mussel species (glochidia) to six chemicals: Implications for daphnids and Utterbackia
imbecillis as surrogates for protection of freshwater mussels (Unionidae). Arch Environ
Contam Toxicol 48:166–173.
59
Morris, T.J. & L.D. Corkum.1996. Assemblage structure of freshwater mussels (Bivalvia:
Unionidae) in rivers with grassy and forested riparian zones. Journal of the North
American Benthological Society 15:576–586.
Neves, R.J. & M.C. Odom. 1989. Muskrat predation on endangered freshwater mussels in
Virginia. Journal of Wildlife Management 53:934-941.
Neves, R.J. 1999. Conservation and commerce: management of freshwater mussel (Bivalvia:
Unionoidea) resources in the United States. Malacologia 41:461-474.
Newton, T.J., D.A. Woolnough & D.L. Strayer. 2008. Using landscape ecology to understand
and manage freshwater mussel populations. Journal of the North American Benthological
Society 27:424-439.
Millington, P.J. & K.F. Walker. 1983. Australian freshwater mussel Velesunio ambiguus
(Philippi) as a biological monitor for zinc, iron and manganese. Marine and Freshwater
Research 34:873-892.
Novotny ,́ V. & Y. Basset. 2000. Rare species in communities of tropical insect herbivores:
pondering the mystery of singletons. Oikos 89:564-572.
Obermeyer, B.K. 1998. A comparison of quadrats versus timed snorkel searches for assessing
freshwater mussels. American Midland Naturalist 139:331-339.
Palmer, M.W., P.G. Earls, B.W. Hoagland, P.S. White & T. Wohlgemuth. 2002. Quantitative
tools for perfecting species lists. Environmetrics 13:121-137.
Poole, K.E.& J.A. Downing. 2004. Relationship of Declining Mussel Biodiversity to Stream-
Reach and Watershed Characteristics in an Agricultural Landscape. Journal of the North
American Benthological Society 23:114-125.
Price, J.E. & S.M. Welch. 2009. Semi-quantitative methods for crayfish sampling: Sex, size, and
habitat bias. Journal of Crustacean Biology 29:208-216.
Purvis, A., Hector, A. 2000. Getting the measure of biodiversity. Nature 405:212-219.
Pusch, M., J. Siefert & N. Walz. 2001. Filtration and respiration rates of two unionid species and
their impact on the water quality of a lowland river. Pages. 318–326. in F. Bauer and K.
Wachtler (editors). Ecology and evolution of the freshwater mussels Unionoida. Springer-
Verlag, Heidelberg, Germany.
R Development Core Team. 2010. R: A language and environment for statistical computing. R
Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org
Remsen Jr., J.V. 1994. Use and misuse of bird lists in community ecology and conservation. Auk
111:225-227.
Reynolds, L., A.T. Herlihy, P.R. Kaufmann, S.V. Gregory & R.H. Hughes. 2003. Electrofishing
effort requirements for assessing species richness and biotic integrity in western Oregon
streams. North American Journal of Fisheries Management 23:450-461.
60
Roecker, E.B. 1991. Prediction error and its estimation for subset—selected models.
Technometrics 33:459-468.
Scharff, N., J.A. Coddington, C.E. Griswold, G. Hormiga & P.D.P Bjørn. 2003. When to quit?
Estimating spider species richness in a Northern European deciduous forest. Journal of
Arachnology 31:246-273.
Schloesser, D.W., T.F. Nalepa & G.L. Mackie. 1996. Zebra mussel infestation of unionid
bivalves (Unionidae) in North America. American Zoologist 36:300-310.
Schloesser, D.W., J. L. Metcalfe-Smith, W. P. Kovalak, G. D. Langton, and R. D. Smithee. 2006.
Extirpation of freshwater mussels (Bivalvia: Unionidae) following the invasion of
dreissenid mussels in an interconnecting river of the Laurentian Great Lakes. American
Midland Naturalist 155:307-320.
Schwalb, A.N. & M.T. Pusch. 2007. Horizontal and vertical movements of unionid mussels in a
lowland river. Journal of the North American Benthological Society 26:261-272.
Shreve, R.L. 1967. Infinite topologically random channel networks. Journal of Geology 75:178-
186.
Smith, D.R., R.F. Villella, D.P. Lemarié & S. von Oettingen. 2001. How much excavation is
needed to monitor freshwater mussels? In R. A. Tankersley, D. I. Warmolts, G. T. Watters,
B. J. Armitage, P. D. Johnson & R. S. Butler (Eds.), Freshwater mollusk symposium
proceedings (pp. 203–218).Columbus: Ohio Biological Survey.
