:2
_

 

THS_

waste
:1.
21709

This is to certify that the
dissertation entitled

HABITAT SELECTION AND RESILIENCE OF STREAM
FISHES IN THE RED CEDAR RIVER WATERSHED,
MICHIGAN

presented by

10 ANNE LATIMORE

has been accepted towards fulﬁllment
of the requirements for the

 

 

Doctoral degree in \ Fisheries and Wildlife
\
- ”j W

Maj or Professor’s Signature
a & Cc, “4 be I 2&5 ’

Date

 

MSU is an Afﬁrmative ActiorVEqual Opportunity Institution

 

LIBRARY
Michigan State
University

 

 

 

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

 

DATE DUE DATE DUE DATE DUE

 

0m 05 2007
10 02 07
mm 40 a 2*2008
JUN, o 9 20:3

113035

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:lClRC/DaIeDue.indd-p.1

HABITAT SELECTION AND RESILIENCE OF STREAM FISHES IN THE RED
CEDAR RIVER WATERSHED, MICHIGAN

By

Jo Anne Latimore

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Fisheries and Wildlife

2005

ABSTRACT

HABITAT SELECTION AND RESILIENCE OF STREAM FISH IN THE RED
CEDAR RIVER WATERSHED, MICHIGAN

By
Jo Anne Latimore

Resilience to disturbance is vital to the sustainability of ecological systems.
However, determination of system resilience in the ﬁeld is less than straightforward
because of its dependence on the nature of the disturbance and on the choice of endpoints
for system recovery. More frequently, we base assessments of ecological system health
on measures of function, such as abundance, productivity, and biomass, and measures of
structure, such as species richness and diversity, which are relatively simple to obtain in
the ﬁeld.

I conducted 14 experimental ﬁsh defaunations in lOO-meter reaches of second- to
fourth-order warmwater streams in the Red Cedar River watershed of lower Michigan.
Stream habitat was left intact. I documented ﬁsh assemblage recovery in these reaches
over the course of a month in order to describe the structure of recovering assemblages
over time and to identify habitat or assemblage characteristics that inﬂuenced resilience.
Fish assemblages recovered remarkably quickly, exhibiting high similarity to pre-
disturbance structure within two days of defaunation. Pool habitat appeared to encourage
recolonization, while pre-disturbance ﬁsh abundance was negatively related to abundance
recovery. No relationship between resilience and multimetric indices commonly used to
assess stream health, such as overall habitat quality or biotic integrity, was evident. Such
metrics, although efﬁcient in terms of cost and time, may not produce a complete picture

of stream health.

The results of these experiments permitted an examination of the relationship
between assemblage structure, function, and resilience as components of ecological
health. There was no relationship between the pre-disturbance vigor or organization of
the ﬁsh community and its resilience to disturbance. While I found that measuring post-
disturbance changes in structure and abundance over time was useful in determining
recovery, these resilience values are applicable only to one type of disturbance at one
spatial and temporal scale. Although vigor and organization may serve as partial
indicators of community health, a general model of the relationship of these variables to
resilience would be invaluable for ecosystem conservation and management.

Finally, the structure of these defaunation experiments provided an opportunity to
investigate habitat selection in a common stream ﬁsh, the creek chub (Semotilus
atromaculatus Mitchill). Traditional methods of assessing habitat suitability rely on
correlating habitat characteristics with the number of individuals occupying the habitat
under natural conditions. Distribution is a behavioral phenomenon, however, and the
majority of these correlations do not consider movement or density-dependent
interactions that may inﬂuence the suitability of a particular habitat, and therefore, the
dynamics of creek chub population recovery following to disturbance. In this study, I
considered both the inﬂuence of physical habitat and density of conspeciﬁcs on
distribution. Creek chubs preferred pools over runs, and runs over rifﬂes. Alternative
methods of measuring selection, including selection indices and immigration rates,
revealed similar insights. Density of creek chubs in the study reaches, which ranged from
zero immediately following defaunation to above pre-defaunation (natural) densities, had

no apparent effect on habitat preferences.

ACKNOWLEDGMENTS

None of the research included here could have been accomplished without the
invaluable assistance of a talented and energetic group of undergraduate technicians:
Kelly DeGrandchamp, Janice Karcher, Kevin Mann, Colleen McLean, Cassandra Meier,
Aaron Shultz, and Alfred Thomson. I truly hope I was able to share something in return.
Fellow graduate students in the Department of Fisheries and Wildlife provided technical
assistance, social support, and intellectual stimulation, especially Bryan Burroughs, Matt
Klungle, Anita Morzillo, Sharon Johnson Schapel, Dave Thomas, and Brad Thompson.
My faculty Guidance Committee held high standards and provided direction throughout
the course of my program at Michigan State University: Mary Bremigan, Tom Coon,
Scott Witter, and most outstandingly, my Major Professor, Dan Hayes, about whom I
could never say enough.

Support for this research was provided through Michigan State University’s
Watershed Action Through Education and Research initiative (MSU-WATER), the
Michigan State University Distinguished Fellowship Program, and the US Department of
Agriculture National Needs Water Science Fellowship Program.

Finally, I express heartfelt appreciation for the support I have received as I
worked toward this goal - at once a mission accomplished and an adventure just
beginning - from my incomparable family: my parents, Terry and Nancy Latimore, my

sister, Janet Latimore, and my husband, Ralph Kridner.

iv

TABLE OF CONTENTS

List of Tables ........................................................................................ vii
List of Figures ....................................................................................... viii
CHAPTER 1. Fish assemblage resilience in experimentally defaunated warmwater
stream reaches ......................................................................................... 1
Abstract ....................................................................................... 1
Introduction ................................................................................... 2
Methods ....................................................................................... 5
Results ......................................................................................... 8
Discussion ................................................................................... 25
Acknowledgments .......................................................................... 35
Literature Cited ............................................................................. 36

CHAPTER 2. The vigor-organization—resilience concept of ecological health: lessons

from temperate warmwater stream ﬁsh commtmities .......................................... 41
Abstract ...................................................................................... 41
Introduction ................................................................................. 42
Methods ...................................................................................... 45
Results ....................................................................................... 45
Discussion ................................................................................... 49
Acknowledgments .......................................................................... 53
Literature Cited ............................................................................. 54

CHAPTER 3. Habitat selection by creek chubs (Semotilus atromaculatus Mitchill)
following experimental defaunation .............................................................. 55

Abstract ...................................................................................... 55

Introduction ................................................................................. 56

Methods ...................................................................................... 59
Results ....................................................................................... 63
Discussion ................................................................................... 76
Acknowledgments .......................................................................... 79
Literature Cited ............................................................................. 80

vi

LIST OF TABLES

Chapter 1.

Table 1. Stream habitat characteristics for the experimental sites. QHEI is the
Qualitative Habitat Evaluation Index (range 0-90; Rankin 1989). Mean width and depth
are reported in meters. Percent channel unit composition calculated as proportion of total
site length ...................................... . ...................................................... 9

Table 2. Total ﬁsh collected during experimental ﬁsh removals, excluding young-of-the-
year. Procedure 51 Score refers to the Michigan Department of Environmental Quality
scoring method for stream ﬁsh assemblages (range -10 to 10; MDEQ 1997) ............... 11

Table 3. Common and scientiﬁc names of all ﬁsh species collected in the course of ﬁsh
removal experiments in the Red Cedar River watershed, Michigan, in 2001 and 2002. . . 12

Chapter 2.

Table 1. Components of ecological health (adapted from Costanza and Mageau 1999) 43
Chapter 3.

Table 1. Experimental ﬁsh removal sites in the Red Cedar River watershed ............... 61
Table 2. Creek chub removal and recovery from experimental stream sites in the Red
Cedar River watershed .............................................................................. 64

vii

LIST OF FIGURES

Chapter 1.

Figure 1. Locations of study sites in the Red Cedar River watershed, Ingham and
Livingston Counties, Michigan. See Table 1 for study site descriptions ..................... 5

Figure 2. Recovery rates of ﬁsh species recolonizing defaunated stream reaches during
the ﬁrst sampling interval following ﬁsh removal (2-4 days), presented as the mean
percentage of pre-removal density recovered per day for the most common species
encountered .......................................................................................... 13

Figure 3. Species-speciﬁc patterns of density recovery over the course of the one-month
recolonization period following defaunation, presented as the total number of individuals
collected at all experimental sites for each sampling interval ................................. 15

Figure 4. Morisita’s (1959) similarity of ﬁsh assemblages sampled during the recovery
period compared to pre-disturbance assemblages. Data points represent the mean value
(3:1 SE) across all 14 experimental stream reaches ............................................. 21

Figure 5. Mean level of ﬁsh density recovery (:hl SE) over time across all 14
experimental stream reaches ....................................................................... 21

Figure 6. Pre-disturbance ﬁsh density(a) and species richness(b) in the 14 experimental
stream reaches relative to the Qualitative Habitat Evaluation Index (QHEI; Rankin
1989) .................................................................................................. 22

Figure 7. Maximum ﬁsh density(a) and maximum species richness(b) recovery in the 14
experimental stream reaches relative to Qualitative Habitat Evaluation Index (QHEI;
Rankin 1989) ......................................................................................... 23

Figure 8. Maximum ﬁsh density recovery compared to the percentage of reach length as
pool habitat in the 14 experimental stream reaches ............................................. 24

Figure 9. Level of maximum ﬁsh species richness recovery compared to the pre-
disturbance richness of the 14 experimental stream reaches .................................. 24

Figure 10. Level of maximum ﬁsh density recovery compared to the pre-disturbance
density of the 14 experimental stream reaches .................................................. 26

Figure 11. Procedure 51 ﬁsh assemblage scores (MDEQ 1997) compared to the level of

maximum ﬁsh density recovery(a) and species richness recovery (b) in the 14
experimental stream reaches ....................................................................... 27

viii

Chapter 2.

Figure 1. Neither ﬁsh assemblage vigor (abundance) nor organization (species richness)
was related to habitat quality as measured by the Qualitative Habitat Evaluation Index
(QHEI; Rankin 1989) ............................................................................... 46

Figure 2. Fish assemblage organization (richness and proportional abundance) was
highly resilient, showing strong similarity to pre-disturbance conditions throughout the
recovery period. Morisita’s (1959) similarity index ranges from 0 (completely dissimilar)
to just above 1 (completely similar) (a). Assemblage vigor (ﬁsh abundance) recovered
more slowly but increased over time (b) ......................................................... 47

Figure 3. No clear relationship was found between pre-disturbance organization and
organization recovery after one month (a). Pre-disturbance vigor was negatively related
to its recovery after one month (b) ................................................................ 48

Figure 4. Paths of assemblage recovery (dotted arrow) following disturbance (bold
arrow) of biota only, as in our experimental ﬁsh removals. (a; 0 represents the mean
value across all sites for each time interval), and both biota and habitat (b) ............... 50

Chapter 3.

Figure 1. Creek chub habitat selection versus density of chubs in the experimental sites
containing pools, rifﬂes, and runs as assessed with (a) Strauss’ (1979) linear selection
index and (b) Manly’s standardized selection ratio (Manly et a1. 2002) ..................... 66

Figure 2. Creek chub habitat selection versus density of chubs in the experimental sites
containing only rifﬂes and runs as assessed with (a) Strauss’ (1979) linear selection index
and (b) Manly’s standardized selection ratio (Manly et a1. 2002) ............................ 67

Figure 3. Creek chub habitat selection versus density of chubs in the experimental sites
containing only pools and runs as assessed with (a) Strauss’ (1979) linear selection index
and (b) Manly’s standardized selection ratio (Manly et al. 2002) ............................ 68

Figure 4. Strauss’s linear selection index for pools plotted against site density for each
observation. Numeric data markers represent individual sites ............................... 69

Figure 5. Strauss’s (1979) linear selection index for runs plotted against site density for
each observation. Numeric data markers represent individual sites ......................... 70

Figure 6. Strauss’s (1979) linear selection index for riﬁles plotted against site density for
each observation. Numeric data markers represent individual sites ......................... 71

Figure 7. Immigration rates of creek chubs colonizing each habitat unit class in streams
containing pools, rifﬂes, and runs. Immigration was calculated for (a) the ﬁrst return

ix

interval, 2 to 6 days after ﬁsh removal, and (b) an interval of one month, the length of the
experiments .......................................................................................... 73

Figure 8. Immigration rates of creek chubs colonizing each habitat unit class in streams
containing only rifﬂes and runs. Immigration was calculated for (a) the ﬁrst return
interval, 2 to 6 days aﬁer ﬁsh removal, and (b) an interval of one month, the length of the
experiments .......................................................................................... 74

Figure 9. Immigration rates of creek chubs colonizing each habitat unit class in streams
containing only pools and runs. Immigration was calculated for (a) the ﬁrst return
interval, 2 to 6 days after ﬁsh removal, and (b) an interval of one month, the length of the
experiments .......................................................................................... 75

CHAPTER 1

Fish assemblage resilience in experimentally defaunated warmwater stream reaches

Abstract

Resilience to disturbance is vital to the sustainability of ecological systems.
Stream ecosystems are noted for their resiliency, due to the broad range of natural
ﬂuctuations in habitat and water conditions to which the biota is adapted. However,
anthropogenic impacts can disturb stream systems by shifting habitat, ﬂow, or water
quality out of typical ranges or by introducing new stressors. We currently have limited
knowledge of stream ﬁsh assemblage resilience, especially the short-term dynamics of
recolonization following disturbance. I conducted 14 experimental ﬁsh defaunations in
100 m reaches of second- to fourth-order warmwater streams in the Red Cedar River
watershed of lower Michigan. Stream habitat was left intact. I documented recovery in
these reaches over the course of a month in order to describe the structure of recovering
assemblages over time and to identify habitat or assemblage characteristics that
inﬂuenced resilience. Assemblage structure was highly similar to the pre-disturbance
state within two days of defaunation. Recovery of ﬁsh abundance was slower, but
approached ﬁtll recovery within one month at most sites. Pool habitat within the
defaunated area appeared to encourage ﬁsh recolonization, while pre-disturbance ﬁsh
abundance was negatively related to abundance recovery. No relationship was found
between species richness and resilience. These results suggest that rapid colonization of
undisturbed stream habitat may be a general property of warmwater ﬁsh assemblages in

low-order streams. Interestingly, no relationship between resilience and multimetric

indices commonly used to assess stream health, such as overall habitat quality or biotic
integrity, was evident. In fact, overall habitat quality as measured by the Qualitative
Habitat Evaluation Index, designed to assess ﬁsh habitat, did not relate to pre-defaunation
ﬁsh assemblage. Such metrics, although efﬁcient in terms of cost and time, may not

produce a complete picture of stream health.

