ECOTONE PROPERTIES AND INFLUENCES ON FISH DISTRIBUTIONS

ALONG HABITAT GRADIENTS OF COMPLEX AQUATIC SYSTEMS

A Dissertation

Presented to the Faculty of the Graduate School

of Cornell University

In Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

by

Nuanchan Singkran

May 2007

© 2007 Nuanchan Singkran

ECOTONE PROPERTIES AND INFLUENCES ON FISH DISTRIBUTIONS

ALONG HABITAT GRADIENTS OF COMPLEX AQUATIC SYSTEMS

Nuanchan Singkran, Ph. D.

Cornell University 2007

Ecotone properties (formation and function) were studied in complex aquatic

systems in New York State. Ecotone formations were detected on two embayment-

stream gradients associated with Lake Ontario during June–August 2002, using abrupt

changes in habitat variables and fish species compositions. The study was repeated at

a finer scale along the second gradient during June–August 2004. Abrupt changes in

the habitat variables (water depth, current velocity, substrates, and covers) and peak

species turnover rate showed strong congruence at the same location on one gradient.

The repeated study on the second gradient in the summer of 2004 confirmed the same

ecotone orientation as that detected in the summer of 2002 and revealed the ecotone

width covering the lentic-lotic transitions. The ecotone on the second gradient acted

as a hard barrier for most of the fish species. Ecotone properties were determined

along the Hudson River estuary gradient during 1974–2001 using the same methods

employed in the freshwater system. The Hudson ecotones showed both changes in

location and structural formation over time. Influences of tide, freshwater flow,

salinity, dissolved oxygen, and water temperature tended to govern ecotone properties.

One ecotone detected in the lower-middle gradient portion appeared to be the optimal

zone for fish assemblages, but the other ecotones acted as barriers for most fish

species.

A spatially explicit abundance exchange model (AEM) was developed to predict

distribution patterns of five fish species in relation to their population characteristics

and habitat preferences across the lentic-lotic ecotones on the two freshwater gradients

associated with Lake Ontario. Preference indexes of each target fish species for water

depth, water temperature, current velocity, cover types, and bottom substrates were

estimated from field observations, and these were used to compute fish habitat

preference (HP). Fish HP was a key variable in the AEM to quantify abundance

exchange of an associated fish species among habitats on each study gradient. The

AEM efficiently determined local distribution ranges of the fish species on one

gradient. Results from the model validation showed that the AEM was able to

quantify most of the fish species distributions on the second gradient.

iii

BIOGRAPHICAL SKETCH

Nuanchan Singkran was born in Chantaburi province located in the eastern part of

Thailand. After finishing high school, Nuanchan entered the faculty of Science at

Chulalongkorn University in 1989 for her undergraduate study in Marine Science with

a major in Marine Biology. Study at the Marine Science Department allowed her to

obtain broad knowledge of both applied science and ecology, especially the

relationship between aquatic creatures and environmental conditions along the coastal

zone of Thailand. After graduation, Nuanchan began her career as an environmental

reporter for Thai and English newspapers during 1993–1996. During her career as a

journalist, Nuanchan received a 2nd place award for an environmental article

supporting Eco-Tourism of Thailand in 1993, and the award for outstanding

environmental news reporter from the Environmental Siam Association in 1994.

Nuanchan’s work in the media provided her a unique opportunity to observe various

environmental problems in Thailand. She realized that the deterioration of natural

resources in the country comes primarily from public lack of understanding and

awareness about the environment. Lack of suitable methodologies to solve problems

and weak environmental policy are additionally significant factors that aggravate

environmental problems. Although Nuanchan likes the challenge of media work, she

needed to be engaged more actively in the problem-solving process. She believed that

academic and research work is really the area that would allow her to contribute to

public education and in-depth research on specific issues.

During 1996–1999, Nuanchan obtained scholarships from the Chin Sophonpanich

Foundation for Environment and from the Biodiversity Research Grant to further her

Master’s program in the multidisciplinary Department of Environmental Science,

Chulalongkorn University. Her Master’s thesis was entitled “Species Composition of

iv

Fish in Mangrove Canals as Reflected from Coastal Land Use at Trat Bay, Thailand”.

Her interest in this topic comes from the fact that a large number of fish species (over

2,000 across the country) have been seriously threatened by human activities,

especially the conversion of coastal areas into shrimp farms, urban districts, or

industrial zones, all done without proper management. Her Master’s research using

fish as a biological indicator revealed that improper land use activities altered the

coastal environment, which in turn affected fish and other living communities in

coastal systems in terms of both abundance and species diversity. After completing

her Master’s degree, Nuanchan received a government scholarship to pursue her PhD

program in Natural Resources at Cornell University in Ithaca, New York in 2001. At

Cornell, Nuanchan conducted her dissertation research on influences of ecotone

properties (formation and function) on species distributions over spatial and temporal

scales. Her research work dealt with both empirical and simulation modeling to

determine ecotone formation in complex aquatic systems and ecotone function on fish

species assemblages. Additionally, she used a simulation model to spatially and

temporally forecast distribution of fish species assemblages across aquatic boundaries.

v

For all fish and other creatures across global boundaries

vi

ACKNOWLEDGMENTS

My graduate committee played vital roles throughout the development of my

dissertation research. I am grateful to all of them for giving me the chance to freely

design and follow my research ideas. My committee Chair, Dr. Mark B. Bain, not

only supported me in any situation during my PhD program at Cornell University, but

also shared his philosophy of thinking and was patient in shaping my research work to

continue on the right track. Without his strong academic support and other help, my

graduate study at Cornell University could not have been completed. As one of my

committee, Dr. Patrick J. Sullivan always gave me instructive comments and helpful

suggestions that greatly improved the quality of my research work. His simple

questions, e.g., “Why is an ecotone is important?” and “How many functions can an

ecotone have?” caused me to think deeply, carefully, and quantitatively on my ecotone

study. Prof. Daniel P. Loucks is not just one of my committee, but also one of the best

teachers I have ever had during my lifetime as a student. Although Prof. Loucks said

that he did not know anything about fish and ecology, his contributions to my research

work deserve praise. Following his suggestion to take some core courses in his

department of Civil and Environmental Engineering, I gained more modeling

experience and recovered my mathematical background to achieve my dissertation

research.

The Royal Thai Government and Thai people whose tax money supported me to

study in the USA deserve any credit for my accomplishments. Additionally, my

special thanks are due to the Lake Ontario Biocomplexity Project (Natural Science

Foundation OCE–0083625), the Hudson River Foundation (GF/01/05), and the

College of Agriculture and Life Sciences at Cornell University, which partly

supported my study and research. I thank Prof. Paul A. Delcourt and Prof. Hazel R.

vii

Delcourt for suggesting ecotone analysis; John Young from ASA Analysis &

Communication, Inc. for providing the beach seine fish and habitat data from the

Utilities’ Hudson River estuary monitoring program; John Ladd from New York State

Department of Environmental Conservation for providing the substrate and shoreline

data; and the Cornell University’s Geospatial Information Repository for providing

submerged aquatic vegetation data.

The completion of my graduate program at Cornell University would be

impossible without the warm support and encouragement from friends and

collaborators. I am thankful to Kristi Arend, Takao Kumazawa, Kathy Mills, James

Murphy, Marci Meixler, Emelia Deimezis, Geof Eckerlin, Gail Steinhart, Carol

Steinhart, Geoff Steinhart, and all the other people I cannot name here for their

support and help during my study period at Cornell. I thank my parents and my

husband, Poonrit Kuakul, for their patience and understanding. I am thankful to my

senior friends, Dolina Millar, Bruce Anderson, Wandee Anderson, and some sisters

and brothers from the Taste of Thai restaurant who made my life in Ithaca not so

lonely and never refused to help me in any manner they could.

viii

TABLE OF CONTENTS

Biographical sketch iii

Dedication v

Acknowledgments vi

List of figures ix

List of tables xii

Chapter One Ecotone properties in the freshwater system associated 1 with Lake Ontario, New York.

Chapter Two Ecotone properties in the Hudson River estuary system 26 New York.

Chapter Three An abundance exchange model of fish assemblage 60 response to changing habitat along embayment-stream gradient of Lake Ontario, New York.

Appendix 2.1. Cumulative abundance and species richness of fish 93 by region (RE) along the Hudson River estuary gradient (RE1–RE12) from the five power plant utilities’ Hudson River estuary monitoring program using beach seine survey during 1974 to 2001.