Smith, K.L. & M.L. Jones. 2005. Watershed-level sampling effort requirements for determining
riverine fish species composition. Can. J. Fish. Aquat. Sci. 62:1580–1588.
Smith, D.R. 2006. Survey design for detecting rare freshwater mussels. J. N. Am. Benthol. Soc.
25:701-711.
Strayer, D.L. 1983. The effects of surface geology and stream size on freshwater mussel
(Bivalvia, Unionidae) distribution in southeastern Michigan, U.S.A. Freshwater Biology
13:253-264.
Strayer, D.L. & D.R. Smith. 2003. A guide to sampling freshwater mussel populations.
American Fisheries Society Monograph 8:1-103.
Strayer, D.L. & H. Malcolm. 2007. Submersed vegetation as habitat for invertebrates in the
Hudson River estuary. Estuaries and Coasts 30:253-64.
Strayer, D.L. 2008. Freshwater Mussel Ecology, A Multifactor Approach to Distribution and
Abundance. University of California Press, Berkeley, California.
Strobl, C., A.L. Boulesteix, A. Zeileis & T. Hothorn. 2007. Bias in Random Forest Variable
Importance Measures: Illustrations, Sources and a Solution. BMC Bioinformatics 2007:8-
25.
61
Theler, J.L. 2000. Animal remains from Native American archaeological sites in western
Wisconsin. Transactions of the Wisconsin Academy of Sciences, Arts and Letters 88:121-
142.
Tiemann, J.S. 2006. Freshwater Mussel (Bivalvia: Unionidae) Survey of the Wakarusa River
Basin, Kansas. Transactions of the Kansas Academy of Science 109:221-230.
Tiemann, J.S., K.S. Cummings & C.A. Mayer. 2009. Timed search technique used to evaluate
freshwater mussel (Bivalvia: Unionidae) species richness in headwater streams: Is a single
one-hour visit enough? Journal of Freshwater Ecology 24:85-92.
U.S. Environmental Protection Agency. 2007. U.S. EPA Biological Indicators of Watershed
Health webpage. Accessed 12/12/2010. http://www.epa.gov/bioindicators.
Vannote, R.L., G.W. Minshall, K.W. Cummings, J.R. Sedell & C.E. Cushing. 1980. The River
Continuum Concept. Canadian Journal of Fisheries and Aquatic Sciences 37:130-137.
Vaughn, C.C. & C.M. Taylor. 2000. Macroecology of a host-parasite relationship. Ecography
23:11-20.
Vaughn, C.C, S.J. Nichols & D.E. Spooner. 2008. Community and foodweb ecology of
freshwater mussels. Journal of the North American Benthological Society 27:409-423.
Vaughn, C.C. 2010. Biodiversity Losses and Ecosystem Function in Freshwaters: Emerging
Conclusions and Research Directions. BioScience 60:25-35.
Villella, R.F. & D.R. Smith. 2005. Two-phase sampling to estimate river-wide populations of
freshwater mussels. Journal of the North American Benthological Society 24:357-368.
Walther, B.A. & J.L. Martin. 2001. Species richness estimation of bird communities: how to
control for sampling effort? Ibis 143:413-419.
Walker, K.F., M. Byrne, C.W. Hickey & D.S. Roper. 2001. Freshwater mussels (Hyriidae) of
Austalasia. In G. Bauer & K. Wächtler (eds), Ecology and Evolution of Freshwater
Mussels, Unionoida. Ecological Studies, Vol. 145. Springer Verlag, 5-31.
Wang, N., C.G. Ingersoll, D.K. Hardesty, C.D. Ivey, J.L. Kunz, T.W. May, F.J. Dwyer, A.D
Roberts, T. Augspurger, C.M. Kane, R.J. Neves & M.C. Barnhart. 2007. Acute toxicity of
copper, ammonia, and chlorine to glochidia and juveniles of freshwater mussels
(Unionidae). Environ Toxicol Chem 26:2036-2047.
Watters, G.T. 1994. An Annotated Bibliography of the Reproduction and Propagation of the
Unionoidea (Primarily of North America). Ohio Biological Survey Miscellaneous
Contributions.
Watters, G.T.1996. Small dams as barriers to freshwater mussels (Bivalvia, Unionoida) and their
hosts. Biological Conservation 75:79-85.