Introduction

Resilience, or the ability to recover from disturbance (Holling 1973, Pimm 1984)
is an important component of ecosystem sustainability. Contanza and Mageau (1999)
emphasized this when they deﬁned sustainability as the ability to maintain structure and
function over time in the presence of disturbance. Stream ecosystems are noted for their
resiliency (Resh et al. 1988), which is often attributed to biological adaptation to the
broad ﬂuctuations in ﬂow (Fisher et al. 1982, Poff et al. 1997), temperature (Hawkins et
a1. 1997), and related variables that streams naturally exhibit (Wallace 1990, Yount and
Niemi 1990). Human impacts within streams and their watersheds, however, oﬁen shift
stream conditions outside the natural range of variation (Likens et al. 1970, Ward and
Stanford 1983, Bain et al. 1988, Niemi et al. 1990, Kinsolving and Bain 1993, Li et al.
1994), or introduce new stresses to the system, such as exotic species (Pringle 1997) or
pollutants (Krumholz and Minckley 1964). The severity of these disturbances may test
the resiliency, and therefore the sustainability, of stream systems.

A substantial body of research exists on the impacts of disturbance on stream
ﬁshes (see Detenbeck et al. 1992 for a review), but it consists primarily of case studies

with multiple ecological impacts and minimal pre-disturbance information. As such, we

currently have limited knowledge of the patterns or mechanisms of stream ﬁsh resilience,
or how resilience is affected by speciﬁc stream or assemblage characteristics.
Experimental studies of resilience are replicable and also have the advantage of being
able to manipulate single variables, thereby revealing the dynamics of recovery in a way
not possible in the majority of case studies. A few experimental studies of stream ﬁsh
response to disturbance have been conducted in which habitat is largely unchanged, but
these studies have addressed resiliency over relatively long time intervals (e.g., re-
sarnpling aﬁer one year). Most of these studies have shown substantial, if not always
complete, recovery (Larimore et al. 1959; Gunning and Berra 1968,1969; Berra and
Gunning 1970; Olmsted and Cloutman 1974; Meﬁ‘e and Sheldon 1990; Ensign et al.
1997). Explorations of short-term dynamics are rare. Peterson and Bayley (1993)
observed ﬁsh assemblages recovering from experimental defaunation in Midwestern
warmwater streams over intervals of thirty minutes to six days in reaches 46-113 m in
length and predicted near-complete recovery would occur within two weeks. In a
similarly designed study, Sheldon and Meffe (1995) defaunated individual warmwater
stream pools and observed recovery within one to two months. The former study also
was designed to detect ecological factors that may inﬂuence recovery, but found that time
since defaunation was the only predictor of the degree of colonization at a particular site.
Stream and ﬁsheries management frequently depends on the use of multimetric
indices that combine a number of variables that have been shown to relate to overall
stream quality. These indices are typically straightforward, replicable, and cost little in
terms of effort or resources, allowing many sites to be assessed quickly. For example, the

quality of ﬁsh habitat in Midwestern streams is often measured using the Qualitative

Habitat Evaluation Index (QHEI; Rankin 1989). Stream ﬁsh assemblages are frequently
assessed using an index of biotic integrity (IBI; Karr 1987). Resilience to disturbance is
clearly a factor in stream integrity and sustainability, yet intentionally disturbing streams
to determine resilience could be counterproductive and costly. Determining whether
current methods for assessing streams reﬂect their ability to recover from disturbance (or
learning how to modify them to do so) would be extremely valuable from a management
standpoint.

I conducted replicable experimental defaunations in undisturbed habitat to address
two major objectives. The ﬁrst was to describe the rate at which ﬁsh assemblages
rebound following this type of disturbance, and the structure of the assemblage over time
as it recovered. An understanding of the rate and pattern of the recovery process provides
a model for setting expectations for other recovering stream ﬁsh assemblages. I expand
on the design of Peterson and Bayley (1983) by repeatedly observing the colonization
process in 100 m reaches over a month’s time. My second objective was to identify
features of ﬁsh assemblages and stream habitat that inﬂuenced recovery. I considered not
only individual stream and assemblage descriptors, but also multimetric indices
commonly used to assess stream quality. Achieving this objective would allow us to
identify less resilient systems where impacts should be avoided, and to determine
whether current methods of stream assessment reﬂect the system’s ability to recover from

disturbance.

Methods

Fourteen ﬁsh removal experiments were conducted in twelve second- to third-
order streams during the summers of 2001 and 2002. To maintain consistency in ﬁsh
fauna, geology, and climatic affects, all work took place in the Red Cedar River
watershed, Ingham and Livingston Counties, Michigan, USA (Figure 1). This
warmwater system is characterized by low gradients (mean = 0.475 m/km; Horton 1969),

glacial till geology, and a high percentage of agricultural land cover. Experimental sites

 

8
11 10 14

12

 

 

8km

 

Figure 1. Locations of study sites in the Red Cedar River watershed, Ingham and
Livingston Counties, Michigan. See Table 1 for study site descriptions.

were selected, using land use data (MDNR 2001) and ﬁeld scouting, to include a range of
stream sizes and habitat characteristics.

I characterized the habitat in each study reach prior to ﬁsh assemblage
manipulation. First, approximately 100 meters of stream were delineated into channel
units (pools, rifﬂes, and runs; Dunne and Leopold 1978). The length of each unit was
recorded, and a transect was established at the midpoint of the unit. At each transect, the
wetted width, depth proﬁle, and substrate composition using the pebble count method
(Wolman 1954) were documented. Habitat quality of the study reach was assessed using
the Qualitative Habitat Evaluation Index (QHEI), a subjective, multimetric stream ﬁsh
habitat assessment index that includes substrate, in-stream cover, channel morphology,
riparian zone quality and bank erosion, and pool and rifﬂe quality (Rankin 1989).

Block nets (3 -mm mesh) were temporarily installed at the upstream and
downstream ends of the study reach during ﬁsh removal to prevent ﬁsh immigration or
emigration. Fish were collected using a Smith-Root backpack-mounted or barge-
mounted electroﬁshing unit, depending on stream depth. Fish were removed using a
minimum of three electroﬁshing passes through the reach. If more than ten individuals
were collected on the third pass, additional passes were conducted. The completeness of
my removal method was conﬁrmed at one site (site 5) by leaving block nets in place
overnight following removal. The study reach was electroﬁshed again the following day.
Very few ﬁsh were found, all of which were small and the majority of which were
young-of-the-year, which were excluded from all analyses in this study.

All ﬁsh collected from the study reach were identiﬁed to species, with the

exception of young-of-the-year ﬁsh, which were excluded from analysis due to the

difﬁculty of collection and identiﬁcation. During removal, ﬁsh were held in large shaded
holding containers ﬁlled with stream water and supplied with aeration. Following
removal, all collected ﬁsh were released at stream sites within the watershed 2-10 km
from the study reach to reduce the chance of their returning to the study site within the
experimental period. Pre-defaunation ﬁsh assemblages were assessed using the
Michigan Department of Environmental Quality’s Procedure 51 scoring method (MDEQ
1997), which closely resembles the index of biotic integrity (IBI; Karr 1987) used in

many Midwestern states.

Return trips were made to each study reach to determine the progress of
recolonization of the defaunated reach. Pilot studies indicated that study reaches could
recover within one week. Therefore, the ﬁrst return trip took place within two days of
ﬁsh removal. Return trip sampling consisted of a single electroﬁshing pass through the
study reach. After identiﬁcation, ﬁsh were returned to the channel unit from which they
were collected to allow colonization to continue unimpeded. Subsequent return trips
were made to each study reach at intervals of one week or less for at least a month.

Species richness (number of species present) and total number of ﬁsh collected
were calculated for removal and recovery samples. Fish collections from the ﬁrst
electroﬁshing pass during the removal were used for comparison with return trip
collections to standardize for collection effort. For analysis, any recovery sample that
met or exceeded removal abundance or richness was considered 100% recovered for that
parameter. By not assigning recovery values greater than 100%, I maintained a

conservative estimate of across—watershed recovery rates.

Resilience was calculated in terms of ﬁsh abundance by comparing ﬁrst-pass
removal counts to return trip counts. I used Morisita’s index of similarity (1M; Morisita
1959) to quantify assemblage resilience by comparing ﬁsh assemblage structure during
recovery to the pre-disturbance state:

00
2: nl in2i

_ i=1

' 0.1+).2)N1N2

Where n, = the number of individuals of species i in a sample; and

N = the total number of individuals in a sample; and

A_ani(n1i-l)
— N1(N1-l)

, based on Simpson’s measure of diversity (Simpson 1949).
This index represents assemblage similarity based on relative abundance of
species in the samples, so that proportionally similar assemblages of species score high
regardless of the total number of individuals in the samples. In addition, its performance
is less inﬂuenced by sample size than other similarity indices (Wolda 1981). Morisita’s
index ranges from zero (completely dissimilar) to slightly greater than one (complete
similarity). Values of 0.70 or greater are generally taken to indicate a high degree of
similarity (Matthews et al. 1988). Patterns of resilience were then related to habitat and

assemblage descriptors, including the QHEI and Procedure 51 , using simple linear

regression.

Results
The 14 study sites, located in twelve separate tributaries of the Red Cedar River

(Table 1), ranged in mean wetted width from 2.6 to 11.1 m and in mean depth from 0.11

 

 

3 c 2 42 E. Q .32 3.8 Ba .ﬁaem— cam 3
a a a mg a.” 3. 6.02833 2
.9 o 3 m2 3, 3 68m «3:538 v.86 58 N.
3 a 2 NS 3 on case .89 :
2. NM 3 and .3 mm 88x 03on x85 58 2
E a a a; a... 8 68m :35 at been 6.5.53— a
2: o o N2. 3“ 2 6.86 see a
8. o o _ S S. a 38m «3.538 6.85 ESE c.
8 2 o as 2.: % v.36 seesaw e
2: o c an... S. 2 :52 380 Ba .555 “a? m
2: o c a; 9m x v.86 can 58 a
8 an o a; S. : v.35 :85 m
a 2 S n2. 3 no 535 83. 528 N
2 a. mm mm... 2 8 base: 83. 038 a
5.: .x. 0:5. .x. .8.— .x £95 :32 52>» 53: 5:5 2.32 an an

 

Swan:
8% ~89 «o 5980:. a.» 33—313 :oEmanoo :5. $585 Boob; £208 5 corona v.8 Snow 23 523 522 .Aawﬂ gm
momé omgv x02: scams—gm «85mm 6323:sz 05 mm EEO .85 38685on 2: 8m moumtoﬁﬁgo “SEE 885m A 033.

to 0.38 m. Substrate was predominantly sand (mean 40%, range 17-59%, of substrate
material), followed by organic material and silt. The study sites varied from
channelized, continuous runs to stable rifﬂe-run sequences. Qualitative Habitat
Evaluation Index (QHEI; Rankin 1989) values for the experimental sites ranged from 18
(poor) to 65 (excellent); the average was 46 (marginally acceptable).

Twenty-ﬁve species of ﬁsh were collected during the ﬁsh removals (Table 2 and
3). Dominant species were creek chub (Semotilus atromaculatus), blacknose dace
(Rhinichthys atratulus), johnny darter (Etheostoma nigrum), brook stickleback (Culaea
inconstans), bluntnose minnow (Pimephales notatus) and white sucker (Catostomus
commersoni). Species richness in the study reaches during removal ranged from 5 to 16
(mean = 9.6, SE = 0.8). Total ﬁsh abundance within the 100-meter reaches ranged from
35 to 489 (mean = 246.0, SE = 40.0). Procedure 51 (MDEQ 1997) ﬁsh assemblage
scores for the study reaches ranged from -8 (poor) to +3 (tending toward excellent; mean
= -4.1, SE = 0.73).