Appendix 2.2. Multiple functions (aggregator, mediator, soft barrier, 105 and hard barrier) of the ecotones (from all frequencies of detection) on the Hudson fish species in each time period during 1974–2001.

Appendix 2.3. Multiple functions (neutral, partly inhibit, and 109 completely inhibit) of the ecotone peripheries (from all frequencies of detection) on the Hudson fish species in each time period during 1974–2001.

References 112

ix

LIST OF FIGURES

Figure 1.1. Ecotone structures and functions. 5

Figure 1.2. Sampling stations along the Sterling gradient during the 9

summer of 2002 (June–August). Observed value of each habitat variable was averaged over the three summer months and compared among stations using different sizes (small, medium, large) of bar charts to display mean magnitude (low, intermediate, or high).

Figure 1.3. Sampling stations along the Floodwood gradient during the 11

summer of 2002 (June–August). Observed value of each habitat variable was averaged over the three summer months and compared among stations using different sizes (small, medium, large) of bar charts to display mean magnitude (low, intermediate, or high).

Figure 1.4. Twelve stations (rectangles) were sampled along the 13

Floodwood gradient in the summer of 2004 (June–August) compared to the eight stations (circles) sampled in the summer of 2002. The ecotone periphery was indicated monthly by a peak absolute difference in the beta diversity (δ beta A) of fish species (June = connected line, July = dashed line, and August = gray line with squared marker) in each study summer.

Figure 2.1. The Hudson River estuary gradient showing 13 regions 31

designated for the five power plant utilities’ Hudson River estuary monitoring program.

Figure 2.2. Frequency of ecotones detected monthly based on the 39

data collection along the Hudson River estuary gradient during 1974–2001.

Figure 2.3. Ecotones (thick bars) and ecotone peripheries (thick lines) 40

detected during 1974–2001on the Hudson River estuary gradient covering 228 river kilometers (RKM) from region (RE) 1 to RE12 are grouped into each frequency class. Numbers of times of the formation of an ecotone and an ecotone periphery in each location are shown in brackets in each frequency class on the right side of the graph.

x

Figure 2.4. Monthly mean values of dissolved oxygen observed along 47 the Hudson River estuary gradient in the detected ecotone (line with solid dot markers) and non-detected ecotone (line with open dot markers) periods.

Figure 2.5. Monthly mean values of salinity observed along the 49

Hudson River estuary gradient in the detected ecotone (line with solid dot markers) and non-detected ecotone (line with open dot markers) periods.

Figure 2.6. Monthly mean values of water temperature observed 50

along the Hudson River estuary gradient in the detected ecotone (line with solid dot markers) and non-detected ecotone (line with open dot markers) periods.

Figure 2.7. Values of dissolved oxygen (DO), salinity, and water 52

temperature were averaged by region over the 116 time periods that the ecotones were detected. Bed areas of two submerged aquatic vegetation (SAV) species (Trapa natans and Vallisneria american) vary along the gradient (data for region 1 were unavailable). Proportions of 10 substrate types were observed in the tidal river estuary.

Figure 3.1. The Floodwood (FL) and Sterling (ST) gradients with eight 65

sampling stations (stars, FL; circles, ST) for monthly collecting of target fish and habitat variables during June–August 2002. The fish and habitat collections were repeated at 12 stations (circles) along the FL gradient during June–August 2004.

Figure 3.2. A conceptual diagram of fish distribution within and among 70

habitats. Migration rates of fish at two life stages in the growing season (yk, 0+ yr old; ak, 1+ yr old) and for all fish in winter (k) among habitats n, n–1, and n+1 are dependent on fish habitat preference in the growing season (yHP, 0+ yr old; aHP, 1+ yr old) and in winter (HP), respectively.

Figure 3.3. Categorized mean values of all habitat variables (except 77

substrate composition) observed monthly during June–August 2002 along the Floodwood gradient.

Figure 3.4. Categorized mean values of all habitat variables (except 79

substrate composition) observed monthly during June–August 2002 along the Sterling gradient.

xi

Figure 3.5. Seasonal distribution pattern of yellow perch over the 81 100-year simulation (A). The fish population size exponentially decreased or increased for the first 10-year period of the simulation when changes of 50% of carrying capacity Cy (B), birth fraction rb (C), or death fraction rd (D) were assigned to the abundance exchange model, but the population size returned to the seasonal pattern over the long-term prediction.

Figure 3.6. Observed abundances of yellow perch (YP), bluegill (BG), 83

bluntnose minnow (BM), logperch (LP), and fantail darter (FTD) along the Floodwood gradient in August 2004 were compared with the mean predicted abundance ± standard deviation of the prediction of the associated fish species in August over the 100-year simulation.

Figure 3.7. Mean predicted abundance of yellow perch (YP), bluegill 85

(BG), bluntnose minnow (BM), logperch (LP), and fantail darter (FTD) among months and stations on the Floodwood gradient over the 100-year simulation.

Figure 3.8. Observed abundance of yellow perch (YP), bluegill (BG), 86

bluntnose minnow (BM), and fantail darter (FTD) along the Sterling gradient in August 2002 were compared with the mean predicted abundance ± standard deviation of the prediction of the associated fish species in August over the 100-year simulation.

Figure 3.9. Mean predicted abundance of yellow perch (YP), bluegill (BG), 88

bluntnose minnow (BM), and fantail darter (FTD) among months and stations on the Sterling gradient over the 100-year simulation.

xii

LIST OF TABLES

Table 1.1. Intensity of each habitat variable was average the three 17

summer months (June–August) of 2002 along the Sterling and Floodwood gradients.

Table 1.2. Relative abundance comparison of each fish species 20

among the downstream (DW, FL1–FL3), ecotonal (ECO, FL4–FL5), and upstream (UP, FL6–FL12) habitats on the Floodwood gradient over the summer months of 2004 (June–August).

Table 2.1. Functions of the ecotones detected ≥5 times at certain 41

regions (RE) and time periods on Hudson fish species assemblages.

Table 2.2. Functions of the ecotone peripheries detected ≥5 times 43

at certain regions (RE) and time periods on Hudson fish species assemblages.

Table 2.3. Results from multivariate analysis of variance for each 46

habitat variable and its delta (absolute difference between minimum and maximum values) between non-detected (ECO = 0) and detected (ECO = 1) ecotone periods and due to the interactions of ECO × RE (river region), ECO × Month, ECO × Year, ECO × RE × Month, ECO × RE × Year, and ECO × Month × Year.

Table 3.1. Initial values of the population variables (Cy, rb, and rd) 73

and individuals of each target fish species for simulating the abundance exchange model.

Table 3.2. Preference index (Pi) for each habitat variable of five 75

modeled fish species at two life stages (0+yr and 1+yr old) estimated from the fish and habitat data observed along the Floodwood gradient over the three summer months (June–August) of 2004.

1

CHAPTER ONE

Ecotone properties in the freshwater system associated with Lake Ontario, New York

ABSTRACT

Ecotone formations were detected on two embayment-stream gradients associated

with Lake Ontario using abrupt changes in habitat variables and peak turnover rates in

fish species compositions along the gradients. The study was conducted on both

gradients over the three summer months (June–August) of 2002, and it was repeated at

a finer scale along the second gradient over the same summer months of 2004. The

results revealed that the ecotones were static in their orientations on both gradients

during the study periods. In the summer of 2002, abrupt changes in the habitat

variables (water depth, current velocity, substrates, and covers) and peak species

turnover rate showed strong congruence at the same location on one of the two

gradients. On the second gradient, only the peak species turnover rate could clearly

define the ecotone periphery, whereas the abrupt changes in the habitat variables

occurred at multiple locations. The repeated study at the finer scale on the second

gradient in the summer of 2004 confirmed the same ecotone orientation as that

detected in the summer of 2002 and revealed the ecotone width covering the lentic-

lotic transition. Four ecotone functions (aggregator, mediator, soft barrier, and hard

barrier) were inferred from the comparison of relative abundance of fish species

among downstream, ecotonal, and upstream habitats. In general, the ecotone acted as

a hard barrier for most of the fish species. The distinct physical discontinuities below

and above the ecotonal habitat appear to be an important factor shaping the ecotone

geometry and impeding fish distribution into the ecotone.