Watters, G.T. & S.H. O'Dee. 1998. Metamorphosis of Freshwater Mussel Glochidia (Bivalvia:
Unionidae) on Amphibians and Exotic Fishes. American Midland Naturalist 139:49-57.
62
Watters, G.T. 1999. Freshwater mussels and water quality: A review of the effects of hydrologic
and instream habitat alterations. Proceedings of the First Freshwater Mollusk Conservation
Society Symposium 1999:261-274.
Weigel, B.M., J. Lyons, L.K. Paine, S.I. Dodson & D.J. Undersander. 2000. Using stream
macroinvertebrates to compare riparian land use practices on cattle farms in Southwestern
Wisconsin. Journal of Freshwater Ecology 15:93-106.
Westman, K., O. Sumari & M. Puriainen. 1978. Electric fishing in sampling crayfish. Freshwater
crayfish 4:251–256.
Williams, J.D., M.L. Warren, K.S. Cummings, J.L. Harris & R.J. Neves. 1993. Conservation
status of freshwater mussels of the United States and Canada. Fisheries 18:6-22.
Wisconsin Department of Natural Resources, Fisheries and Aquatic Sciences Research Program.
2005. Guidelines for sampling freshwater mussels in Wadeable streams. Wisconsin
Department of Natural Resources. Madison, Wisconsin.
Woodcock, T., T. Mihuc, E. Romanowicz & E. Allen. 2006. Land-use effects on catchment- and
patch-scale habitat and macroinvertebrate responses in the adirondack uplands. American
Fisheries Society Symposium 48:395-411.
Zahner-Meike, E. & J.M. Hanson. 2001. Effect of muskrat predation on naiads. In: Ecology and
Evolution of the Freshwater Mussels Unionoida (G. Bauer and K. Wachtler, Eds).
Springer. New York.
Zigler, S.J., T.J. Newton, J.J. Steuer, M.R. Bartsch & J.S. Sauer. 2008. Importance of physical
and hydraulic characteristics to unionid mussels: a retrospective analysis in a reach of large
river. Hydrobiologia 598:343–360.
63
APPENDIX A: MUSSEL DEMOGRAPHIC FIELD DATA SHEET
(ILLINOIS NATURAL HISTORY SURVEY)
EPA Site No.: _________________________ Field No.:_________ Page: _________________
Stream:________________________________ Drainage:___________________________________
Common Location:_________________________________________________________________________
Lat, Long:________________________________ County:______________________ State:______
Date:______________________________________ T ______ R _____ sec. _____ Quarter ____
Collectors:_________________________________________________ Person-hours:___________
Remarks:_________________________________________________________________________________
_________________________________________________________________________________________
________________________________________________________________________
Mean (ft) Max (ft) North
Width: _______ ______
Depth: _______ ______
Length: _______ ______
Substrate (%) (Total = 100%):
Bedrock _______ Boulder _______
Gravel _______ Sand _______
Cobble _______ Silt _______
Clay _______ Other _______
Flow: Low Mod High
Clarity: Low Mod High
Sampling: Fair Good High
Live Corbicula: (none) 1 2 3 (abundant)
Live Dreissena: (none) 1 2 3 (abundant) Photo: none digital ________________
64