I calculated species-speciﬁc recovery rates for the most abundant ﬁsh species.
First, I considered the recovery rate during the ﬁrst interval following removal (two to
four days), by calculating the percentage of removal density recovered per day during
that ﬁrst interval. This ﬁrst-interval recovery rate represents the maximum rate observed
for each species, and ranged from 44% recovery/day (SE = 22%) for johnny darters to
9% recovery/day (SE = 1%) for rock bass (Figure 2). Second, I generated a pattern of
density recovery over the duration of the one-month experimental period for each of
these species by summing the number of individuals collected at each sampling interval

for all experiments. While some species, such as the central mudminnow and the mottled

10

 

2: E 2: 2: 2: 2: 2: 8 an 2: mm NM 2: ".0339: ouaatnsa< ..\o .52
2: ne 2: an 8— 2: 2: 2: an an 2: «a 2. 2: . "b.9593— mnoE—oa o\.. in:
N- n- w- v- v- m- h- v- v- w- 330m E 9:539:

«mm :N has a: 92 mm— ME 53. mm aav new 2: £930 .85.

b
\O

In
("t
*0 ‘0
n I
N
ea
0
M
N
I

 

 

 

2 N 2 2 N N N N 2 z m 2 S N "325:. £33

2. N o N c N o o c _ c o o N o 52.. 33532
2. o o N N N c N c N _ N c o o 88 :8“.
N... _ o o o o o o o o o N o o o 325 No: .5582
2. o o N N o o o o _ c c o o o 55% 2805
2. _ o o o N c o o o o N o o N 83:3 gem
N... N o o o o o o c o N N o o N 32.3 NONE
N... o o o o o o o o c N o o o o .25 38%
2 o c o o o o o N N o o o o o 38 58535.0.
.3 o o o o N N N c a N c o o o 3852 ”8%:
N... o N o _ o _ N N N c _ _ N _ swim
a... o o N N _ N N. o N N o N N o 5:283 32.8
N... 3 o N c o o N o 2 N c c N N mam seem
a... 2. o N N o c o N N N o o o o 3:3 520
N... z N c o o N 2 N o o o N N N 385532 .358
an NN c N. _ N. N N o _N N N N _ N 3:5 929.85
3. _ 2 N N _ o N N N NN N o o o beam 85:50
:2 N o o _ o o o o 22 NN o z o N 5:5 32.53.
92 o o 8 ON NN o o N N o 2 2 N o seam 8322
N2 NN N «N NN N N N o N N o N N N .325 223
N2 NN N a N o N N 2 NN mm o NN N N 3252 3855
NNN N a o N N o N NN N .N c N: X N 83225 Noam
3N NN NN NN N 2 N NN 2. 3 a _ N. 2 2 3:5 .852
3:. 2 NN : NN 2 2 NN N NN N N N2. N2 No :26 N85
v.2. NN NN NN : N N : NN N Na NN NNN o2 N 88 085.85
8% z 2 2 : 3 N N N N N N. N N N 8.8%
mmOho<

:82 ea

 

Ath ONE x: 8 3- omaﬁv 33383.3 ﬁe BEEN com @0508 3203 £350 Euogohém mo Hangman 595:2
05 3 2&2 28m 3 0.530on .Eo>-oﬁ.mo-m§o> mamas—88 £3382 can EcuEtoaxo wage @8828 32m 38. .N 2an

11

Table 3. Common and scientiﬁc names of all ﬁsh species collected in the course of ﬁsh
removal experiments in the Red Cedar River watershed, Michigan, in 2001 and 2002.

 

Family

Scientiﬁc Name

Common Name

 

Catostomidae

Centrarchidae

Cottidae
Cyprinidae

Gasterosteidae
Ictaluridae

Percidae

Salmonidae
Umbridae

Catostomus commersom'
Hypentelium nigricans
A mbloplites rupestris
Lepomis cyanellus
Lepomis macrochirus
Micropterus dolomieu
Cottus bairdi
Campostoma anomalum
Luxilus cornutus
Margarzls'cus margarita
Notropis atherinoides
Notropis hudsonius
Opsopoeodus emiliae
Pimephales notatus
Rhinichthys atratulus
Semotilus atromaculatus
Culaea inconstans
Ameiurus melas
Ameiurus nebulosus
Etheostoma caeruleum
Etheostoma nigrum
Percina maculata
Oncorhynchus mykiss
Umbra limi

12

White Sucker
Northern Hog Sucker
Rock Bass

Green Sunﬁsh
Bluegill
Smallmouth Bass
Mottled Sculpin
Central Stoneroller
Common Shiner
Pearl Dace
Emerald Shiner
Spottail Shiner
Pugnose Minnow
Bluntnose Minnow
Blacknose Dace
Creek Chub

Brook Stickleback
Black Bullhead
Brown Bullhead
Rainbow Darter
Johnny Darter
Blackside Darter
Rainbow Trout
Central Mudminnow

 

{g 60%j
.t' 50% -
E 40%~
E
g 30% q jig-,1;
3 20% a w
a! 10% j Vt: :i r 1
ml ' “ A
‘ '0 ° ‘ ‘ ~$ -0 o e
do °° as? 0* 9°53 \é‘sﬁ 9°69 0&6 9‘9 03¢ 006 00%
(‘6x 9 00 6’0 0 \a" .‘9 , 6° 6‘ 06% 00$ °d~
6‘ <9 ° 6 a) \‘9 {a -o
5 N“ 96‘ Co ~t- 43° 6‘ \ o \
\° %\ 9° 6" 6‘0 V“
9 Q C?

Figure 2. Recovery rates of ﬁsh species recolonizing defaunated stream reaches during
the ﬁrst sampling interval following ﬁsh removal (2-4 days), presented as the mean
percentage of pre-removal density recovered per day for the most common species
encountered.

13

sculpin, increased steadily and approached or reached recovery over a month’s time,
several other species did not (Figure 3). It is important to note that the apparent failure of
creek chubs, white suckers, and blacknose dace to approach recovery is strongly
inﬂuenced by the data from a single site (Site 2, Corwin Road Tributary). Likewise, the
erratic patterns demonstrated by green sunﬁsh, bluntnose minnows, and rock bass are
partly the result of data from Site 14 (East Branch, Red Cedar River), where these species
were abundant but which was not sampled on days 11 or 25.

Stream ﬁsh assemblage structure recovered rapidly at all sites (Figure 4). Within
two days of disturbance, the mean level of Morisita’s (1959) similarity (IM) between
removal assemblages and recovering assemblages across all streams was 0.83 (SE =
0.048), and the similarity of assemblages to pre-disturbance conditions remained high
throughout the study. Species richness, a component of assemblage structure, followed a
similar pattern. The number of ﬁsh in the experimental reaches took more time to
recover (Figure 5). After two days, the mean level of numerical recovery across all
streams was 38% (SE = 8%). Densities continued to increase throughout the one-month
experimental period, averaging 69% (SE = 9%) recovery aﬁer 32 days.

I compared the QHEI scores of the study reaches to removal ﬁsh assemblages and
the recovery of abundance and species richness to look at habitat inﬂuences on ﬁsh
assemblage structure or recovery. In no case was there a signiﬁcant relationship (Figures
6 and 7). However, in streams without distinct pool habitat, only two out of seven (29%)
reached complete abundance recovery, while 6 out of 7 (86%) with pools did (Figure 8).

Pre-removal assemblage characteristics had variable effects on recovery. There

14

Blacknose Dace

Number

 

Removal 2 4 1 1 18 25 32
Days Slnce Removal

Blackside Darter

Number

 

Removal 2 4 1 1 18 25 32
Days Slnce Removal

Figure 3. Species-speciﬁc patterns of density recovery over the course of the one-month
recolonization period following defaunation, presented as the total number of individuals
collected at all experimental sites for each sampling interval.

Bluntnose Minnow

 

B
.e
E
:3
z .
E
i
147%
Removal 2 4 11 18 25 32
Days After Removal
Brook Stickleback
200
150
h
3
E 100
5
z

50

    

1 8 25 32

Days Slnee Removal

Removal 2 4 1 1

Figure 3 (cont’d).

Central Mudminnow

Number

 

Remowl 2 4 1 1 18 25 32
Days After Removal
Creek Chub
450
400
350
300
i 250
S 200
z 150
1 00
50
0
Removal 2 4 1 1 18 25 32
Days Slnce Removal
Figure 3 (cont’d).

301

 

151

Number

 

 

 

 

 

 

Green Sunﬁsh

 

 

 

 

 

 

 

 

 

 

 

 

 

Removal

 

 

 

 

l

2

 

 

 

4 11 18
Days After Removal

Johnny Darter

 

 

 

 

 

 

 

 

 

 

32

 

 

 

 

 

Removal

Figure 3 (cont’d).

2

4 11 18
Days Since Removal

18

 

Mottled Sculpin

Num ber

 

Removal
Days Since Removal

Rainbow Darter

Number

 

25 32

Removal 2 4 11 18
Days Since Removal

Figure 3 (cont’d).

Rock Bass

 

 

B
E
3
2
Removal 2 4 11 18 25 32
Days After Removal
White Sucker
I-
3
E
3
z

 

 

2 4 11 18 25 32

Days Since Removal

Figure 3 (cont’d).

20

Morisita's Index

 

 

 

T l r l l l 7

-5 10 15 20 25 30 35

Days Post-Disturbance

Figure 4. Morisita’s (1959) similarity of ﬁsh assemblages sampled during the recovery
period compared to pre-disturbance assemblages. Data points represent the mean value
(11 SE) across all 14 experimental stream reaches.

90%
80%

E 70%
60%
50%
40%
30%
20%
10%
0%

Density Recove

L4

 

 

l l T T l l l

5 10 15 20 25 30 35
Days Post-Distu rbance

Figure 5. Mean level of ﬁsh density recovery (:l:1 SE) over time across all 14
experimental stream reaches.

21

(a)

400 -

 

 

 

 

3‘ R2 = 0.023 .
8
a 200 a ’ .
> 9 0 e
0 e
g 100 1 0 z
n: . 9
0 V 1 i ' l
0 20 40 60 80
QHEI Habitat Score
(b)
20 1 R2 = 0.027
g J P = 0.577 ’
g 15 o
. o o
E 10 . .: 9
9 ° ° 2
° 5 - ° 0
5
m 0 i l i l
0 20 40 60 80

QHEI Habitat Score

Figure 6. Pre-disturbance ﬁsh density(a) and species richness(b) in the 14 experimental
stream reaches relative to the Qualitative Habitat Evaluation Index (QHEI; Rankin 1989).

22

(a)

0))

Density Recovery

Richness Recovery

1 00%
80%
60%
40%
20%

0%

1 00%
80%
60%
40%
20%

0%

l

 

 

 

 

~ 0
O
7 O O
1 ’ R2 = 0.023
P = 0.605
0 20 40 60 80
QHEI Habitat Score
a O O O O. O
9 e
— O
’ e
_ O
l R2 = 0.002
— P = 0.872
0 20 40 60 80

QHEI Habitat Score

Figure 7. Maximum ﬁsh density(a) and maximum species richness(b) recovery in the 14
experimental stream reaches relative to Qualitative Habitat Evaluation Index (QHEI;

Rankin 1989).

23

100% e e 0 es 0

 

E 80%
>
O
3 60%
M
g 40%
g 0
3 20%
0% 1 l 1 1 F 1
0 10 20 30 40 50

% Pool

Figure 8. Maximum ﬁsh density recovery compared to the percentage of reach length as
pool habitat 1n the 14 experimental stream reaches.

 

 

100% - e e e e e e e
E" o 9
g 80%~ e
3 . °
33 60%- °
§ 40%—
E
g 20%~
n:
0% l T l 1
0 5 1O 15 20

Removal Richness

Figure 9. Level of maximum ﬁsh species richness recovery compared to the pre-
disturbance richness of the 14 experimental stream reaches.

24

was no relationship between pre-removal species richness and richness recovery (Figure
9). Pre-removal ﬁsh abundance did appear to inﬂuence abundance recovery, however
(Figure 10). Reaches with relatively low numbers of ﬁsh (3100) were able to recover
completely within the experimental period, while reaches with relatively high ﬁsh
numbers (>175) generally did not recover numbers fully. Finally, Procedure 51 ﬁsh

assemblage scores did not appear to relate to ﬁsh assemblage recovery (Figure 11).

Discussion

The sandy, low-gradient, and oﬁen channelized streams of the Red Cedar River
watershed are typical of agricultural watersheds in Michigan and elsewhere in the
Midwestern United States (Richards et al. 1996, Iowa State University 1997). Qualitative
Habitat Evaluation Index (QHEI; Rankin 1989) values for the experimental stream
reaches ranged from 18 to 65. Values below 45 (7 sites) indicate poor ﬁsh habitat, scores
of 45—60 (6 sites) are average, and scores of 60-90 (3 sites) are considered excellent. This
index suggests that the study reaches provided minimally acceptable ﬁsh habitat on
average.

The ﬁsh assemblages observed in the experimental sites were dominated by
species common in warmwater streams of the eastern United States (Trautman 1981).
Five of the six numerically dominant species are tolerant of degraded stream quality
(Rhinichthys atratlulus, blacknose dace; Semotilus atromaculatus, creek chub;
Etheostoma nigrum, johnny darter; Pimephales notatus, bluntnose minnow; and
Catostomus commersoni, white sucker) and four are omnivorous (creek chub, blacknose

dace, bluntnose minnow, and white sucker; MDEQ 1997), indicating a generalist,

25

   
  

100% w 9 R2 = 0.7033
5' 80% _ P = 0.0002
0
60% -
E
3. 40% —
'2 20°/ '
o _
8
0% l 1 l l

 

 

0 100 200 300 400
Removal Density

Figure 10. Level of maximum ﬁsh density recovery compared to the pre-disturbance
density of the 14 experimental stream reaches.

26

(a)

100% -
80% ~
60% -
40% -

Density Recovery

20% -

 

0%

.00.. O

 

(b)

100% -
80% -
60% ~
40% -
20% -

Richness Recovery

 

0%

Procedure 51 Fish Score

00..