2

INTRODUCTION

The ecotone concept is almost as old as the field of ecology (Zalewski et al. 2001)

although ecologists have used the concept loosely on a broad range of boundary

conditions. This practice has impeded the development of theory and the recognition

of patterns that could advance the understanding and use of the ecotone idea (Strayer

et al. 2003). Nevertheless, ecotone studies in ecosystems have remained prominent

and appealing (e.g., Clements 1897, Leopold 1933, Odum 1971, Risser 1990, Johnson

et al. 1992, Winemiller and Leslie 1992, Lidicker 1999, Fortin et al. 2000, Willis and

Magnuson 2000, Harding 2002, Cadenasso et al. 2003, Martino and Able 2003, Miller

and Sadro 2003, Ries et al. 2004) because of the important roles of ecotones on

biodiversity (di Castri and Hansen 1992, Neilson et al. 1992, Lachavanne 1997), biotic

interactions (Wiens et al. 1993), flows of materials and nutrients in ecosystems

(Hansen and di Castri 1992), and the ecological implications for natural resources

management at the individual, population, and ecosystem levels (Sisk and Haddad

2002).

Clements (1897) was the first who viewed an ecotone as the transition zone

between different plant communities. He observed that the vegetation in this zone was

often accentuated in size and density. Leopold (1933) later explained that shape and

spatial configuration of an ecotone influence species richness and organism

abundance. Successive ecotone studies (e.g., Lay 1938, Johnston and Odum 1956)

worked with these early ideas. Odum (1971) defined an ecotone as a transition zone

between two or more communities, and he labeled the increases in both species

richness and abundance at the ecotone as a consequence of the “edge effect”. At

present, an ecotone is widely viewed as a transition zone between adjacent ecological

systems with influences on organism distributions going beyond the traditional “edge

3

effect” (e.g., Holland 1988, Naiman and Décamps 1990, Fortin et al. 2000, Fagan et

al. 2003, Strayer et al. 2003).

From over 900 empirical ecotone studies in terrestrial ecosystems (Ries et al.

2004), ecotones provide diverse influences on abundance distributions of organisms in

and around the ecotones. For instance, results from bird communities (Sisk and

Margules 1993, Baker et al. 2002, Watson et al. 2004) showed that the expected

increase in density and species richness in ecotones was not easily detected because

bird species responded differently, i.e., edge avoiders, edge exploiters, and habitat

generalists. Although distinct changes in bird assemblages were observed at an

ecotone, conditions in the ecotone appeared to be neutral to some species. Birds have

been widely used as ecotone response variables in terrestrial ecosystems (Ries et al.

2004) because of their mobility, diversity, and diverse habitat specializations. Fish are

also mobile organisms that often display high diversities in species richness and life

histories and, thus, appear to be an excellent ecotone indicator (Kirchhofer 1995,

Schiemer et al. 1995) in aquatic ecosystems. However, fewer ecotone investigations

have been conducted in aquatic ecosystems (e.g., Przybylski et al. 1991, Willis and

Magnuson 2000, Martino and Able 2003).

Whereas locations of most terrestrial ecotones can be visually identified (e.g., an

ecotone between grassland and forest) before hypothesis testing, it is more difficult to

determine locations of aquatic ecotones. In aquatic ecosystems, environmental

conditions are complex in space and time due to variations in hydrology (Wiens 2002)

and climate-linked seasonal changes that can alter other aquatic habitat variables (e.g.,

water temperature, growth of aquatic plants and algae). Consequently, ecotone

properties (formation and function) in water systems are highly related to system

variability, and they reflect the dynamics of organism assemblages responding to the

environmental changes over space and time.

4

The objectives of this study were thus to detect the ecotone formation on two

embayment-stream gradients and determine the ecotone function on fish species

assemblages. An ecotone is viewed as a dynamic structure in an ecosystem that

constitutes a unique habitat and possibly a unique community (Fagan et al. 2003). In

this study, it is defined as a zone of transition in the physicochemical properties and

interactions among habitats and species. An ecotone may form, decline, disappear,

shift location, or change in structure or function due to the influence of changing

physicochemical conditions, species use of habitat, and interactions among species

(Holland 1988, Delcourt and Delcourt 1992, Wiens 1992, Kolasa and Zalewski 1995).

Relevant hypotheses and aquatic ecotone studies

The status of an ecotone may differ depending on the spatial scale considered and

the prior delineation of the system of interest (Kolasa and Zalewski1995).

Consequently, scientists across fields of study may view ecotone structure differently

under different terms, such as an edge, a transition zone, or a boundary (e.g., Strayer et

al. 2003). Hence, it is necessary to clearly specify the spatial structure of an ecotone

(Kolasa and Zalewski 1995, Cadenasso et al. 2003) in a study. Ecotone structure may

be defined according to dimensionality, grain size, sharpness, or contrast. Based on

dimensionality (Strayer et al. 2003), an ecotone may be considered either as a thin line

between adjacent habitats or as a zone with the same dimension as its adjacent habitats

(Figure 1.1a). An ecotone may be defined according to different grain sizes between

adjacent habitats (Figure 1.1b) or sharpness of formation (Figure 1.1c) depending on

changes in the ecological process across the ecotone (Cadenasso et al. 1997, Bowersox

and Brown 2001, Cadenasso et al. 2003, Fagan et al. 2003, Strayer et al. 2003).

Lastly, an ecotone may form between adjacent habitats with high or low habitat

contrast (Figure 1.1d; Cadenasso et al. 2003, Strayer et al. 2003).

5

Figure 1.1. Ecotone structures and functions. Ecotone structure may be defined based on dimensionality (a.1: a line, a.2: a broader zone), grain size (b.1: a fine grain size, b.2: a coarse grain size), sharpness (c.1: a sharp formation, c.2: a gradual formation), or contrast (d.1: high contrast, d.2: low contrast). As in a.2, an ecotone may function as an aggregator attracting some species into the ecotone, a hard barrier completely blocking some species into the ecotone, a soft barrier partially allowing some species to enter the ecotone, or a mediator having no effect on flux of certain species. (Figures 1a–1d are reproduced from Strayer et al. 2003, A classification of ecological boundaries, Copyright, American Institute of Biological Sciences.)

6

In this study, ecotone structure was considered as a zone having the same

dimension as its adjacent habitats, and its formation was detected on each study

gradient using both habitat changes and fish species turnover rate as indicators. These

are based on the hypothesis that an aquatic ecotone can be determined using abrupt

changes in either habitat conditions or fish species composition. An ecotone generally

provides diverse permeability (Wiens 1992, Ross et al. 2005) on flux of organisms.

This role is particularly significant for mobile animals, as they can migrate to their

optimal habitats in and around an ecotone. According to various permeability of an

ecotone on species distributions, an ecotone may simultaneously facilitate, inhibit, or

have no effects on distributions of multiple species into the ecotone (Wiens et al.

1985, Wiens 1992, Kolasa and Zalewski 1995, Haddad 1999, Lidicker 1999, Ries and

Debinski 2001, Matthysen 2002, Strayer et al. 2003).

Four hypothetical functions of aquatic ecotones (Figure 1.1) on fish species

distributions along the study gradients were proposed as follows. An ecotone may act

as an aggregator for particular species by attracting them into the ecotone.

Alternately, it may act as a hard barrier by completely blocking distributions of some

other species into the ecotone, or as a soft barrier by partially allowing some species to

enter the ecotone. In the latter case, most species populations entering the ecotone

will return to the habitats from which they came, causing few of them to be observed

in the ecotone. Lastly, an ecotone may act as a mediator, having no effect on the flux

of particular species between the ecotone and the adjacent habitat(s).

Some previous studies reported about the influences of aquatic ecotones in

particular ecosystem settings and the way changing characteristics of habitat and

species interact. For instance, the transitions between downstream stations and Lake

Balaton (Hungary) acted as the zones aggregating the highest fish density and species

richness (Przybylski et al. 1991). Similarly, Willis and Magnuson (2000) studied

7

patterns in fish species composition across the interface between lakes and streams in

Wisconsin and concluded that both species richness and density of fish species

composition were the highest at the lake-stream ecotones. Their study suggested that

differences in the hydrologic and geomorphic properties of habitat strongly influence

the patterns of change in fish communities.