APPENDIX B: FRESHWATER MUSSEL SPECIES AND THEIR ABUNDANCE
AT EACH OF THE 18 WADEABLE-STREAM SITES
Species Site
ATF03 BE19 BEF02 BEP01 BERB01 BEZZ02 BO08 BO09 OZC01
A. confragosus 1 0 0 0 0 0 0 0 0
A. ferrusacianus 0 0 7 0 0 0 3 3 0
A. marginata 0 0 0 0 0 1 0 0 0
A. plicata 9 0 0 0 0 11 187 9 0
A. suborbiculata 1 0 0 0 0 0 0 0 0
A. viridis 0 0 0 0 0 0 1 0 0
C. tuberculata 0 0 0 0 0 0 0 0 0
F. flava 0 4 0 37 0 5 105 66 0
L. cardium 0 1 15 9 6 0 1 17 0
L. complanata 43 4 19 3 24 3 0 5 6
L. compressa 0 0 0 0 0 0 1 1 0
L. costata 0 0 0 0 0 0 0 0 0
L. fragilis 0 4 2 7 0 6 0 0 2
L. siliquoidea 0 1 102 18 2 1 604 6 0
L. subrostrata 0 0 0 0 0 0 0 0 1
L. teres 5 0 0 0 0 0 0 0 0
M. nervosa 3 0 0 0 0 0 0 0 0
O. reflexa 10 0 0 0 0 0 0 0 0
P. alatus 0 0 0 0 0 0 0 0 0
P. capax 1 0 0 0 0 0 0 0 0
P. grandis 0 2 0 0 0 3 1 0 8
P. ohensis 4 0 0 0 0 0 0 0 0
P. sintoxia 0 57 0 0 1 8 0 0 0
Q. Nodulata 109 0 0 0 0 0 0 0 0
Q. pustulosa 0 63 0 2 0 0 0 0 0
Q. quadrula 277 1 0 0 0 1 0 0 26
S. undulates 0 1 0 0 0 2 0 0 0
T. donaciformis 14 0 0 0 0 0 0 0 0
T. parvus 1 0 61 0 0 2 0 0 2
T. truncate 4 11 0 0 0 0 0 0 0
T. verrucosa 0 12 0 0 0 0 0 0 0
U. tetralasmus 0 0 0 0 0 0 0 0 0
V. ellipsiformis 0 0 0 0 0 0 0 0 0
V. lienosa 0 0 0 0 1 0 39 4 0
Total 459 161 206 76 34 43 942 111 45
65
Species Site
BPK12 BPKP05 DKK01 Dk20 EID04 EIDE01 EIE10 EIEI01 EZZH02
A. confragosus 0 0 0 0 0 0 0 0 0
A. ferrusacianus 1 174 0 0 0 15 0 31 3
A. marginata 0 0 0 0 1 1 0 0 0
A. plicata 0 178 1 0 0 124 1 3 0
A. suborbiculata 0 0 0 0 0 0 0 0 0
A. viridis 1 0 0 0 0 0 0 0 1
C. tuberculata 3 0 0 0 0 0 0 0 0
F. flava 24 267 2 8 1 94 3 2 11
L. cardium 19 2 22 16 37 72 13 25 9
L. complanata 3 20 6 1 1 9 1 8 0
L. compressa 0 0 0 0 0 0 0 1 3
L. costata 1 1 0 0 0 0 0 0 0
L. fragilis 0 0 1 0 5 0 0 0 0
L. siliquoidea 4 57 3 9 0 4 0 78 108
L. subrostrata 0 0 0 0 0 0 0 0 0
L. teres 0 0 21 10 0 1 0 1 0
M. nervosa 3 0 0 0 0 0 0 0 0
O. reflexa 10 0 0 0 0 0 0 0 0
P. alatus 0 0 5 6 0 0 0 0 0
P. capax 0 0 0 0 0 0 0 0 0
P. grandis 0 38 0 0 0 0 0 0 0
P. ohensis 0 0 0 0 0 0 0 0 0
P. sintoxia 5 1 1 8 2 62 1 0 0
Q. nodulata 0 0 0 0 0 0 0 0 0
Q. pustulosa 42 0 5 1 1 0 2 0 0
Q. quadrula 16 0 1 0 1 4 3 0 0
S. undulates 5 58 1 0 0 39 1 2 0
T. donaciformis 0 0 0 0 0 0 0 0 0
T. parvus 0 0 0 0 0 0 0 0 0
T. truncate 0 0 0 0 0 0 0 0 0
T. verrucosa 52 0 0 0 4 0 2 0 0
U. tetralasmus 0 2 0 0 0 0 0 1 0
V. ellipsiformis 0 0 0 2 0 0 0 0 0
V. lienosa 2 73 0 0 0 0 0 0 0
Total 178 873 54 54 53 425 27 152 135
IN ILLINOIS
66
APPENDIX C: NUMBER OF MUSSEL SPECIES ACCUMULATED OVER SAMPLING
EFFORT AT 18 WADEABLE-STREAM SITES
Sites Sampling effort (man-hours)
2 4 6 8 10 12 14 16
1 5 6 6 6 6 6 6 6
2 3 6 6 6 6 6 6 6
3 4 7 7 7 7 8 8 8
4 9 10 11 12 12 12 12 13
5 9 9 9 9 9 11 11 11
6 5 8 9 10 10 11 11 11
7 3 4 4 4 4 5 6 6
8 3 4 4 5 5 5 5 5
9 4 6 9 10 10 11 12 12
10 5 6 8 8 8 9 9 9
11 5 5 5 7 8 9 10 10
12 4 5 6 6 10 12 14 14
13*
3 3 3 6 8 9 9
/
14 1 2 3 4 6 7 7 9
15 7 9 9 9 9 9 9 10
16 2 3 5 5 6 6 6 6
17 9 11 14 15 16 17 17 18
18 7 8 12 13 13 13 13 13
Note: * denotes that sampling was stopped at 14 man-hours due to a storm at Site 13.