0000

 

-10

l l l

-5 0 5
Procedure 51 Fish Score

Figure 11. Procedure 51 ﬁsh assemblage scores (MDEQ 1997) compared to the level of
maximum ﬁsh density recovery(a) and species richness recovery (b) in the 14
experimental stream reaches.

27

adaptable assemblage. The Michigan Department of Environmental Quality’s (1997)
Procedure 51 ﬁsh assemblage assessment can range from -10 to‘ 10 for a particular site.
A score of 5 or higher is “excellent”, -5 or lower is “poor”, and anything in between is
“acceptable”. Of the 14 experiment sites, six ﬁsh assemblages rated poor, eight were
acceptable, and none was excellent. The generalist nature of the assemblages may reﬂect
the quality of habitat available in the watershed and also suggests a degree of adaptation
to disturbance. For example, the assemblages are dominated by cyprinids, a mobile
group that appears to be able to rapidly colonize available habitats following disturbance
(Storck and Momot 1981, Smithson and Johnston 1999, Larson et al. 2002). Indeed,
creek chubs, blacknose dace, and bluntnose minnows were among the fastest-recovering
species in the ﬁrst few days following removal in these experiments.

Stream ﬁsh assemblages recovered rapidly from defaunation at the scale of this
experimental disturbance, with recovering assemblages exhibiting highly similar
structure to removal assemblages (mean IM=O.83) within two days, regardless of any
variation in recovery rates among species. Assemblage similarity greater than 0.70 is
typically considered high (Matthews et al. 1988, Walser et al. 2000). The restricted
movement paradigm (Gerking 1953) suggests that adult stream ﬁsh tend not to move
from a home pool or restricted reach. However, recent studies have demonstrated that
while many stream ﬁsh may remain in a home channel unit most of the time, regular
exploratory trips are commonplace (Smithson and Johnston 1999, Larson et al. 2002).
The speed of recovery in this study also suggests that a portion of the individuals of each

species is mobile, rather than restricted in their movements.

28

I suspect that the lag in abundance recovery behind structural similarity is at least
partly due to decreasing ﬁsh abundance throughout the vicinity of the experimental
reaches as individuals dispersed into the empty habitat. Again, this trend suggests that a
portion of each species is making exploratory trips outside of the home channel unit at
any given time. By not assigning recovery values greater than 100% when individual
samples did, in fact, occasionally exceed removal abundance, I maintained a conservative
estimate of across-watershed recovery rates. Some portion of the detected lag in
abundance recovery may be an artifact of this approach.

Prior studies of stream ﬁsh assemblage resilience have addressed long-term
potential for recovery, typically returning to the disturbed site after a year (Gunning and
Berra 1968,1969; Berra and Gunning 1970; Meffe and Sheldon 1990). Given my results,
it is not surprising that these earlier researchers observed full or nearly full recovery.
Peterson and Bayley (1993) suggested that assemblage recovery may occur much more
quickly. They investigated the recovery of ﬁsh assemblages from experimental
defaunation in Midwestern warmwater streams in reaches 46-113 m long at a time scale
of 05-140 hours. The assemblages in these streams, which had several species in
common with my sites, reached a proportional similarity index (Schoener 1968) of 0.70
in 60-140 hours (2.5-5.8 (1) following defaunation and recovered a projected 90% of ﬁsh
abundance within 100-270 h (4-11 d). Similarly, Sheldon and Meffe (1995) observed the
recovery of ﬁsh assemblages in individual pools within 1 to 2 months. These results
agree with the rapid colonization of available undisturbed stream habitat that I observed,
suggesting that this may be a general property of warmwater ﬁsh assemblages in low-

order streams.

29

Although ﬁsh assemblages generally recovered quickly, there was variation in the
rate and completeness of recovery among streams, allowing me to investigate potential
effects of habitat quality on ﬁsh assemblages and resilience. The QHEI was not related
to pre-defaunation assemblage structure (species richness, ﬁsh abundance, or Procedure
51 score) or resilience in my streams. It is curious that the ﬁsh assemblage did not
correlate to the QHEI because this index was designed to evaluate ﬁsh habitat in streams
and indicate the quality of assemblage that the stream could support (Rankin 1989). The
QHEI was created for Ohio warmwater streams similar to those in the Red Cedar River
watershed of lower Michigan. Stauffer and Goldstein (1997) similarly found no
relationship between QHEI and ﬁsh assemblage metrics (including IBI score, species
diversity, species richness, evenness, and percent of total individuals within trophic
groups) in a study of 18 prairie streams. While they noted that the QHEI was developed
for streams in forested ecoregions that conform to the river continuum concept (V annote
1980), rather than prairie streams that have somewhat different trophic dynamics, the
authors also suggested that the metrics within the QHEI that are most important to ﬁsh
assemblages may be masked in the total score due to redundancy and correlation.
Sharing this concern, 1 evaluated individual metrics from the index and repeated the
analyses. Although I found no strong relationships between individual metrics and pre-
removal assemblage characteristics, the presence of pool habitat strongly encouraged
recovery of ﬁsh abundance.

The positive relationship between the availability of pool habitat and ﬁsh
assemblage recovery suggests that habitat inﬂuences the recovery process. Deeper, slow-

moving pool habitat can serve as cover for stream ﬁsh (Harvey and Stewart 1991) and

30

may have encouraged individuals to colonize the study reaches. This habitat preference
suggests that pool habitat, which is typically lost when streams are channelized to
facilitate drainage, should be preserved in streams where ﬁsh conservation is a priority.
Further, multimetric approaches like the QHEI might mask the most important
parameters for ﬁsh assemblages. This emphasizes the importance of combining
biological sampling with physical indices such as the QHEI to assess the state of stream
ecosystems, despite the additional effort required to obtain meaningful results (Stauffer
and Goldstein 1997).

I was also interested in pre-removal assemblage inﬂuences on resilience.
Although the individuals present during the removal were not part of the colonist source
pool, I hypothesized that the pre-removal assemblage may indicate the structure of the
source pool and the suitability of the stream habitat within the study reaches. Sampling
of stream habitat upstream and downstream of the study reaches typically revealed ﬁsh
assemblages similar to those found within the study reaches (unpublished data). My
ﬁnding that pre-removal species richness was unrelated to the rate and extent of richness
recovery is consistent with previous work (see Detenbeck et al. 1992 for a review);
species-poor streams recovered richness as well as speciose systems. This supports my
earlier suggestion that members of all species in this system, both those considered
mobile, such as cyprinids, and more sedentary species, such as darters, regularly travel
outside individual channel units. These movements encourage the colonization of newly
available habitat (Larson et al. 2002).

The negative relationship between pre-removal ﬁsh abundance and abundance

recovery in my study reaches is worthy of consideration. Study reaches with fewer than

31

100 individuals per 100 m generally recovered abundance rapidly and completely, while
reaches with more than 175 individuals tended not to recover fully within the month of
observation. This may have been related to the quality of habitat available in adjacent
stream reaches which provided potential colonists for the empty habitat. Study reaches
with initially small populations may have represented relatively poorer habitat than
source reaches, and simple diffusive movement ﬁ'om higher-quality, more densely
populated source reaches may have been adequate for assemblage recovery. Conversely,
densely populated study reaches may have been slow to recover in numbers because
those reaches may have represented the best available local habitat. The majority of local
ﬁsh may have been concentrated at that site, and following my removal of those
individuals, adjacent source areas did not support a sufficient number of potential
colonists to permit complete abundance recovery. This is consistent with the view that
streams are heterogeneous systems composed of differing patches of habitat due to local
disturbances and colonization dynamics (Townsend 1989, Melo et al. 2003). This
heterogeneity combined with the apparent mobility of stream ﬁshes emphasizes the need
for stream conservation over long reaches that can support a variety of habitat types to
meet the needs of a diverse biota.

Finally, the quality of ﬁsh assemblages as measured by Procedure 51 (MDEQ 1997)
was unrelated to the recovery of species richness or abundance after defaunation. The
Procedure 51 score is derived from four taxonomic metrics, three trophic metrics, two
tolerance metrics, and one reproductive metric. Due to the rapid recovery of assemblage
structure, I can state that there were not great differences in colonization ability among

taxa in the study streams at the temporal and spatial scale I observed. Further, Detenbeck

32

et al. (1992) found that trophic guild was not a good predictor of resilience. Therefore,
since seven of ten Procedure 51 metrics were unlikely to relate to resilience of my ﬁsh
assemblages, the lack of relationship is not surprising.

While the rapid recovery I observed should not be used to justify disturbance to
streams or stream ﬁshes (Sheldon and Meffe 1995), it does have positive implications for
stream ﬁsh management. For example, small-scale ﬁsh kills due to temporary pollutant
inputs should recover quickly without aid as long as source populations are available
nearby and habitat remains unimpaired. Likewise, high resilience suggests that restored
stream reaches from which ﬂow is temporarily diverted should be rapidly colonized if
relatively short reaches are restored at a time and source ﬁsh populations are unimpeded.

We are left with the question of how well current assessment techniques evaluate the
health of stream ﬁsh communities. Multimetric indices such as the QHEI, and Procedure
51 and other indices of biotic integrity, are widely used to assess the health of stream
ecosystems (Karr 1987, Lyons 1992, Stauffer and Goldstein 1997). They are efﬁcient in
terms of cost and time, but may not tell the whole story. Ecosystem health has been
equated with sustainability, the ability to maintain structure and function in the presence
of disturbance (Meyer 1997, Costanza and Mageau 1999). This suggests, logically, that
resilience is a primary component of stream health, although it is not a part of any stream
assessment index of which I am aware. A review of the recovery of temperate stream
ﬁsh communities from disturbance led Detenbeck et al. (1992) to conclude that currently
we can not predict recovery times, because they are a function of disturbance-speciﬁc,
site-speciﬁc, and species-speciﬁc factors. It is not clear that the resilience of a system

can be predicted using any kind of universal metric or index. However, it does appear

33

that there are a number of characteristics of organisms (e.g., life history, dispersal
ability), assemblages (e. g., spatial heterogeneity, degree of isolation), and disturbances
(e.g., temporal and spatial scale, residual effects) that we can identify and use to predict,
in a relative sense, the ability of a system to recover from disturbance (Detenbeck et a1.
1992, Peterson and Bayley 1993, Sheldon and Meffe 1995). For example, ﬁsh
assemblage recovery from a temporary, or pulse, disturbance such as an experimental
defaunation is typically rapid, while long-term, or press, disturbances such as
channelization may permanently disrupt the system unless mitigation actions are taken
(Detenbeck et al. 1992). The exceptionally rapid recovery I observed in these
experiments is likely attributable to (1) undisturbed habitat, (2) a mobile colonist
assemblage dominated by cyprinids and centrarchids, and (3) the relatively small spatial
scale of the disturbance (Sheldon and Meffe 1995). Additional scales and types of
disturbance should be investigated to expand our understanding of ﬁsh assemblage
resilience and its relevance to stream health.

Further, my results suggest that the relationship between resilience and metrics
typically used to assess stream health, such as assemblage structure and overall habitat
quality, is not a simple one. Valuable insights into overall stream health could potentially
be gained by combining physical and biological assessments with experimental
determinations of the ability of these systems to recover from disturbance. With
improved understanding, metrics may be developed to estimate recovery potential

without the need to disturb the stream system.

34

Acknowledgments

Invaluable assistance was provided in the ﬁeld and laboratory by Kelly
DeGrandchamp, Janice Karcher, Kevin Mann, Colleen McLean, Cassandra Meier, Aaron
Shultz and Alfred Thomson. This research was supported by Michigan State University
while supported by a University Distinguished Fellowship from Michigan State
University and a National Needs Water Science Fellowship from the United States

Department of Agriculture.

35

Literature Cited

Bain, M. B., J. T. Finn, and H. E. Booke. 1988. Strearnﬂow regulation and ﬁsh
community structure. Ecology 69:382-392.

Berra, T. M., and G. E. Gunning. 1970. Repopulation of experimentally decimated
sections of streams by longear sunﬁsh, Lepomis megalotis (Raﬁnesque).
Transactions of the American Fisheries Society 99:776-781.

Costanza, R., and M. Mageau. 1999. What is a healthy ecosystem? Aquatic Ecology
33: 105-1 15.

Detenbeck, N., P. DeVore, G. J. Niemi, and A. Lima. 1992. Recovery of temperate-
stream ﬁsh communities from disturbance: a review of case studies and synthesis
of theory. Environmental Management 16:33-53.

Dunne, T., and L. B. Leopold. 1978. Water in environmental planning. W.H. Freeman,
San Francisco, CA.

Ensign, W. E., K. N. Leﬁwich, P. L. Angermeier, and C. A. Dolloff. 1997. Factors
inﬂuencing stream ﬁsh recovery following a large-scale disturbance. Transactions
of the American Fisheries Society 126:895-907.

Fisher, S. G., L. J. Gray, N. B. Grimm, and D. E. Busch. 1982. Temporal succession in a
desert stream following ﬂash ﬂooding. Ecological Monographs 52:93-110.

Gerking, S. D. 1953. Evidence for the concepts of home range and territory in stream
ﬁshes. Ecology 34:347-365.

Gunning, G. E., and T. M. Berra. 1968. Repopulation of a decimated stream segment by
the sharpﬁn chubsucker. The Progressive Fish-Culturist 30:92-95.

Gunning, G. E., and T. M. Berra. 1969. Fish repopulation of experimentally decimated
sections in the headwaters of two streams. Transactions of the American Fisheries
Society 111:305-308.