Features of aquatic habitats, such as water depth, current velocity, substrates, and

covers, have been reported as important factors influencing fish distribution and

species richness (Zalewski et al. 2001). Rakocinski (1992) revealed that water depth,

emergent stem density, salinity, water temperature, and dissolved oxygen influenced

both temporal and spatial changes in the community structure of marsh-edge fishes

across the marsh-open water ecotone in a Louisiana estuary. Environmental variables

related to habitat size and salinity showed the greatest correspondence with the

changes in the structures of fish assemblages across a marine-freshwater ecotone in

Tortuguero National Park on the Caribbean coast of Central America (Winemiller and

Leslie 1992). In addition, the variability in species richness and abundance of fish

assemblages in and around an ocean-estuarine ecotone on the river-bay-ocean gradient

in southern New Jersey were influenced by habitat heterogeneity, water depth, and the

fish species responses to the environmental gradient (Martino and Able, 2003). Miller

and Sadro (2003) reported an alternative basis of an estuary-stream ecotone in Oregon

as a habitat that aggregated juvenile anadromous fish during their physiological

transition.

8

MATERIALS AND METHODS

Study gradients and field observations

Field observations were conducted along two embayment-tributary stream

gradients, Sterling (ST) and Floodwood (FL), located on the southeastern coast of

Lake Ontario, New York. The ST gradient includes Sterling Pond and Sterling Creek.

The total watershed area associated with the ST gradient was 210.76 km2, while the

surface area of Sterling Pond alone was 0.38 km2. Sterling Creek ranges in width

from 3 m (upstream) to more than 20 m (entering Sterling Pond) with a mean annual

discharge of 1.9 m3/s (37-year record—USGS gauge 04232100 above the study area).

The FL gradient comprises Floodwood Pond and Sandy Creek. Total watershed area

for the FL gradient was 671.82 km2, while the surface area of Floodwood Pond alone

was 0.078 km2. Sandy Creek ranges in width from 6 m (upstream) to more than 50 m

(entering Floodwood Pond) with a mean annual discharge of 7.8 m3/s (37-year

record—USGS gauge 04250750 above the study area).

A Garmin Etrex Venture global positioning system was used to mark the locations

of sampling stations on both gradients. In the summer months (June–August) of 2002,

eight transect stations on the ST gradient were selected (between 43°20'34.5" N,

76°41'54.7" W and 43°16'51.6" N, 76°37'37.6" W) using a stratified random sampling

method to cover all habitat types. The sampling stations (ST with number) were near

the connection between Lake Ontario and Sterling Pond (ST1), the connection

between Sterling Pond and the mouth of Sterling Creek (ST2), the downstream (ST3

and ST4), the midstream (ST5 and ST6), and the upstream (ST7 and ST8) of Sterling

Creek (Figure 1.2). The length of the entire sampling gradient was about 16.3 km.

9

Figure 1.2. Sampling stations along the Sterling gradient during the summer of 2002 (June–August). Observed value of each habitat variable was averaged over the three summer months and compared among stations using different sizes (small, medium, large) of bar charts to display mean magnitude (low, intermediate, or high). Four substrate types were mud (Md), sand (Sd), gravel-cobble (Gv-Cb), and rock-bedrock (Rb). The three graphs below show the turnover rate in fish species composition (Beta A—connected line) and the peak absolute difference in the beta A, (δ beta A—dashed line) along the gradient in the associated months of observation.

10

The same stratified random method was used to select eight transect stations

(between 43°43'16" N, 76°12'10" W and 43°48'47" N, 76°4'31" W) on the FL

gradient. These sampling stations were at the connection between Lake Ontario and

Floodwood Pond (FL1), the connection between Floodwood Pond and the mouth of

Sandy Creek (FL2), the downstream (FL3, FL4, and FL5), the midstream (FL6), and

the upstream (FL7, FL8) of Sandy Creek (Figure 1.3). The length of the entire

sampling gradient was about 18.6 km.

Both abundance and species richness of fish assemblages were observed once a

month in the summer of 2002 (June–August) along the FL and ST gradients. Fish

were sampled with electrofishing gear at all FL and ST stations. At the sampling

stations where water depth was ≥1.0 m (ST1–ST6, FL1–FL4), the power supply gears

(a 3500-W generator and a transformer—the DC voltage output controller of 250–350

V) and the electrode array were installed on a 4.9-m shallow draft, flat bottom boat.

The boat hull served as the cathode array, and two 1-m-diameter rings with dropper

anodes were suspended in the water off the front of the boat. A 20-m-diameter

circular area was repeatedly electrofished for 15 minutes per station. While catching

fish, the boat speed was controlled, as closely as possible, at the average walking

speed of people, 4.85 km/hr (Fruin 1971), and this speed is assumed to be similar to

the walking pace of the fieldwork crew in shallow water stations (ST7–ST8 and FL5–

FL8, <1.0 m deep).

Because the boat could not access the shallow water stations, the generator and

transformer for generating electricity with the same DC voltage output as used on the

boat were instead installed on shore. One end of a cathode arraying with 4-m steel

cables was connected to the transformer and its other end was centrally positioned in a

sampling station and occasionally repositioned. Using this configuration, the power

supply gears generated a strong electric current across the sampling station.

11

Figure 1.3. Sampling stations along the Floodwood gradient during the summer of 2002 (June–August). Observed value of each habitat variable was averaged over the three summer months and compared among stations using different sizes (small, medium, large) of bar charts to display mean magnitude (low, intermediate, or high). Four substrate types were mud (Md), sand (Sd), gravel-cobble (Gv-Cb), and rock-bedrock (Rb). The three graphs below show the turnover rate in fish species composition (Beta A—connected line) and the peak absolute difference in the beta A, (δ beta A—dashed line) along the gradient in the associated months of observation.

12

For catching fish, one end of a 100-m electricity cable was connected to the

transformer. The other end of the cable was connected to a handheld steel ring (0.5-m

diameter) anode, and it was maneuvered through the sampling station for manually

shocking fish. A 60-m-long channel segment was sampled from downstream to

upstream by repeating passes for 15 minutes in each station. Juveniles and adults of

each fish species caught at each station were enumerated and their total length was

measured to the nearest millimeter.

In the same summer months of 2002, the habitat variables were measured once a

month in all stations where the fish species were collected, as follows. Water depth

(m) and water temperature (°C) were randomly measured five times in each sampling

station. Current velocity (m/s) was randomly measured five times within each station

using a Marsh-McBirney 2000-model (Bovee 1982). Bottom substrates (%) were

visually observed once a month in all sampling fish stations on each study gradient; at

the first observation, substrate grain sizes (diameter) were measured and classified into

four types modified from Pusey et al. (1993): mud (<1 mm diameter), sand (1–16 mm

diameter), gravel-cobble (16–128 mm diameter), and rock-bedrock (>128 mm

diameter). The average percentage of each substrate type and average abundance of

each cover type was calculated from their 10 random observations in each station.

The abundance of aquatic plants, algae, and woody debris was coded on a scale of 0–

3: 0 = absent, 1 = low, 2 = common, and 3 = high.

In the three summer months (June–August) of 2004, the fish collections were

repeated at 12 sampling stations (FL1–FL12; Figure 1.4: squares) along the FL

gradient. The 12 sampling stations were located between 43°43'20.6" N, 76°12'13.3"

W and 43°51'11.7" N, 75°56'44" W with a total sampling gradient length of 28.2 km.

Within each of the 12 sampling stations, three to six locations were randomly chosen

for monthly collecting abundance and species richness of fish on a finer scale. In

13

other words, the fish assemblages were collected monthly at a total of 60 locations

within the 12 stations along the FL gradient. With the finer scale of sampling,

juveniles and adults of fish assemblages were collected for 10 minutes at each of the

60 locations using the same fishing gear and sampling protocols as used in the summer

of 2002. Individuals of each fish species collected from all locations in each station

were accumulated, enumerated, and measured for their total length to the nearest

millimeter.

Figure 1.4. Twelve stations (rectangles) were sampled along the Floodwood gradient in the summer of 2004 (June–August) compared to the eight stations (circles) sampled in the summer of 2002. The ecotone periphery was indicated monthly by a peak absolute difference in the beta diversity (δ beta A) of fish species (June = connected line, July = dashed line, and August = gray line with squared marker) in each study summer. The ecotone comprises the lower ecotone periphery (e1), the ecotonal (ECO) habitat, and the upper ecotone periphery (e2). According to the ecotone orientation detected in the summer of 2004, the major downstream (DW) habitat was from stations 1 to 3, and the major upstream (UP) habitat was from stations 4 to 5.