IN ILLINOIS
67
APPENDIX D: THE DESCRIPTIONS OF ENVIRONMENTAL VARIABLES OF 18
WADEABLE-STREAM SITES
Variables Descriptions
Site-specific
Channel width Mean channel width (m)
SD of width Standard deviation of channel width
Water depth Mean thalweg depth (m)
SD of depth Standard deviation of water depth
Water clarity Rank of water clarity (1-dark, 3-mucky, 5-stained, 7-moderate, 9-clear)
Flow rate Rank of flow velocity (1-still, 3-trickle, 5-smooth, 7-ripple, 9-turbulent)
Water temperature Water temperature at sampling time (oC)
% of silt-dominant points % of points dominated by silt (0.004-0.062 mm)
% of sand-dominant points % % of points dominated by sand (0.062-2.0 mm)
% of gravel-dominant points % of points dominated by gravel (2-64 mm)
% of cobble-dominant points % of points dominated by cobble (64-256 mm)
% of boulder-dominant points % of points dominated by boulder (>256 mm)
% of clay-dominant points % of points dominated by claypan
% of logs-dominant points % of points dominated by logs (>7.6 cm diameter)
% of vegetation % of non-woody vegetation in the stream
Substrate diversity Shannon diversity index of substrate
Reach-level
Dist_Class Disturbance classes of streams (1-low, 2-medium, 3-high)
Gradient Mean gradient of channel (degree)
IBI Index of biotic integrity
68
Variables Descriptions
Riparian zone-level (60 m from both stream banks)
R_lu21 % non-row crop in the local riparian zone
R_lu41 % upland forests in the local riparian zone
R_lu50 % open water in the local riparian zone
R_lu610 % shrub wetland in the local riparian zone
R_lu62 % non-wooded wetland in the local riparian zone
R_perm_mean Soil permeability of local riparian zone
R_qg12 % local riparian zone underlain by Alluvium
Rt_area_km2 Total riparian zone area (km
2)
Rt_bd203 % bedrock in the total riparian zone (from depth 100 to 200 m)
Rt_bd204 % bedrock in the total riparian zone (from depth 200 to 400 m)
Rt_lu11 % industrial area in the total riparian zone
Rt_lu12 % residential area in the total riparian zone
Rt_lu13 % non-used urban area in the total riparian zone
Rt_lu30 % open land in the total riparian zone
Rt_lu50 % open water in the total riparian zone
Rt_lu610 % shrub wetland in the total riparian zone
Rt_lu62 % non-wooded wetland in the total riparian zone
Rt_qg6 % total riparian zone underlain by fine ground-moraine
Watershed –level
W_lu12 % residential urban area in the local watershed
W_lu21 % non-row agriculture in the local watershed
W_lu30 % open land in the local watershed
W_lu41 % upland deciduous forest in the local riparian zone
W_lu50 % open water in the local watershed
W_lu62 % non-wooded in the local watershed
69
Variables Descriptions
W_qg12 % local watershed underlain by Alluvium
W_qg3 % local watershed underlain by fine end-moraine
W_qg6 % local watershed underlain by fine ground-moraine
Wt_bd204 % bedrock in the total watershed (from depth 200 to 400 m)
Wt_lu22 % row crop in the total watershed
Wt_lu50 % open water in the total watershed
Wt_lu610 % shrub wetland in the total watershed
Wt_lu62 % non-wooded Wetland in the total watershed
Wt_lu70 % barren land in the total watershed
Wt_qg6 % total watershed underlain by fine ground-moraine
Catchment area Total watershed area (km2)
Stream order (Shreve 1967) Number of the 1st-order streams in a catchment, i.e., sum of upstream links
Upstream_length Length of channels in upstream watershed (km)
Damdnst_length Distance from closest dam to target segment (km)
Watershed_located The name of watersheds where sites are located
Base flow index The ratio of annual base flow volume to annual total flow volume (%)
KFACT Erodibility (susceptibility of soil particles to detachment by water)
OMAVE Average organic matter content in the soil (g/cm3)
BDAVE Average value of soil bulk density (g/cm3)