Harvey, B. C., and A. J. Stewart. 1991. Fish size and habitat-depth relationships in
headwater streams. Oecologia 87:336-342.

Hawkins, C. P., J. N. Hogue, L. M. Decker, and J. W. Feminella. 1997. Channel
morphology, water temperature, and assemblage structure of stream insects.
Journal of the North American Benthological Society 16:728-749.

Holling, C. S. 1973. Resilience and stability of ecological systems. Annual Review of
Ecology and Systematics 4:1-23.

36

Horton, W. M. 1969. Species composition and distribution of the Red Cedar River
system with detailed descriptions of collecting stations. Master of Science.
Michigan State University, East Lansing, Michigan.

Iowa State University. 1997. Stewards of our streams: riparian buffer systems. Iowa State
University Extension.

Karr, J. R. 1987. Biological monitoring and environmental assessment: a conceptual
framework. Environmental Management 11:249-256.

Kinsolving, A. D., and M. B. Bain. 1993. Fish assemblage recovery along a riverine
disturbance gradient. Ecological Applications 3:531-544.

Krumholz, L. A., and W. L. Minckley. 1964. Changes in the ﬁsh population in the Upper
Ohio River following temporary pollution abatement. Transactions of the
American Fisheries Society 93:1-5.

Larimore, R. W., W. F. Childers, and C. Heckrotte. 1959. Destruction and re-
establishment of stream ﬁsh and invertebrates affected by drought. Transactions
of the American Fisheries Society 88:261-285.

Larson, G. L., R. L. Hoffman, and S. E. Moore. 2002. Observations of the distributions of
ﬁve ﬁsh species in a small Appalachian stream. Transactions of the American
Fisheries Society 131 :791-796.

Li, H. W., G. A. Lamberti, T. N. Pearsons, C. K. Tait, J. L. Li, and J. C. Buckhouse.
1994. Cumulative effects of riparian disturbances along high desert trout streams

of the John Day Basin, Oregon. Transactions of the American Fisheries Society
123:627-640.

Likens, G. E., F. H. Borrnann, N. M. Johnson, D. W. Fisher, and R. S. Pierce. 1970.
Effects of forest cutting and herbicide treatment on nutrient budgets in the
Hubbard Brook watershed-ecosystem. Ecological Monographs 40:23-47.

Lyons, J. 1992. Using the index of biotic integrity (IBI) to measure environmental quality
in warmwater streams of Wisconsin. NC-149, North Central Forest Experiment
Station, USDA Forest Service, St. Paul, Minnesota.

Matthews, W. J ., R. C. Cashner, and F. P. Gelwick. 1988. Stability and persistence of ﬁsh
faunas and assemblages in three Midwestern streams. Copeia 1988:945-955.

Meffe, G. K., and A. L. Sheldon. 1990. Post-defaunation recovery of ﬁsh assemblages in
southeastern blackwater streams. Ecology 71 :657-667.

37

Melo, A. S., D. K. Niyogi, C. D. Matthaei, and C. R. Townsend. 2003. Resistance,
resilience, and patchiness of invertebrate assemblages in native tussock and
pasture streams in New Zealand after a hydrological disturbance. Canadian
Journal of Fisheries and Aquatic Science 60:731-739.

Meyer, J. L. 1997. Stream health: incorporating the human dimension to advance stream
ecology. Journal of the North American Benthological Society 16:439-447.

Michigan Department of Environmental Quality (MDEQ). 1997. GLEAS Procedure #51:
Qualitative biological and habitat survey protocols for wadable streams and
rivers. Surface Water Quality Division, Great Lakes and Environmental
Assessment Section, Lansing, Michigan.

Michigan Department of Natural Resources (MDNR). 2001. Spatial Data Library.
(http://www.dnr.state.mi.us/spatialdatalibrary/index.htrn). Accessed 12 April
2001.

Morisita, M. 1959. Measuring of interspeciﬁc association and similarity between
communities. Memoirs of the Faculty of Science, Kyushu University, Series E
(Biology) 3:65-80.

Niemi, G. J ., P. DeVore, N. Detenbeck, D. Taylor, A. Lima, J. Pastor, J. D. Yount, and R.
J. Naiman. 1990. Overview of case studies on recovery of aquatic systems from
disturbance. Environmental Management 14:571-587.

Olmsted, L. L., and D. G. Cloutman. 1974. Repopulation after a ﬁsh kill in Mud Creek,
Washington County, Arkansas following pesticide pollution. Transactions of the
American Fisheries Society 1:79-87.

Peterson, J. T., and P. B. Bayley. 1993. Colonization rates of ﬁshes in experimentally
defaunated warmwater streams. Transactions of the American Fisheries Society
122: 199-207.

Pimm, S. L. 1984. The complexity and stability of ecosystems. Nature 3072321-326.
Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Richter, R. E.
Sparks, and J. C. Stromberg. 1997. The natural ﬂow regime: a paradigm for river

conservation and restoration. BioScience 47:769-784.

Pringle, C. M. 1997. Exploring how disturbance is transmitted upstream: going against
the ﬂow. Journal of the North American Benthological Society 16:425-438.

Rankin, E. T. 1989. The Qualitative Habitat Evaluation Index [QHEI]: rationale,

methods, and application. Ohio Environmental Protection Agency, Columbus,
Ohio.

38

Resh, V. H., A. V. Brown, A. P. Covich, M. E. Gurtz, H. W. Li, G. W. Minshall, S. R.
Reice, A. L. Sheldon, J. B. Wallace, and R. C. Wissmar. 1988. The role of

disturbance in stream ecology. Journal of the North American Benthological
Society 7:433-455.

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

Schoener, T. W. 1968. The Anolis lizards of Bimini: resource partitioning in a complex
fauna. Ecology 49:707-726.

Sheldon, A. L., and G. K. Meffe. 1995. Short-term recolonization by ﬁshes of
experimentally defaunated pools of a coastal plain stream. Copeia 1995:828-83 7.

Simpson, EH. 1949. Measurement of diversity. Nature 163: 688.

Smithson, E. B., and C. E. Johnston. 1999. Movement patterns of stream ﬁshes in a
Ouachita highlands stream: an examination of the restricted movement paradigm.

Transactions of the American Fisheries Society
128:847-853.

Stauffer, J. C., and R. M. Goldstein. 1997. Comparison of three qualitative habitat indices

and their applicability to prairie streams. North American Journal of Fisheries
Management 17:348-361.

Storck, T., and W. T. Momot. 1981. Movements of the creek chub in a small Ohio
stream. Ohio Journal of Science 81 :9-13.

Townsend, C. R. 1989. The patch dynamics concept of stream community ecology.
Journal of the North American Benthological Society 8:36-50.

Trautrnan, M. B. 1981. The ﬁshes of Ohio: with illustrated keys, Revised edition. Ohio
State University Press.

Vannote, R. L., G. W. Minshall, K. W. Cummins, J. R. Sedell, and C. E. Cushing. 1980.
The river continuum concept. Canadian Journal of Fisheries and Aquatic Science
37:130-137.

Wallace, J. B. 1990. Recovery of lotic macroinvertebrate communities from disturbance.
Environmental Management 14:605-620.

Walser, C. A., B. Falterrnan, and H. L. Bart. 2000. Impact of introduced rough shiner

(Notropis baileyi) on the native ﬁsh community in the Chattahoochee River
system. American Midland Naturalist 144:393-405.

39

Ward, J. V., and J. A. Stanford. 1983. The serial discontinuity concept of lotic
ecosystems. Pages 29-42 in T. D. Fontaine III and S. M. Bartell, editors.
Dynamics of lotic ecosystems. Ann Arbor Science, Ann Arbor.

Wolda, H. 1981. Similarity indices, sample size and diversity. Oecologia 50:296-302.

Wolman, M. G. 1954. A method of sampling coarse river-bed material. Transactions of
the American Geophysical Union 35:951-956.

Yount, J. D., and G. J. Niemi. 1990. Recovery of lotic communities and ecosystems from

disturbance - a narrative review of case studies. Environmental Management
14:547-569.

40

CHAPTER 2
The vigor-organization-resilience concept of ecological health: lessons from temperate

warmwater stream ﬁsh communities

Abstract

An ecologically healthy system is able to maintain vigor, or ﬁrnction, and
organization, or structure, in the presence of disturbance (resilience). Measures of
ecological vigor, such as abundance, productivity, and biomass, are relatively simple to
obtain in the ﬁeld, as are measures of organization, such as species richness and diversity.
However, determination of system resilience in the ﬁeld is less than straightforward
because of its dependence on the nature of the disturbance and on the choice of endpoints
for system recovery. I investigated the vigor, organization, and resilience of temperate
warmwater stream ﬁsh communities subject to controlled disturbance to determine what
relationships, if any, existed among these variables and stream habitat quality. I
measured community resilience in a number of ways, including level of and time to
recovery of ﬁsh density, richness, diversity, and community structure. I found no
signiﬁcant relationship between habitat quality and vigor or organization of the study
communities or their recovery, although some individual physical parameters were
signiﬁcant. Interestingly, there was no relationship between the pre-disturbance vigor or
organization of the ﬁsh community and its resilience to disturbance. While I found that
measuring post-disturbance changes in structure and abundance over time was useful in
determining recovery, these resilience values are applicable only to one type of

disturbance at one spatial and temporal scale. Although vigor and organization may

41

serve as partial indicators of community health, a measurement of resilience to various
types of disturbance would be invaluable for stream ﬁshery management and warrants

further research.

Introduction

Ecosystem management is based on our ability to assess the state of an ecological
system and, if necessary, take steps to maintain or improve the quality of the system.
Rich debate over assessment methods exists in the literature (Costanza 1992). This
debate is due, in part, to the difficulty in devising assessment methods that are
straightforward, inexpensive, and easily understood by stakeholders, yet comprehensive
and rigorous.

The concepts of ecosystem integrity and health were developed as frameworks for
assessment. Karr (1995) deﬁned ecosystem integrity as “the capacity to support and
maintain a balanced, integrated, adaptive biologic system having the full range of
elements and processes expected in the natural habitat of the region”. Therefore,
integrity is the maintenance of structure and function in the context of evolutionary and
biogeographic forces (Meyer 1997). Costanza and Mageau (1999) deﬁne a healthy
system as one that can maintain its function (vigor) and structure (organization) over time
in the presence of external stress (resilience). They go on to explain this concept in
another way: “A healthy ecosystem is one that can develop an efﬁcient diversity of
components and exchange pathways (high organization) while maintaining some
redundancy or resilience as an insurance against stress, and substantial vigor to quickly

recover or utilize stress in a positive manner”. They also equate health with

42

sustainability, noting that sustainability refers not to a system lasting forever, but
achieving its maximum life span within natural successional processes.

Measures of ecological vigor, such as abundance, productivity, and biomass, are
relatively simple to obtain in the ﬁeld, as are measures of organization, such as species
richness and diversity (Table 1). However, determination of system resilience in the ﬁeld
is less than straightforward. Resilience can be measured as the magnitude of disturbance
from which a system can recover (Holling 1973) or as the time required for a system to
recover from a disturbance (Pimm 1984). Comparisons among systems are difﬁcult
because they depend on the nature of the disturbance and on the choice of endpoints used
to determine system recovery. These complications may explain why ﬁeld measurement

of resilience has not received much attention as a health assessment tool.

Table 1. Components of ecological health (adapted from Costanza and Mageau 1999).

 

Variable Type Measures Methods
Vigor Static Abundance, Productivity, Direct measurement
Biomass
Organization Static Species richness, Direct measurement
Diversity, IBI
Resilience Dynamic Time, Path, or Level of Measurement over
Recovery time, Simulation
modeling

 

Because resilience refers to the ability of a system to recover structure and
function following disturbance, it serves as a dynamic stress test for assessing system

health and complements static measures of structure and function. Inclusion of resilience

43

as an indicator of system health is logically appealing because of its role in sustainability.
While system structure and function may appear healthy at a particular point in time, if
the system is not resilient to disturbance, it is not sustainable or healthy.

I feel that the concept of ecological health is a useful metaphor for
communication among scientists and with the public, despite its potential ﬂaws
(Wicklum and Davies 1995). It is not my goal in this paper to enter into the debate over
the ecosystem health concept; for an in-depth discussion of the beneﬁts and potential
pitfalls of the ecological health concept, see Constanza et al. (1992). Assessment of the
component variables of ecosystem health -— vigor (function), organization (structure), and
resilience — is relevant on ecosystem, assemblage, and population scales, and the health

concept provides an excellent framework for discussion.

I investigated the vigor, organization, and resilience of temperate warmwater
stream ﬁsh assemblages subject to controlled disturbance, focusing on the recovery
process. I had three goals in mind during our study. First, I investigated whether habitat
assessment measures, commonly used to indirectly assess stream health, were good
predictors of the vigor or organization of the ﬁsh assemblage. Second, I looked for
relationships among vigor, organization, and resilience that might further our
understanding of ﬁsh assemblage recovery from disturbance. Finally, I used the recovery
patterns observed in our experiments to develop a hypothesis about the recovery path

taken by stream ﬁsh assemblages based on the nature of the disturbances they might face.