14

Ecotone analysis

Turnover rate or beta diversity (Whittaker 1975, Gauch 1982, Delcourt and

Delcourt 1992, Winemiller and Leslie 1992, Wagner 1999, Hoeinghaus et al. 2004) in

species composition of fish was used to detect ecotone formation on each study

gradient in the summers of 2002 and 2004. Detrended correspondence analysis

(DCA) in CANOCO 4.0 software (ter Braak and Smilauer 1998) was used to compute

the beta diversity (hereafter called beta A) based on both occurrence and abundance of

fish. In the computation, the sample scores on the DCA ordination axes are rescaled

into the average standard deviation (SD) units of species turnover (Hill 1979). A

gradient length of 1.0 SD represents approximately one-fourth of a complete turnover

rate in species composition on the gradient (Gauch 1982). Species appear, rise to their

modes, and disappear over a distance of about 4 SD. That is, if the beta A on the

DCA’s axis 1 (that accounts for the most variation) is ≥4.0 SD, it indicates that species

compositions at both extreme ends of a gradient represent completely different

communities (Gauch 1982). This indicates complete formation of at least one ecotone

on the gradient. The peak absolute difference in the beta A (δ beta A) indicates an

ecotone periphery (a margin where a sharp transition between adjacent habitats

occurs).

Abrupt changes in the study habitat variables, the other ecotone indicator, were

compared among stations, months, and the station-month interactions along each

gradient in the summer of 2002 using a repeated measures 2-way ANOVA in SPSS

version 13 software. The habitat variables with significant differences among stations

(p≤0.05) were further analyzed using Tukey’s honestly significant differences (HSD)

test. The statistical results showing patterns of variations in the habitat variables along

each gradient in the summer of 2002 were summarized graphically in low, medium,

15

and high intensity levels to allow visual synthesis of habitat and fish assemblage shifts

along the gradients (Figures 1.2 and 1.3).

With the finer scale of the field observations along the FL gradient in the summer

of 2004, the abundance by species of the fish assemblages observed in and around the

FL ecotone was used to interpret ecotone function. In the process, each fish species

with ≥5 individuals observed on the FL gradient over the three summer months (June–

August) was selected to compare its percentage abundance among three major

habitats: ecotonal (ECO—area within the ecotone peripheries), downstream (DW),

and upstream (UP) habitats. The four hypothetical ecotone functions (aggregator,

mediator, soft barrier, and hard barrier) were then interpreted from relative abundances

of fish among the three major habitats. This study proposed that an ecotone acts as an

aggregator for a fish species with >60% abundance observed in the ECO habitat and

as a mediator for fish species with 30–60% abundance observed in the ECO habitat.

Concurrently, an ecotone acts as a soft barrier for a fish species with 5–30%

abundance observed in the ECO habitat. Lastly, an ecotone acts as a hard barrier for a

fish species with <5% abundance observed in the transitional habitat.

RESULTS

Ecotone formation

In the summer of 2002, 2,071 individual fish from 39 species were collected along

the ST gradient. Both fish abundance and species richness were used to estimate the

fish species turnover rate (beta A) along the gradient. Results from the DCA analysis

showed that the beta A was 2.1 SD, 12.6 SD, and 12.8 SD in June, July, and August,

respectively (Figure 1.2). The beta A values >4.0 SD in July and August indicated the

complete changeover in fish species composition on the gradient in these months. In

16

June, no change (the beta A = 0) in species composition of fishes was detected from

ST1 to ST6 (Figure 1.2), but a clear increase of the beta A occurred between ST6 and

ST7. As a result, a peak of an absolute difference in the beta A (δ beta A) was

detected between ST6 and ST7 in all months indicating this location as a persistent

zone of transition in fish species composition on the ST gradient.

Significant habitat variations along the ST gradient were detected among stations,

months, and station-month interactions (the repeated measures 2-way ANOVA,

p≤0.05). Significant differences among stations were found for all habitat variables

except possibly current velocity (p=0.065). Habitat differences across the three

sampling months were minor and significant only for current velocity, algae, and

aquatic plants. These three habitat variables were the only ones showing significant

station-month interactions, which were expected because of plant growth and

declining flows during the study period. The intensity values of all habitat variables

were averaged over the three summer months (Table 1.1), and they were graphically

compared among the ST stations (Figure 1.2) based on the results from Tukey’s HSD

post hoc. Consistent with the peak transition in fish species composition, abrupt

changes of all habitat variables except current velocity were also detected between

ST6 and ST7, reflecting this location as a zone of transition in habitat conditions as

well (Figure 1.2).

In the 2002 summer, 1,988 individual fish from 42 species were collected along

the FL gradient, from which the turnover rate in fish species composition on this

gradient was estimated. Results from the DCA analysis revealed the complete fish

turnover rate on the FL gradient over the study period according to the beta A value

(Figure 1.3) which exceeded 4.0 SD in June (6.3 SD), July (4.8 SD), and August (5.4

SD). The peak δ beta A occurred between FL5 and FL6 over the three summer

months marking this location as a distinct zone of transition in fish species

17

composition on the FL gradient. Significant differences among stations were found

for all habitat variables on this gradient (the repeated measures 2-way ANOVA,

p≤0.05). Additionally, changing intensities of all habitat variables except substrate

composition were significantly different among the sampling months.

Table 1.1. Intensity of each habitat variable was averaged over the three summer months (June–August) of 2002 along the Sterling and Floodwood gradients.

1 2 3 4 5 6 7 8

Sterling gradientWater depth (m) 1.71 2.06 2.72 2.92 2.39 1.66 0.33 0.48Current velocity (m/sec) 0.11 0.12 0.11 0.11 0.19 0.18 0.22 0.20Substrate (%) mud 90 90 100 100 90 75 2 0 sand 10 6 0 0 5 10 47 25 gravel-cobble 0 4 0 0 3 8 45 50 rock-bedrock 0 0 0 0 2 7 6 25Cover1

algae 1.17 2.30 2.43 1.97 1.87 2.10 1.07 1.00 aquatic plants 1.97 2.27 1.90 2.43 2.03 2.10 0.30 0.37 woody debris 0.00 0.23 0.00 0.00 0.60 0.73 1.73 1.70

Floodwood gradientWater depth (m) 2.69 1.94 2.20 1.50 0.54 0.17 0.46 0.44Current velocity (m/sec) 0.19 0.17 0.15 0.15 0.28 0.34 0.25 0.27Substrate (%) mud 90 100 100 12 5 0 0 0 sand 10 0 0 48 14 5 5 5 gravel-cobble 0 0 0 40 40 35 45 46 rock-bedrock 0 0 0 0 41 60 50 49Cover1

algae 0.80 1.47 0.10 0.10 1.27 1.67 1.37 0.97 aquatic plants 1.43 2.23 0.93 0.73 0.93 0.20 0.33 0.53 woody debris 0.33 0.73 1.27 0.67 0.20 0.13 0.80 0.60

StationHabitat variable

1Occurrence and abundance of each cover type was coded: 0 = absent, 1 = low, 2 = common, and 3 = high.

18

Similar to the patterns observed on the ST gradient, current velocity, algae, and

aquatic plants were significantly different by station-month interactions on the FL

gradient. Again, these were expected due to plant growth and declining flows during

the study period. The intensity values of all habitat variables were averaged over the

three summer months of 2002 (Table 1.1), and they were graphically compared among

the FL stations (Figure 1.3) based on the results from Tukey’s HSD post hoc. While

the peak fish turnover rate was detected between FL5 and FL6 over the study period,

abrupt changes in the habitat variables mostly occurred at the multiple stations (Figure

1.3). Variability and inconsistency of habitat changes along the FL gradient made it

difficult to use the habitat variables to clearly define ecotone location on this gradient.

Abrupt changes in both habitat variables and fish species composition between

ST6 and ST7 on the ST gradient indicated the ECO habitat between the two stations.

With respect to the peak turnover rate in fish species composition that was consistently

detected between FL5 and FL6 over the three summer months, the ECO habitat on the

FL gradient was most likely to be situated between these two stations. The field

observations were repeated at a finer scale along the FL gradient in the summer of

2004 to: 1) test if the ecotone would be detected at the same location on the gradient

with the complex mix of habitat changes during the summer period and 2) determine

the ecotone width and ecotone functions.