44

Methods

I conducted a series of ﬁeld experiments to investigate the resilience of stream
ﬁsh assemblages to disturbance in the Red Cedar River watershed, Michigan, USA, a
warmwater system characterized by agricultural and, secondarily, urban development.
The experiments described here are reported in greater detail in Chapter 1. I assessed
stream habitat using the Qualitative Habitat Evaluation Index (QHEI; Rankin 1989). I
disturbed the systems by removing all ﬁsh from 100 m reaches at 14 sites in second- and
third-order streams of varying habitat quality and ﬁsh assemblage structure. The
composition of the pre-disturbance assemblage was recorded. I then returned to these
sites several times within a month to monitor ﬁsh assemblage recovery via catch-and-
release sampling. Throughout the study, ﬁsh abundance and species richness represented
vigor and organization, respectively. Resilience was measured as the maximum percent
recovery of vigor and organization during the study period and by calculating Morisita’s
(1959) index of similarity to compare pre-disturbance assemblage structure to that of

recovery samples.

Results

Neither pre-disturbance ﬁsh assemblage vigor, measured as the number of
individuals (range = 23-342), nor organization (species richness; range = 5-18) was
related to habitat quality as assessed using the QHEI (range: 16-65; Figure 1). Following
ﬁsh removal, assemblage organization was highly resilient, resulting in an average

Morisita’s similarity of 0.83 across the watershed within 2 days of ﬁsh removal (Figure

45

2a). This similarity was maintained through the course of the experiments. In
comparison, total ﬁsh abundance recovered more slowly but was increasing toward
complete recovery by the end of a month on average (Figure 2b). Pre-disturbance
organization did not inﬂuence organization recovery (Figure 3a), but pre-disturbance

vigor was negatively related to vigor recovery (Figure 3b).

 

 

 

400 1 -- 20
O
o
300 ~ 0 r_ 15
I. O O O .
’ V or
3' 200 + . ° 0 -~ ° '9 . .
> o 0 .0 . o 0 Organization
9 o
100 ~ . o . -t 5
. o
9 8
0 I I I 0
0 20 40 60 80

QHEI Hablht Score

Figure 1. Neither ﬁsh assemblage vigor (abundance) nor organization (species richness)
was related to habitat quality as measured by the Qualitative Habitat Evaluation Index

(QHEI; Rankin 1989).

46

(a)

 

 

 

 

 

 

g 0.8 ~ f’ I I T
g 0.6 -
C
o
‘8 0.4 -
.5
2
ﬂ
2’ 0.2 -
O
0 I I T I l l
0 5 1O 15 20 25 30 35
Days Post-dleturbance
(b)
100% -
80% 4
Z‘
c
g 60% d I I I/{/—{
a: I I
3 40% ~
.9
>
20% -
0% l I I I I I I
0 5 10 15 20 25 30 35

Days Poet-diaturba nee

Figure 2. Fish assemblage organization (richness and proportional abundance) was
highly resilient, showing strong similarity to pre-disturbance conditions throughout the
recovery period. Morisita’s (1959) similarity index ranges from 0 (completely dissimilar)
to just above 1 (completely similar) (a). Assemblage vigor (ﬁsh abundance) recovered
more slowly but increased over time (b).

47

(a)

(b)

Richneee Recovery

Vigor Recovery

1 00%

80%

60%

40%

20%

0%

1 00%

80%

60%

40%

20%

0%

 

 

 

     

 

~ 0 O O O O O O
’ o
- o
O O
_ o
O 5 10 15 20
Pro-disturbance Organization (Richneae)
- o o
R2 = 0.7033
- P = 0.0002
9
0 100 200 300 400

Pro-disturbance Vigor (Abundance)

Figure 3. No clear relationship was found between pre-disturbance organization and
organization recovery after one month (a). Pre-disturbance vigor was negatively related
to its recovery after one month (b).

48

Discussion

Neither ﬁsh assemblage vigor nor organization was related to habitat quality as
measured by the multimetric Qualitative Habitat Evaluation Index (QHEI). This index
was designed to assess stream habitat quality for supporting ﬁsh, and is often used in
place of biological sampling. The lack of relationship between QHEI score and the ﬁsh
community metrics suggests that habitat evaluation alone, at least via this method, is not
sufﬁcient to assess the health of a stream system, and, in fact, may be misleading. Pre-
disturbance assemblage organization, measured as species richness, was not a predictor
of richness recovery. Fish abundance (vigor) was negatively related to recovery,
suggesting that densely populated stream reaches may be more sensitive to disturbance.
Further discussions of the implications of the above ﬁndings are available in Chapter 1.

A conceptual model linking the component variables of ecological health — vigor,
organization, and resilience — can be developed to better understand and predict the path
of recovery after disturbance. In our ﬁsh removal experiments, both the vigor and
organization of the ﬁsh assemblage were decreased to zero. I predicted that, with habitat
and colonist pools intact, organization would recover more quickly than vigor
(abundance) as the stream reach was recolonized (Figure 4a). Our data support this
hypothesis. The recovery of assemblage organization, as measured with Morisita’s
similarity index, was nearly complete within 2 days after ﬁsh removal. Vigor, measured
as number of individuals, recovered more slowly and approached recovery within a

month.

49

Biotic Biotic plus

   
     

 

 

 

 

Disturbance Habitat
12 Disturbance
' 1
1 . :° .3
0.8‘ ‘¢¢."‘ ."‘.
0.0 - 5
0.4 «5
02;? .
' 0. ...?.Y: I T T T 1
0 0.2 0.4 0.0 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Organization Organization

Figure 4. Paths of assemblage recovery (dotted arrow) following disturbance (bold
arrow) of biota only, as in our experimental ﬁsh removals. (a; 0 represents the mean
value across all sites for each time interval), and both biota and habitat (b).

50

I suggest that the path of recovery taken by the ﬁsh assemblage is inﬂuenced by
the nature of the disturbance. The direction and extent of change due to disturbance
would affect both the path and rate of recovery. For example, if I had not only removed
ﬁsh from the experimental stream reaches, but also disturbed the habitat, I expect that
assemblage vigor would recover more quickly than organization (Figure 4b). Degraded
habitat would no longer be suitable for the full complement of ﬁsh species available in
the adjacent colonist pools, so the experimental habitat would likely be rapidly colonized
by generalist species, perhaps in substantial numbers, while less tolerant species would
not recolonize until the habitat had recovered. As a result, the path of resilience would be
different than in cases where habitat is undisturbed, and the time scale of biotic recovery
would be greater as it would depend on habitat recovery.

This conceptual model of ﬁsh assemblage recovery has application in making
comparisons between systems and for understanding the mechanisms behind the
ecological recovery process. However, from a management standpoint, a complex
system descriptor that requires measurement over time and is dependent on a number of
extraneous variables, such as the type and scale of disturbance, is not particularly
desirable in terms of cost and time. So, although system resilience is logically an
important component of system health, it may not be feasible to measure in the ﬁeld
when a large number of systems are being considered. Because comparison depends on
equivalent disturbances, it would be necessary to deliberately disturb the set of systems of
interest, and then frequently resample these systems to determine progress along the path

of recovery.

51

This large investment of effort led Costanza and Mageau (1999) to suggest
simulation modeling as the best way to determine system resilience and make
comparisons among systems. However, even this approach requires a large initial
investment in gathering data to construct and validate the model. Further, prediction of
system recovery would again be dependent on the nature of the disturbance and the
choice of endpoints for determining recovery. Any framework using resilience as a
relative measure of health in a set of systems, such as warmwater stream ﬁsh
communities, would require the adoption of a particular disturbance type against which to
measure recovery, and would have to be, at most, regional in scope, because variables
such as mobility of potential colonists and habitat heterogeneity vary from region to
region.

A review of the recovery of temperate stream ﬁsh communities from disturbance
led Detenbeck et al. (1992) to conclude that currently we can not predict recovery times,
because they are a function of disturbance-speciﬁc, site-speciﬁc, and species-speciﬁc
factors. It is not clear that the resilience of a system can be predicted using any kind of
universal metric or index. However, it does appear that there are a number of
characteristics of organisms (e.g., life history, dispersal ability), communities/ ecosystems
(e. g., spatial heterogeneity, degree of isolation), and disturbances (e.g., temporal and
spatial scale, type of disturbance, residual effects) that we can identify and use to predict,
in a relative sense, the ability of a system to recover from disturbance. I feel that further
investigation of the features of ecological systems and the disturbances that affect them
will allow us a better understanding of system resilience and health, and should be

pursued. Although vigor and organization may serve as partial indicators of health, a

52

clearer understanding of resilience dynamics would be invaluable for stream ﬁshery

management.

Acknowledgments

I was assisted in the ﬁeld by a talented group of undergraduates: K.
DeGrandchamp, J. Karcher, K. Mann, C. McLean, C. Meier, A. Shultz, and A. Thomson.
Funding was provided by Michigan State University and the USDA National Needs

Water Science Program.

53

Literature Cited

Costanza, R. 1992. Toward an operational deﬁnition of ecosystem health. Pages 239-256
in R. Costanza, B. G. Norton, and B. D. Haskell, editors. Ecosystem health: new
goals for environmental management. Island Press, Washington, DC.

Costanza, R., B.G. Norton, and B.D. Haskell, editors. Ecosystem health: new goals for
environmental management. Island Press, Washington, DC.

Costanza, R., and M. Mageau. 1999. What is a healthy ecosystem? Aquatic Ecology
33:105-115.

Detenbeck, N., P. DeVore, G. J. Niemi, and A. Lima. 1992. Recovery of temperate-
stream ﬁsh communities from disturbance: a review of case studies and synthesis
of theory. Environmental Management 16:33-53.

Holling, C. S. 1973. Resilience and stability of ecological systems. Annual Review of
Ecology and Systematics 4:1-23.

Karr, J. R. 1995. Ecological integrity and ecological health are not the same. Pages 97-
109 in P. Schulze, editor. Engineering within ecological constraints. National
Academy of Engineering. National Academy Press, Washington, DC.

Meyer, J. L. 1997. Stream health: incorporating the human dimension to advance stream
ecology. Journal of the North American Benthological Society 16:439-447.

Morisita, M. 1959. Measuring of interspeciﬁc association and similarity between
communities. Memoirs of the Faculty of Science, Kyushu University, Series E
(Biology) 3:65-80.

Pimm, S. L. 1984. The complexity and stability of ecosystems. Nature 307:321-326.

Rankin, E. T. 1989. The Qualitative Habitat Evaluation Index [QHEI]: rationale,
methods, and application. Ohio Environmental Protection Agency, Columbus,

Ohio.

Wicklum, D., and R. W. Davies. 1995. Ecosystem health and integrity? Canadian Journal
of Botany 73:997-1000.

54

CHAPTER 3
Habitat selection by creek chubs (Semotilus atromaculatus Mitchill) following

experimental defaunation

Abstract

Ideal free distribution theory suggests that animals will occupy the most suitable
habitats available. Suitability can be deﬁned as the ﬁtness that a particular habitat affords
the individuals that occupy it. Therefore, an understanding of habitat selection under a
range of conditions will yield information about the relative suitability of adjacent
habitats and potentially reveal habitat characteristics for emphasis in species
conservation. Traditional methods of assessing habitat suitability rely on correlating
habitat characteristics with the number of individuals occupying the habitat under natural
conditions. Distribution is a behavioral phenomenon, however, and the majority of these
correlations do not consider movement or density-dependent interactions that may
inﬂuence the suitability of a particular habitat. In this study, I considered both the
inﬂuence of physical habitat and density of conspeciﬁcs on distribution during
experimental defaunation of ﬁshes from warmwater stream reaches. Habitats were left
intact, and the distribution and movement of one species, the creek chub (Semotilus
atromaculatus Mitchill), were tracked as densities increased via habitat recolonization.
As observed in previous studies, chubs preferred pools over runs, and runs over rifﬂes.
Alternative methods of measuring selection, including selection indices and immigration

rates, revealed similar insights. Density of creek chubs in the study reaches, which

55

ranged from zero immediately following defaunation to above pre-defaunation (natural)

densities, had no apparent effect on habitat preferences.

Introduction

Habitat selection by animals results in distribution patterns that can be observed in
the ﬁeld and laboratory. These distribution patterns yield information about the
suitability of different available habitats for the species of interest. Field observations of
animals in their natural environments, and laboratory and ﬁeld manipulations of animals
and habitat, provide clues about the particular characteristics of a given habitat that make
it more or less suitable for the species.

Habitat suitability can be deﬁned as the ﬁtness that a particular habitat affords the
individuals that occupy it (F retwell and Lucas 1970). Therefore, suitability is the product
of birth, growth, and death rates and reproductive success for a species in a particular
habitat. Traditional methods of calculating habitat suitability use the density of
individuals at a site as a proxy for these more complicated variables, based on the
assumption that animals will occupy the most suitable habitat available. This assumption
is based on theories of species distribution, including the ideal free distribution described
by Fretwell and Lucas (1970). Ideal free distribution theory suggests that if all
individuals are free to move into any habitat, they will settle in the habitat most suitable
to them (i.e., offering the highest ﬁtness).

For example, the habitat suitability indices (HSIs) produced by the US. Fish
and Wildlife Service primarily are based on literature reviews of distributions and

correlations with physical habitat characteristics. These data are used to determine the

56

best habitat for a species (e. g., McMahon 1982). HSIs do not consider density-dependent
effects on habitat suitability, such as competition for resources, nor do they consider
behavior or changes in distribution over time; however, these indices frequently form the
basis of habitat management and species conservation efforts. Further, these models are
largely untested and often refuted in experimental studies (Hubert and Rahel 1989).

Describing habitats by physical attributes alone may not be sufﬁcient to explain
distributions, and is possibly one of the biggest shortcomings of suitability indices like
the H81. Distribution is a behavioral phenomenon, the result of individual movement.
Animal movement can be inﬂuenced not only by the physical environment but also by
interactions with other animals, many of which are mediated by density (Fraser and
Emmons 1984, Strange et al. 1993).

In a study of stream ﬁshes, Belanger and Rodriguez (2002) suggested that
measures of local movement, such as habitat-speciﬁc immigration and loss rates, provide
useful indicators of habitat quality and provide a more reliable mechanistic basis for
understanding than point observations of density. They argue that simple correlations of
density to habitat type are weakened by transient ﬂuctuations in abundance, seasonality,
territorial interactions, and density dependence of habitat selection. Their movement-
based model predicted lower turnover rates among ﬁsh occupying optimal habitat than
among those in sub-optimal habitat. The authors reported that immigration rates were
strongly correlated with species-speciﬁc habitat preferences observed in previous studies,
and provided an integrated assessment of habitat quality as perceived by ﬁsh. They

suggested that because immigration rate seems to reﬂect the attractiveness of a unit,

57

habitat types that tend to show high immigration rates would be colonized most rapidly
following transient declines in density, such as experimental defaunation.

In this study, I considered both the inﬂuence of physical habitat characteristics
and density of conspeciﬁcs on distribution during recolonization following experimental
defaunation of ﬁshes from stream reaches. Habitats were left intact, and the distribution
and movement of one species, the creek chub (Semotilus atromaculatus Mitchill), were
tracked as densities increased via habitat recolonization. I sought to answer the following
questions. What are the habitat preferences of creek chubs in our study streams? Do
different methods of measuring suitability yield similar results? Finally, do habitat
preferences change with the density of individuals in the stream reach?

Stream ﬁshes are well suited for examining the behavioral mechanisms and
habitat characteristics that lead to species distributions, because habitat units are arranged
in a linear fashion, distinct habitat units are typically easily deﬁned, and many species are
abundant and easy to collect and identify. While early studies suggested that stream
ﬁshes exhibit only limited movement within streams (Gerking 1953), more recent work
has documented extensive movement by at least a portion of the populations of many
species (Smithson and Johnston 1999, Albanese 2001). Movement among habitat units
by stream ﬁshes allow them to sample available habitat and occupy the most suitable
units available, making them more likely to conform to an ideal free distribution
(Fretwell and Lucas 1970).

I chose to focus on creek chubs, the most abundant minnow in the eastern United
States (Scott and Crossman 1973). They are widespread, are relatively long-lived, and

adults are often the top predator in small streams, making life stages of creek chubs

58

important on many levels of the trophic web (Copes 1978). Their ubiquity and
abundance have made them a candidate “sentinel species” for assessing stream health
(Fitzgerald et al. 1999), and make them an ideal species for examining distribution
mechanisms in the ﬁeld and laboratory. Further, the preference of adult creek chubs for
pool habitat has been noted in an HSI (McMahon 1982) and other studies (Moshenko and
Gee 1973, Copes 1978, Storck and Momot 1981, Hubert and Rahel 1989, Harvey and

Stewart 1991) and forms a basis for comparison for the results of this study.

Methods

I conducted ten ﬁsh removal experiments in 100 m reaches of nine different
streams in the Red Cedar River watershed, Michigan, during the summers of 2001 and
2002. These experiments are a subset of those described in Chapter 2; only sites that
contained at least two of the three possible major channel unit types (pools, rifﬂes, or
runs; Dunne and Leopold 1978) were considered for this analysis.

The ten experimental stream sites averaged 103 m in length (range: 96.1-117.6 m)
and 502 m2 in surface area (range: 297.5-1090.12 mz‘ Table 1). Wetted width ranged
from 0.9 — 13.8 m (median: 3.8 m). Five of the ten sites included pool, rifﬂe, and run
habitat. Three contained only rifﬂes and runs, and two contained only pools and runs.

I removed all ﬁsh from the study sites via electroﬁshing. Species and pool, rifﬂe,
or run location within the study site were recorded for each individual collected. I
relocated all ﬁsh collected during each removal to another stream site 2-10 km away from
the study site to discourage their return to the study site; creek chubs generally move less

than 1 km during spring and summer (Storck and Momot 1981). I then re-sampled each

59

study site frequently during the following month to track ﬁsh assemblage recovery. See
Chapter 2 for a detailed description of the methods used to characterize the stream habitat

and conduct the ﬁsh removal experiments.

60

 

.25. 3.5 .5.

 

 

NN... - .N... N c . 2. .. 2 e3. 8. S. .555 «am 3

3... 3... - N . o No N c NNN 8. N .85 :03 N.
33— «58:30

N... - N4... N c N E .. N 5 S. 3. 5.86 Son. N.

N... 8... .N... N . . NN s .. N. N 8. a... 0.86 .25 . .
.33. 352

N... S... N... N N N N4 NN N No 3 we 5.86 58 c.
83. :35 5>

2... 2... .N... v . N NN NN NN NNN N... 3. 5.86 U22.5.3. N

N... N . .c - N . c a o. .. 8.: N... Z. 0.86 82.3on o

.N... 2... - N N c 3 «N o NNN N: N... 0.86 :85 N
base;

3... N... 3.... e N N N o. 2. EN S. N .33. 528 N
0535

2... :5 N... N N o .N N N NNN No. 3 .33. 038 .

=3. OER .2... 23. 8E... n.8,. :3. .x. 05.5.: .2... ox. .38. :5 E 0552 a

Swan; 52>»
3% 0432 mag as: ea

 

Hosp—8m? 53M .8on com 05 E Noam 3688 3mm Ecuatomxm A 033.

61

I assessed habitat selection by creek chubs both with Strauss’ linear selection

index (Strauss 1979),
Li = oi — in ,

and Manly’s standardized selection ratio (Manly et al. 2002),
I
Bi = v‘vi/ Xvi/i ,
i=1

Where 0, = sample proportion of used units in category i,

it, = proportion of the available units in category i, and

M = o/ m, the selection ratio.

Strauss’ linear selection index simply shows the difference between the
proportional availability of a habitat and its proportional use. A positive value indicates
that a habitat was used in greater proportion than it was available, suggesting positive
selection for that habitat. Manly’s standardized selection ratio converts the same data
into a ratio that is then standardized to represent the probability that an individual will use
a particular habitat type (Manly et al. 2002). Both the linear selection index and the

selection ratio are commonly used to quantify selection among habitats by individuals.

62

Immigration rate (individuals per day) was calculated as

I_(nt1—nt0)
tl—tO '

The linear correlation between the total number of creek chubs removed from
each study site and the number collected in the ﬁrst electroﬁshing pass during
defaunation was very strong (y = 0.819x — 11.931, R2 = 0.95). Because of this strong
linear relationship, the number of creek chubs collected during the single electroﬁshing
passes conducted during recovery sampling of the study sites was used as an estimator of
the total density of creek chubs at each site.

I used a general linear model where site was a blocking factor and density was a
continuous covariate to quantify the effects of these variables on creek chub habitat
preferences. Interaction terms were evaluated to determine if the response of habitat
selection to density varied across sites. Selection for pools, riffles, and runs were

evaluated separately.

Results

The total number of creek chubs removed per site ranged from 10 to 198
individuals, resulting in a median density of 0.071 ﬁsh/m2 (range: 0.021-0.62 ﬁsh/m2).
By comparing ﬁrst-pass numbers from the removal to subsequent single-pass return
catches, I determined that creek chubs recovered to or above removal densities at six out

of ten sites during the month-long duration of the experiments (Table 2).

63

Table 2. Creek chub removal and recovery ﬁom experimental stream sites in the Red

 

 

 

Cedar River watershed.
$39 Removal First pass no. of thy
# Name Total Densi Removal Max. %
Removed (ﬁsh/m ) Recovery Recovery
1 Dobie Road Tributary 66 0.19 31 107 >100
2 Corwin Road Tributary 198 0.62 161 38 24
3 Sloan Creek 40 0.07 1 l 22 >100
6 Sycamore Creek 27 0.03 16 13 81
9 Kalamink Creek, Van Orden Rd 10 0.03 6 83
10 Doan Creek, Noble Road 13 0.02 6 >100
l 1 Deer Creek 63 0.12 20 37 > 100
12 Dean Creek, Columbia Road 51 0.12 23 41 >100
13 Wolf Creek 20 0.07 12 58
14 East Branch, Red Cedar River 13 0.02 5 >100

 

64

Both Strauss’ linear selection index and Manly’s standardized selection ratio
indicated that in stream reaches with pools, rifﬂes, and runs available, creek chubs
generally preferred pools over runs, and runs over riﬁles (Figure 1). If no distinct pool
habitat existed, creek chubs continued to prefer runs over rifﬂes (Figure 2). Interestingly,
in the two sites with only runs and pools, runs were preferred over pools (Figure 3).

These preferences were not inﬂuenced by changes in creek chub density over the
course of the experiments. Density in the study sites ranged from zero (immediately
following defaunation) to 0.41 ﬁsh/m2. When selectivity was plotted against density of
creek chubs in the experimental reaches, it appears that selectivity is strongest at low
densities, and the distribution becomes more even as densities increase (Figures 1-3).
However, when the data are examined on a site-by-site basis, it becomes clear that stream
site has a much stronger effect on selectivity than does density. For example, when
Strauss’s selection index is plotted against site density (Figure 4), sites with broad ranges
of creek chub density during the experiment (i.e., sites 1 and 2) show a fairly constant
strength of creek chub selection for pools. Sites where densities remained relatively low
showed substantial ﬂuctuations in selection strength. Similar site-clumped distributions

were observed with runs and rifﬂes (Figures 5 and 6).

65

A
3

Strauss' Selection Index

0))

Manly's Selection Ratio

1
0.8 if
0.6

—1

 

 

 

 

0,4 4 o o Pools
0 2 4 ° ° ° o R
. O ‘8 O O o O O uns
0, . .“e o. O. O xRifﬂes
o
_02 X? X XX XX XX )s< X X
{a
-0.4 p
‘0-6 7 I T T 1 I l
0 0.1 0.2 0.3 0.4 0.5 0.6
Site Density (ﬂshlm‘Z)
1 r"
O
0.75 4
o o o o o Pools
0 o O O R
0.5 -. £0 8. o o . o uns
e x Rimes
O . ..
0.25 - . ° x’ O
x
x x ’S< x
0 «x<—:x><r>-x . . . ?‘ .
0 0.1 0.2 0.3 0.4 0.5 0.6

Site Density (fidrlm‘2)

Figure 1. Creek chub habitat selection versus density of chubs in the experimental sites
containing pools, rifﬂes, and runs as assessed with (a) Strauss’ (1979) linear selection
index and (b) Manly’s standardized selection ratio (Manly et al. 2002).

66

(a)

 

 

 

 

0.6 1
e o
x X e
.3 0.4 A e
E e
c 0.2 - X
° 0
g 0 4 ‘. x 0 9 9 Runs
3 ,gi‘x . x x x Riﬂles
' -0.2 . o
g x
.8 -0.4 ~ x
in e X
x x
'0.6 I I I I I
0 0.01 0.02 0.03 0.04 0.05
Site Density (flehlm‘2)
(b)
1 — 00x 0 e e

e
0 e
a
g 0.75 - x °
C
o

x R

E 0.5 - ; ’ ”"8
(g e x Rlﬂles
.0
2’ _
5 0.25 . x
s X

x

0 ——>xx-< r x I x 1 x .
0 0.01 0.02 0.03 0.04 0.05

Site Density (ﬁshlm‘Z)

Figure 2. Creek chub habitat selection versus density of chubs in the experimental sites
containing only rifﬂes and runs as assessed with (a) Strauss’ (1979) linear selection index
and (b) Manly’s standardized selection ratio (Manly et al. 2002).

67

(a)

 

 

 

0.3 -
m
5 0.2 «
13
.5 O... O 9
c 0.1 a 0
° 0
g 0 o Pools
‘3 o 0 Runs
- -0.1 . °
g- 0000 0 °
3 -0.2 4
to
000
.0-3 I I I I I I
0 0.02 0.04 0.06 0.08 0.1 0.12
Slte Density (ﬁsirlm‘Z)
(b)
1 '1...
o
2 0.75 ~ 0 e
o O
E 0.5 “ o Pools
8 o . 0 Runs
D“ o o
>
T: 0.25 -
a
E
0 ‘m I I I I I I

 

0 0.02 0.04 0.06 0.08 0.1 0.12
Site Density (ﬂehlm‘Z)

Figure 3. Creek chub habitat selection versus density of chubs in the experimental sites
containing only pools and runs as assessed with (a) Strauss’ (1979) linear selection index
and (b) Manly’s standardized selection ratio (Manly et al. 2002).

68

 

 

5 5
0.8 4
34345
0.6 ~ 5
5
'o
5
g 0.4 ~ 53
3 5
3. 1
° 1 1
in 02 1
g . I 4 2 1
8 1 2 1 2
iii 22 a
2
o _
8
008 8
-0.2 .
4488
0.4 I I T I T 1
0 0.1 0.2 0.3 0.4 0.5 0.6

Site Density (ﬂshlm‘2)

Figure 4. Strauss’s linear selection index for pools plotted against site density for each
observation. Numeric data markers represent individual sites.

69

0.6 -

 

 

66
6
0.4 -
6
6
E
0.2 ~
5 5 5
3 85 8
c 77779 9
O 68 22 2 1 2
g o 4 5 1 2 1 1
'5 7 1 2 1
U)
3 8 1
a
E
w -0.2 - 33
4
-0.4 4
444
6
33
.0-6 I I I I I I
0 0.1 0.2 0.3 0.4 0.5 0.6

Site Density (ﬂehlm‘2)

Figure 5. Strauss’s (1979) linear selection index for runs plotted against site density for
each observation. Numeric data markers represent individual sites.