Over the three summer months (June–August) of 2004, 11,318 individual fish

from 43 species were collected from 12 stations along the FL gradient (Figure 1.4:

squares). The complete changeover in fish species composition occurred over the

summer of 2004 according to the beta A values of 5.0, 4.4, and 4.2 SD in June, July,

and August respectively. With the finer scale of the study, two peaks of the δ beta A

were detected. One was between FL5 and FL6 over the three months, while the other

was between FL3 and FL4 in July and August (Figure 1.4: Summer 2004). Results

19

from both summer studies revealed that the ECO habitat on the FL gradient was

situated within the two ecotone peripheries (i.e., FL3–FL4 and FL5–FL6 peripheries)

and had the width from FL4 to FL5 (Figure 1.4). In relation to the ecotone orientation

(the ECO habitat plus its ecotone peripheries), a DW habitat was from FL1 to FL3 and

an UP habitat was from FL6 to FL12 (Figure 1.4).

Ecotone function

Relative abundances of fish species were compared among the DW, ECO, and UP

habitats to interpret ecotone function on fish species distribution along the FL gradient

in the summer of 2004. Of the total 43 fish species, 31 species with ≥5 individuals

collected along the gradient over the three summer months were selected for

comparison (Table 1.2). The FL ecotone acted as an aggregator for bluntnose minnow

(Pimephales notatus), brook stickleback (Culaea inconstans), largemouth bass

(Micropterus salmoides), logperch (Percina caprodes), smallmouth bass (Micropterus

dolomieui), tessellated darter (Etheostoma olmstedi), and white sucker (Catostomus

commersoni). Each of these species was observed >60% in the ECO habitat. The

ecotone acted as a mediator for common shiner (Notropis cornutus), fathead minnow

(Pimephales promelas), pumpkinseed (Lepomis gibbosus), and rock bass (Ambloplites

rupestris). These species were observed 30–60% in the ECO habitat. The ecotone

acted as a soft barrier for bluegill (Lepomis macrochirus), creek chub (Semotilus

atromaculatus), cutlips minnow (Exoglossum maxillingua), golden shiner

(Notemigonus crysoleucas), longnose dace (Rhinichthys cataractae), and yellow perch

(Perca flavescens). These fish species showed 5–30% abundance in the ECO habitat.

The ecotone acted as a hard barrier for the remaining 14 fish species (Table 1.2), each

of these species showed <5% abundance in the ECO habitat.

20

Table 1.2. Relative abundance comparison of each fish species among the downstream (DW, FL1–FL3), ecotonal (ECO, FL4–FL5), and upstream (UP, FL6–FL12) habitats on the Floodwood gradient over the summer months of 2004 (June– August).

DW ECO %UP

Alewife (Alosa pseudoharengus ) 100 0 0 7Black crappie (Pomoxis nigromaculatus ) 100 0 0 13Blacknose dace (Rhinichthys atratulus ) 0 2 98 885Blacknose shiner (Notropis heterolepis ) 100 0 0 7Bluegill (Lepomis macrochirus ) 78 22 0 521Bluntnose minnow (Pimephales notatus ) 6 69 25 466Bowfin (Amia calva ) 95 0 5 10Brook stickleback (Culaea inconstans ) 0 96 4 15Brown bullhead (Ictalurus nebulosus ) 89 0 11 74Comely shiner (Notropis amoenus ) 100 0 0 13Common carp (Cyprinus carpio ) 100 0 0 25Common shiner (Notropis cornutus ) 6 53 41 315Creek chub (Semotilus atromaculatus ) 0 11 89 473Cutlips minnow (Exoglossum maxillingua ) 0 20 80 1,736Fantail darter (Etheostoma flabellare ) 0 2 98 826Fathead minnow (Pimephales promelas ) 0 39 61 45Golden shiner (Notemigonus crysoleucas ) 83 17 0 41Largemouth bass (Micropterus salmoides ) 26 74 0 334Logperch (Percina caprodes ) 9 91 0 74Longnose dace (Rhinichthys cataractae ) 0 9 91 2,000Northern pike (Esox lucius ) 100 0 0 10Pumpkinseed (Lepomis gibbosus ) 65 35 0 531Redside dace (Clinostomus elongatus ) 0 0 100 19Rock bass (Ambloplites rupestris ) 12 58 30 313Sand shiner (Notropis stramuneus ) 0 0 100 28Smallmouth bass (Micropterus dolomieui ) 18 82 0 16Spottail shiner (Notropis hudsonius ) 100 0 0 13Stonecat (Noturus flavus ) 7 0 93 32Tessellated darter (Etheostoma olmstedi ) 4 90 6 307White sucker (Catostomus commersoni ) 0 82 18 1,884Yellow perch (Perca flavescens ) 77 23 0 255

Fish species Abundance (%) by habitat Total individuals1

1Cumulative individuals of each fish species collected along the gradient over the three summer months.

21

DISCUSSION

The peak turnover rates in fish species compositions indicated the ecotone

peripheries between ST6 and ST7 on the ST gradient and between FL5 and FL6 on the

FL gradient over the three summer months of 2002. However, using the abrupt

changes in habitat variables as the ecotone indicator yielded different results between

the two gradients. The congruence of the abrupt changes in habitat conditions and fish

species composition at the ecotone periphery on the ST gradient revealed a strong fish-

habitat relationship (e.g., Przybylski et al. 1991, Winemiller and Leslie 1992,

Rakocinski et al. 1992, Wagner 1999, Willis and Magnuson 2000, Gido et al. 2002,

Martino and Able 2003) across lentic-lotic transitions on the gradient. This

consequence conformed to the hypothesis that the ecotone formation could be detected

using abrupt changes in either habitat conditions or fish species composition. In

contrast, the similar pattern of coincidental abrupt changes in both fish species

composition and the habitat conditions was not detected on the FL gradient. While the

peak turnover rate in fish species composition was detected at the same location, the

changes in most of the habitat variables varied along the FL gradient.

Using habitat condition to detect ecotone formation is difficult when the changes

in the habitat variables are gradual (Hansen et al. 1988). In the summer of 2002,

average intensities of many habitat variables showed gradual or no changes at the FL

ecotone periphery. The complex mix of gradual and abrupt changes in habitat

conditions along the FL gradient diminished the efficiency of this indicator to define

ecotone location. From this result, the peak turnover rate in fish species composition

appears to be the most powerful indicator for depicting the FL ecotone. This is based

on the ideas that changes in species assemblages across a transition zone can

22

apparently reflect both gradual and abrupt changes in habitat conditions (Armand

1992, Strayer et al. 2003, Hoeinghaus et al. 2004).

This study revealed that there were no shifts in ecotone location over the summer

periods. The ecotone peripheries (that indicated the ecotone orientation between lentic

and lotic systems) on both gradients were static in their locations throughout each

study summer period. The fix in location of the ecotones may be due to

environmental discontinuities (Pinay et al. 1990, Delcourt and Delcourt 1992). In this

study, unidirectional water flow from upstream towards downstream on the gradients

promotes mixing of materials and results in spatiotemporal heterogeneities and

physical discontinuities (Ward and Stanford 1983, Pinay et al. 1990) at the particular

gradient portions. The strong physical discontinuities (e.g., abrupt changes in water

depth, current velocity, substrate composition, and stream geomorphology) can

contribute to the stable formation of ecotones over space and time (Pinay et al. 1990).

Previous studies on similar gradient patterns by Przybylski et al. (1991) and Willis

and Magnuson (2002) concluded that river mouths connecting with lakes appeared to

be the ecotones with the highest abundance and species richness of fish observed, and

these conformed to the edge effect (Odum 1971). However, no quantitative methods

were used to define the ecotone formation in those studies. For this study, the ecotone

formation on both gradients was defined using statistical methods. The orientations of

both ST and FL ecotones were detected between the downstream and midstream

stations where the transitional conditions between lentic and lotic systems occurred

instead of at the stations between Lake Ontario and the embayments. These findings

indicated that differences in the sizes of the study gradients, habitat conditions, or

geographical regions (Kolasa and Zalewski 1995) could also contribute to the different

formation patterns and characteristics of the aquatic ecotones.

23

Methods of collecting fish species assemblages are another factor that can

influence the determination of the ecotone locations on the lentic-lotic gradients.

From this study, the ecotone periphery on the ST gradient was detected at the location

where the sampling procedure was changed from boat to manual electrofishing.