70

0.6 -

0.4 i

.0
N
r

N

2221 22 1 2

Strauss' Selection Index
0

.C'>

N

J
.3
-tl

3331

6
444
6
-0.4 -

66

-0.6 I I I I I I
0 0.1 0.2 0.3 0.4 0.5 0.6

Density (ﬂshlm‘Z)

 

 

Figure 6. Strauss’s (1979) linear selection index for rifﬂes plotted against site density for
each observation. Numeric data markers represent individual sites.

71

Analysis of variance further shows that the effect of site on habitat selection is
signiﬁcant, while there is no effect of density. Selection for pool habitat by creek chubs
did not show any site by density interaction (df=6, F=0.55, P=O.7683), indicating that the
response to density was similar across all sites. A reduced model, dropping the
interaction term, showed that selection varied across sites (df=6, F=13.54, P<0.0001) but
did not vary with density (df=1, F=0.15, P=O.7042). Selection for run habitat similarly
did not show any site by density interaction (df=9, F=1.26, P=0.2919). The reduced
model showed that selection again varied across sites (df=9, F=8.49, P<0.0001) but not
with density (df=1, F=0.26, P=0.6155). Finally, no interaction of density with site was
found in selection for rifﬂes (df=7, F=0.86, P=0.5465). The reduced model indicated that
selection did not vary with density (df=1, F=0.16, P=0.6936) or site (df=7, F=2.25,
P=0.0551).

Immigration rates for pools, rifﬂes and runs were calculated twice — once using
collections from the ﬁrst observation (typically 2, but as many as 6 days aﬁer
defaunation), and using results from the last observation (after approximately a month).
Both methods revealed the same patterns as Strauss’ and Manly’s selection indices.

Pools had higher immigration rates than riffles, and rifﬂes had higher rates than runs
when all three were available (Figure 7). In sites with no pools, creek chubs immigrated
into runs at a higher rate than rifﬂes (Figure 8). Where only pools and runs existed, creek

chubs moved into runs with much higher frequency than pools (Figure 9).

72

(a)

 

 

 

12 -
It;
.2 10 “
‘5
E 3 ‘ .
o lRiﬂies
& 6- DRuns
g 4‘ lPools
i!
.9
E 2-
E N D

0 ‘ I I I %

1 2 9 10 11
Site
(b)

1.6
3 1.4
E
E 1.2]
I; 1IRifiies
g 0.8 DRuns
_§ 0.6 lPools
g, 0.4
E
g 0.2 D

o I - I - I I

1 2 9 10 11
Site

Figure 7. Immigration rates of creek chubs colonizing each habitat unit class in streams
containing pools, rifﬂes, and runs. Immigration was calculated for (a) the ﬁrst return
interval, 2 to 6 days aﬁer ﬁsh removal, and (b) an interval of one month, the length of the
experiments.

73

(a)

1.2 1

 

0.8 -
Riiiles
[:1 Runs

 

0.6 ~

0.4 ~

0.2 -

 

Immigration Rate (fishlday)

 

 

 

 

 

 

 

 

Site

0))
0.3 I

0.25 -

 

 

0.2 n
Riﬂ‘les

0.15 -
CIRuns

 

0.1 -

 

0.05 ~

immigration Rate (fishlday)

 

 

 

 

 

 

 

Site

Figure 8. Immigration rates of creek chubs colonizing each habitat unit class in streams
containing only rifﬂes and runs. Immigration was calculated for (a) the ﬁrst return
interval, 2 to 6 days aﬁer ﬁsh removal, and (b) an interval of one month, the length of the

experiments.

74

(a)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2.5 7
3‘.
3 2 .
E
e
E
3 1.5 -1
2
s 1 -
'3
5.
E 0.5 ~
§
0 1
12 14
Site
(b)
1 -
3‘, 0.9 -
S 0 3 -
a .
E 0.71
3 0.6 ~
.2 0.5 -
E 0.4 -
E 0.34
.9 -
s 3'?
0 i l I
12 14
Site

c] Runs
I Pools

1:) Runs
I Pools

Figure 9. Immigration rates of creek chubs colonizing each habitat unit class in streams
containing only pools and runs. Immigration was calculated for (a) the ﬁrst return
interval, 2 to 6 days after ﬁsh removal, and (b) an interval of one month, the length of the

experiments.

75

Discussion

The habitat preferences of creek chubs (Semotilus atromaculatus Mitchill) have
been documented by several investigators (Moshenko and Gee 1973, Copes 1978, Storck
and Momot 1981, Hubert and Rahel 1989, Harvey and Stewart 1991). My observations
agree with previous ﬁndings that adult creek chubs prefer deeper pool habitat to
shallower runs and riffles. This preference is likely due to deeper water providing cover
from terrestrial predators (Harvey and Stewart 1991).

All methods I used to quantify habitat suitability for creek chubs (Strauss’s linear
selection index, Manly’s standardized selection ratio, and immigration rate) yielded
similar insights. Close agreement of immigration trends with density-based habitat
preferences from earlier studies also were discovered by Belanger and Rodriguez (2002)
in their study of stream salmonids. The density of individuals in a habitat unit is the
result of movement into and out of the unit, thus agreement between the two approaches
could be expected. However, a focus on movement encourages observation of
individuals over time, an approach that can integrate seasonality and behavioral responses
to population density and community composition. Such an approach provides a much
more mechanistic understanding than habitat suitabilities based on point counts and
physical habitat attributes alone, which form the basis of most habitat suitability indices.
(HSIs).

The current HSI for creek chubs (McMahon 1982) lists the following variables as
important for the species: percent pool, pool class, percent cover, winter cover, gradient,

width, turbidity, pH, production, temperature, DO, ﬂow, substrate, shade, percent stream

76

bank vegetation, and depth. This list was based on literature review and author opinion.
A cumulative HSI score for a particular site can be calculated from the above variables
by following the procedure described in McMahon (1982). Hubert and Rahel (1989)
found that neither HSI scores nor the majority of component HSI variables correlated
with the biomass of creek chubs in their study sites. The availability of pools was the
only variable included in the H81 that correlated with creek chub biomass. They also
discovered that ﬂow variability, the amount of submerged vegetation, and average
velocity were signiﬁcant, yet not included in the index. They suggest that the failure of
the H81 may be due to non-additivity of the component variables and regional variation.
These weaknesses further support the concept of a habitat suitability assessment that
includes behavior and assemblage variables.

I found that creek chub habitat preference was site-speciﬁc and density-
independent. Fretwell and Lucas (1970), when describing the theory of ideal free
distribution, note that habitat suitability might decline with increasing density. They
suggest that the most suitable habitat will decline with increasing density until it is
equivalent to the next most suitable habitat. Animals will then move into both units
equally, until those units decline to the level of the third most suitable habitat, and so on.
The result is that with increasing density of individuals in an area, distribution among
habitat types should become more even. This was not the case with creek chubs in the
Red Cedar River watershed under the densities observed. Fraser and Sise (1980)
similarly found that evenness did not signiﬁcantly increase with density in adult creek

chubs inhabiting pools varying in the amount of cover in a second order stream.

77

The design of this study allowed observation of creek chub habitat selection over
a range of densities and across many stream sites, allowing separation of site and density
effects. These data revealed trends in ﬁsh habitat selection that could not be observed if
density was not manipulated, as is the case with most HSI data, or if a single site was
used. There is a clear preference for pools over runs (except in the deepest streams,
where no riffles existed), and for runs over rifﬂes, and this did not vary with density.
This is not to suggest that the tenets of the ideal free distribution do not hold, but only
that suitability, in this case, is density-independent within the range of densities observed.

The site-speciﬁc variation in habitat preference suggests that the relative
availability and characteristics of habitats at a particular site are the primary determinants
of preference. For example, in this study, creek chubs showed stronger preference for
runs than pools only in deeper streams where the average run was similar in depth to
pools in other, shallower streams, and rifﬂes were missing altogether. In these streams,
the deepest areas may harbor larger, piscivorous ﬁshes (including the occasional very
large creek chub) that use the depths as cover from terrestrial predators (Harvey and
Stewart 1991). The presence of predatory ﬁshes reduces the suitability of pools for most
creek chubs. Furthermore, time of day may inﬂuence the suitability of particular habitat
types. Fraser and Emmons (1984) found that blacknose dace (Rhinichthys atratulus)
would share pools with predatory ﬁsh during the day, but would move away from these
occupied pools at night, when darkness reduced the risk posed by terrestrial predators.

My study considered creek chub distribution only during daylight hours; however, it is

78

likely that habitat suitability for chubs may shift over a 24-hour period. Proportional
availability, and the habitat available adjacent to the study reaches may also play a role.

Why should we care about habitat selection? First, these observations provide
information about the basic biology of the species. From a management standpoint,
knowledge of how to determine the best habitat characteristics for a species, gleaned
from observations in the ﬁeld or laboratory, supports habitat conservation and restoration.
Further, understanding habitat selection under a variety of conditions will help us
anticipate responses to disturbance and a time frame for recovery, because distribution is
a product of behavior. For these reasons, continued research into how best to interpret

species movement and distribution to determine habitat suitability is warranted.

Acknowledgments

Invaluable assistance was provided in the ﬁeld and laboratory by Kelly
DeGrandchamp, Janice Karcher, Kevin Mann, Colleen McLean, Cassandra Meier, Aaron
Shultz and Alfred Thomson. This research was supported by Michigan State
University’s Watershed Action Through Education and Research initiative (MSU-
WATER), a University Distinguished Fellowship from Michigan State University and a
National Needs Water Science Fellowship from the United States Department of

Agriculture.

79

Literature Cited

Albanese, B. 2001. Movement of ﬁshes in a network of streams and implications for
persistence. Doctoral Dissertation. Virginia Polytechnic Institute and State
University, Blacksburg, Virginia.

Belanger, G., and M. A. Rodriguez. 2002. Local movement as a measure of habitat
quality in stream salmonids. Environmental Biology of Fishes 64: 1 55-164.

Copes, F. C. 1978. Ecology of the creek chub. Reports on the fauna and ﬂora of
Wisconsin, no. 12. The Museum of Natural History, Stevens Point, Wisconsin.

Dunne, T., and L. B. Leopold. 1978. Water in environmental planning. W.H. Freeman,
San Francisco, CA.

Fitzgerald, D. G., R. P. Lanno, and D. G. Dixon. 1999. A comparison of a sentinel
species evaluation using creek chub (Semotilus atromaculatus Mitchill) to a ﬁsh
community evaluation for the initial identiﬁcation of environmental stressors in
small streams. Ecotoxicology 8:33-48.

Fraser, D. F ., and T. E. Sise. 1980. Observations on stream minnows in a patchy
environment: a test of a theory of habitat distribution. Ecology 61:790-797.

Fraser, D. F., and E. E. Emmons. 1984. Behavioral response of juvenile blacknose dace
(Rhinichthys atratulus) to varying densities of predatory creek chub (Semotilus
atromaculatus). Canadian Journal of Fisheries and Aquatic Sciences 41 :364-370.

F retwell, S. D., and H. L. Lucas, Jr. 1970. On territorial behavior and other factors
inﬂuencing habitat distribution in birds. 1. Theoretical development. Acta
Biotheoretica 19:16-36.

Gerking, S. D. 1953. Evidence for the concepts of home range and territory in stream
ﬁshes. Ecology 34:347-365.

Harvey, B. C., and A. J. Stewart. 1991. Fish size and habitat-depth relationships in
headwater streams. Oecologia 87:336—342.

Hubert, W. A., and F. J. Rahel. 1989. Relations of physical habitat to abundance of four
nongame ﬁshes in high-plains streams: a test of habitat suitability index models.
North American Journal of Fisheries Management 9:332-340.

Manly, B. F. J ., L. L. McDonald, D. L. Thomas, T. L. McDonald, and W. P. Erickson.

2002. Resource selection by animals: statistical design and analysis for ﬁeld
studies, 2nd edition. Kluwer, Boston.

80

McMahon, T. E. 1982. Habitat suitability index models: creek chub. USDI Fish and
Wildlife Service.

Moshenko, R. W., and J. H. Gee. 1973. Diet, time and place of spawning, and
environments occupied by creek chub (Semotilus atromaculatus) in the Mink
River, Manitoba. Journal of the Fisheries Research Board of Canada 30:357-362.

Scott, W. B., and E. J. Crossman. 1973. Freshwater ﬁshes of Canada. Bulletin of the
Fisheries Research Board of Canada. 184.

Smithson, E. B., and C. E. Johnston. 1999. Movement patterns of stream ﬁshes in a
Ouachita highlands stream: an examination of the restricted movement paradigm.
Transactions of the American Fisheries Society 128:847-853.

Storck, T., and W. T. Momot. 1981. Movements of the creek chub in a small Ohio
stream. Ohio Journal of Science 81 :9-13.

Strange, E. M., P. B. Moyle, and T. C. Foin. 1993. Interactions between stochastic and

deterministic processes in stream ﬁsh community assembly. Environmental
Biology of Fishes 36:1-15.

Strauss, R. E. 1979. Reliability estimates for Ivlev's electivity index, the forage ratio, and
a proposed linear index of food selection. Transactions of the American Fisheries
Society 108:344-352.

81

S

uIiiijiIiiiiiiiiiiI