However, the ecotone periphery on the FL gradient was detected at the location where

the same sampling procedure was used (i.e., manual electrofishing). Although the

change from boat to manual electrofishing between the deep and shallow stations was

unavoidable in this study, the same electricity output was generated for stunning fish

with a similar pace of locomotion in all stations, and thorough samplings were

conducted to cover all habitat types in each station. From these efforts, the method

constraint for collecting fish between the deep and shallower stations on the gradients

might only trivially affect the results. Additionally, Rakocinski et al. (1992) indicated

that differences in fish species assemblages along environmental gradients represent

species-specific differences in habitat use rather than gear bias.

Ecotone distinctiveness can be scale dependent (Johnston et al. 1992, Wiens 1992,

Fagan et al. 2003), thus a proper spatial sampling design is required to determine an

ecotone (ecotone peripheries plus ecotone width—ECO habitat). For example, an

ecotone that is very abrupt on a community scale may be too small to be detected on a

continental scale because the resolution of data collection (on the continental scale) is

too broad (Johnston et al. 1992, Fagan et al. 2003). Hence, the scale of data

observation must be commensurate with the scale of an ecotone in a study (Johnston et

al. 1992, Wiens 1992, Fortin et al. 2000, Csillag et al. 2001, Cadenasso et al. 2003).

Using the fish species turnover rate as the ecotone indicator, the results from the field

observations along the FL gradient in both summers of 2002 and 2004 confirmed that

the FL ecotone was static at the same location during the summer periods. With the

finer and proper scale of study in the second summer, the entire ecotone was

24

successfully determined, and it was situated within the ecotone periphery detected in

the summer of 2002 (Figure 1.4).

Conditions in habitats adjacent to an ecotone play a strong role in governing the

ecotone geometry and permeability (Stamps et al. 1987, Wiens 1992, Fortin 1994,

Fagan et al. 2003). In this study, the ECO habitat on the FL gradient represents a

mixed zone of transition between the lentic and lotic systems. Based on the ecotone

orientation, the lower portion of the ECO habitat is most likely to be influenced by the

transition between the embayment and downstream conditions at the lower ecotone

periphery (between FL3 and FL4, Figure 1.4: squares). Likewise, the upper portion of

the ECO habitat tended to be affected by the transition between downstream and

midstream conditions at the upper ecotone periphery (between FL5 and FL6, Figure

1.4: squares). Consequently, the ECO habitat appears to be a unique area providing

diverse permeability (Wiens et al. 1985, Wiens 1992, Cadenasso et al. 1997, Ross

2005) for the distribution of fish species assemblages on the gradient.

Of the four ecotone functions (i.e., aggregator, mediator, soft barrier, and hard

barrier) on fish species assemblages, the “edge effect” of an increase in fish abundance

and species richness (Odum 1971) was not observed at the FL ecotone. In contrast,

the ecotone acted as a hard barrier allowing very low permeability for most of the fish

species observed in the summer of 2004 (Table 1.2). The physical discontinuities in

space or patch contrast appear to be another important factor affecting the species

distribution (Pinay et al. 1990, Forman and Moore 1992, Wiens 1992, Wiens 2002).

From the field observations, small cascades and long runs existing above the upper

ecotone periphery possibly obstruct the distribution of some stream fish species into

the ecotone. In addition, the decrease of water depth observed at the lower ecotone

periphery acted as a hard obstacle for many fish species to migrate into the ecotone.

While the FL ecotone provided low-to-zero permeability for most fish species, it was

25

concurrently a nursery ground for seven species (i.e., bluntnose minnow, brook

stickleback, largemouth bass, logperch, smallmouth bass, tessellated darter, and white

sucker). These particular species aggregated (>60%) in the ECO habitat during the

study summer, and most of the individual fish by species were young (0+ year-old).

The fundamental knowledge about ecotone formations on the two embayment-stream

gradients and the ecotone functions on fish species assemblages are useful for better

understanding ecotone properties in aquatic systems and future applications in

resource management (e.g., Petts 1990, Sisk and Haddad 2002) and fish habitat

restoration (Kirchhofer 1995, Kolasa and Zalewski 1995).

Overall, the results from this study show that the turnover rate in fish species

composition appears to be the best ecotone indicator on the embayment-stream

gradients. While abrupt changes in habitat conditions and species composition may

coincidentally occur at the same location reflecting the strong fish-habitat

relationships, the habitat variables showed less efficiency in defining the ecotone

formation on the gradient with the complex mix of habitat conditions. An appropriate

scale of study is required to determine ecotone width. The distinct physical

discontinuities below and above the ecotone detected on the FL gradient contributed to

low permeability of the ecotone on flux of the fish species. This study is the first

effort of its type using both biotic and non-biotic variables to detect ecotone formation

in the freshwater system associated with Lake Ontario. To explore ecotone properties

in a different water system, both functions and formations of ecotones were detected

along the tidal Hudson River, one of the world’s largest marine-freshwater gradients.

The methodology of study, results, and discussion are presented in Chapter 2.

26

CHAPTER TWO

Ecotone properties in the Hudson River estuary, New York

ABSTRACT

Ecotone properties were determined on the Hudson River estuary gradient during

1974–2001. Turnover rate in fish species composition was used to detect ecotone

formations and abundances of fish species assemblages in and around ecotones were

used to interpret ecotone functions. From the results, the Hudson ecotones showed

both changes in location and structure spatially and temporally. In some time periods,

the ecotones changed in their breadth (from broader to narrower or vice versa) or

considerably declined to be just ecotone peripheries. The ecotone peripheries mostly

detected at the lower gradient portion might result from the sharp changes in dissolved

oxygen (DO) and water salinity within this gradient range. The ecotone peripheries

mostly detected at the upper gradient portion might result from the abrupt changes in

DO, water temperature, SAV abundance, and dam constructions within this gradient

range. The ecotones were highly detected at the lower-middle portion of the gradient

and their formations tended to relate with the changes in salinity, tide, and freshwater

flow. Functional patterns of the Hudson ecotones were not distinctly different over

time. One ecotone mostly detected in the lower-middle gradient portion appeared to

be the optimal zone for distributions of diverse fish assemblages. However, the other

detected ecotones acted as barriers for most fish species. Likewise, the ecotone

peripheries inhibited distributions of most fish species. These results might relate with

complex migratory behaviors and life histories of fish and their responses to variations

in the estuarine conditions in space and time.

27

INTRODUCTION

Ecotone studies have been prominent and appealing (e.g., Clements 1897, Leopold

1933, Odum 1971, Risser 1990, Winemiller and Leslie 1992, Lidicker 1999, Willis

and Magnuson 2000, Harding 2002, Martino and Able 2003, Ries et al. 2004) because

of the important roles of ecotones on biodiversity (di Castri and Hansen 1992, Neilson

et al. 1992, Lachavanne 1997), biotic interactions (Wiens et al. 1993), flows of

materials and nutrients in ecosystems (Hansen and di Castri 1992), and the ecological

implications for natural resources management at the individual, population, and

ecosystem levels (Sisk and Haddad 2002). Clements (1897) was the first who viewed

an ecotone as the transition zone between different plant communities. He observed

that the vegetation in this zone was often accentuated in size and density. Leopold

(1933) later explained that shape and spatial configuration of an ecotone influence

species richness and organism abundance. Working with these early ideas, Odum

(1971) defined an ecotone as a transition zone between two or more communities and

he labeled the increases in both species richness and abundance at the ecotone as a

result of the “edge effect”.

At present, an ecotone is widely viewed as a transition zone between adjacent

ecological systems and its influences on organism distributions have been considered

beyond the traditional edge effect. Additionally, scientists may use diverse terms to

call a transition zone depending on the spatial scale considered and the prior

delineation of the system of interest. In this study, an ecotone periphery is used to

represent a narrow transition zone or a boundary with one less dimension than a patch

(e.g., Stamps et al. 1987, Fagan et al. 1999, Strayer et al. 2003), whereas an ecotone

refers to a broader transition zone (Fagan et al. 1999) comprising ecotonal habitat and

ecotone peripheries. An ecotone may form, decline, disappear, shift locations, or

change in terms of structure (from an ecotone to be just an ecotone periphery) or

28

function due to the influence of changing physicochemical conditions, species use of

habitat, and interactions among species (Holland 1988, Delcourt and Delcourt 1992,

Wiens 1992, Kolasa and Zalewski 1995).

As a zone of transition in biotic and physicochemical conditions, an ecotone

provides different permeability on flux of organisms. In this study, ecotone

permeability is considered in terms of ecotone function on organism distributions.

Other than attracting particular organisms, the mechanism that produces the edge

effect may simultaneously inhibit dispersion of other organisms into an ecotone. The

same ecotone may be neutral and has no effect on distribution of some other species

into the ecotone. Dynamic properties (formation and function) of ecotones can be

obviously observed in an estuary, a complex ecosystem where the interaction between

marine and freshwater systems occur, and tide and water flow play strong influence.

As one of the world’s important estuaries, the tidal Hudson River drainage serves

more than 200 fish species (Stanne et al. 1996, Daniels et al. 2005) from freshwater,

estuarine, and marine assemblages. Other than providing variety of habitats for

aquatic creatures, the Hudson River estuary has been heavily altered by human

activities, such as urbanization, shoreline development, dam construction, power plant

operation, and water diversion for commercial, industrial, and agricultural purposes

(Limburg et al. 1986, Barnthouse et al. 1988, Schmidt and Cooper 1996, Wall and

Phillips 1998, Baker et al. 2001, Phillips et al. 2002, Daniels et al. 2005).

The tidal Hudson River portions where the intermediate conditions between fresh

and saline waters meet are apparently vital ecotones for living of various fish species,

especially in their first year of life (Weinstein et al. 1980, Stanne et al. 1996). At the

lower-middle portion of the tidal Hudson River, the locations and spatial structures of

the Hudson ecotones tend to vary in space and time due to the changes in salinity

regime, tide, and freshwater runoff. At the upper portions of the tidal Hudson River,

29

effects of dam constructions and water pollutions, e.g., low dissolve oxygen, high

coliform densities (Boyle 1979), and large loads of polychlorinated biphenyls

discharged from several general electric factories above the Federal Dam (Baker et al.

2001) may contribute to sharp formations of ecotone peripheries affecting fish

distributions below and above the Federal Dam. Although ecotone formations and

their functions on organism distributions can be used as a baseline for resources

managers to maximize healthy habitats for species assemblages of interest, these

ecotone properties have never been revealed in the Hudson system.

The first objective of this study was to detect ecotone formations on the Hudson

River estuary gradient. Spatial structures of the Hudson ecotones were determined

under the hypothesis that complex mix of abrupt and gradual changes in habitat

conditions might contribute to the formation of an ecotone (comprising ecotonal

habitat and ecotone peripheries), whereas abrupt changes in habitat conditions might

contribute to a narrow formation of an ecotone periphery alone. Over space and time,

some ecotones may be fixed in location, new ecotones may form, or former ecotones

may decline or change in their breadths. The study habitat variables were dissolved

oxygen, salinity, water temperature, two major types of submerged aquatic vegetation

(Trapa natans and Vallisneria americana), and bottom substrates of water. Changing

patterns of these habitat variables in and around the ecotone pay strong role on

distribution patterns of fish assemblages along the Hudson River estuary gradient.

The second objective of the study was to detect ecotone functions on the Hudson

fish species assemblages over time. Ecotone functions on fish species assemblages are

interpreted from observed abundance of fish by species in and around an ecotone. An

ecotone may act as an aggregator for particular fish species. Concurrently, it may act

as a hard barrier completely blocking distribution of some fish species into the

ecotone, or as a soft barrier partially allowing some fish species to enter the ecotone.

30

In the latter case, most species populations entering the ecotone will return to the

habitats from which they came, causing few of them to be observed in the ecotone.

An ecotone formed somewhere on the Hudson River estuary gradient may act as a

mediator for marine and freshwater fish species with broad tolerance on changing

salinity. Over space and time, an ecotone may decline in its structure to be just an

ecotone periphery and may completely or partly inhibit distribution of particular fish

species between adjacent habitats. Concomitantly, an ecotone periphery may be

neutral and has no effect on fish distributions between adjacent habitats it separates.

MATERIALS AND METHODS

Study gradient and data sources

The Hudson River estuary starts from Battery at river kilometer (RKM) 1 in

Manhattan, New York City, to the Federal Dam at RKM 246 in Troy, Albany, New

York State (Figure 2.1). In this study, the gradient was divided into four zones. The

lowest zone (RKMs 1–19) was normally polyhaline with salinity concentration of 18–

30 ppt (Ristich et al. 1977, National Oceanic and Atmospheric Administration 1989).

At this zone, the river channel was approximately 1 km wide and 9–15 m deep (Coch

and Bokuniewicz 1986). The above three zones (the lower, the middle, and the upper

zones) were classified following Schmidt et al. (1988).

31

Figure 2.1. The Hudson River estuary gradient showing 13 regions designated for the five power plant utilities’ Hudson River estuary monitoring program.

32

At the lower zone (RKMs 19–61), the river channel was 1.0–1.5 km wide and 13–

17 m deep (Coch and Bokuniewicz 1986). At the middle zone (RKMs 61–122), the

river channel was 0.5–1.0 km wide and 18–48 m deep (Coch and Bokuniewicz 1986).

At the upper zone (RKMs 122–246), the river channel was 0.6–1.2 km wide and 15–

31 km deep (Coch and Bokuniewicz 1986). Salinity in water varies temporally

between mesohaline and oligohaline in the lower and the middle zones depending on

tide and freshwater flow, while water was usually fresh in the upper zone (Strayer et

al. 2004).

Fish data used to determine ecotone properties on the Hudson River estuary

gradient were from the beach seine survey (BSS) during 1974–2001 of the Utilities’

Hudson River estuary monitoring program. The Utilities collectively represents the

five power plant utilities including Central Hudson Gas and Electric Corporation,

Consolidated Edison Company of New York Inc, New York Power Authority, Niagara

Mohawk Power Corporation, and Southern Energy New York. Based on the

monitoring program, the Hudson River estuary was divided into 13 regions (RE)

starting from RE0 (RKMs 1–19) to RE12 (RKMs 201–246; Figure 2.1).

The monitoring program used stratified random samplings to conduct the BSS

along the tidal river gradient. According to the stratification method, fish assemblages

were randomly sampled along the shore zone (≤3-m water depth) using a 30.5-m total

length beach seine with a 0.5-cm bag mesh size. At each sampling point, one end of

the net (of the beach seine) was held on shore and the other end was towed

perpendicularly away from shore by boat sweeping approximately 450 m2 of sampling

area. Main purpose of the survey was to obtain abundance and distribution data of

young of the year of target fish species including Atlantic tomcod (Microgadus

tomcod), American shad (Alosa sapidissima), striped bass (Morone saxatilis), and

33

white perch (Morone americana) along inshore zone of the Hudson River estuary

(Texas Instruments 1981).

At least three observations were taken at each region in each sampling period of

the BSS. Yearly numbers of the observations were 2,331–3,648 during 1974–1980,

500–800 during 1981–1984, and 1,000 during 1985–2001. The BSS was done weekly

from: April to the 2nd week of December during 1974–1975 and 1978–1979, April

through November during 1976–1977, April to the 3rd week of August and then

biweekly through November in 1980. During 1981–1995, the BSS was done biweekly

each year. The biweekly survey schedules were from: August to the 2nd week of

October during 1981–1983, last week of July to the 1st week of October in 1984, the

2nd week of July to the 3rd week of November during 1985–1986, the 4th week of June

to the 2nd week of November in 1987, June through October during 1988–1992 and

1994–1995, and last week of June to the 1st week of November 1993. During 1996–

2001, the BSS was done biweekly each year from July to the 3rd week of October.

The BSS method in detail can be found from the Utilities’ early or recent monitoring

programs (e.g., Texas Instruments 1977, ASA Analysis & Communication 2004).

Habitat data including dissolved oxygen (DO), salinity, water temperature,

submerged aquatic vegetation (SAV), and substrate composition were used to describe

habitat conditions along the Hudson River estuary gradient (RE1–RE12) in this study.

The first three habitat variables were from the Utilities’ BSS, and they were in-situ

measured at approximately 0.3 m below the water surface and 15 m from the shoreline

using a YSI Model 57 Dissolved Oxygen Meter (Texas Instruments 1976). Mean

value for each of the three habitat variables at each sampling location and sampling

week, weighted by stratum volume, was calculated using equation (1) as

34

∑=

=rwn

iirwrwrw n

1W/1W , (1)

where Wrw is the mean of a habitat variable at river mile r during bi-week w of the

BSS, Wirw is a habitat measurement for location i at river mile r during bi-week w, and

nrw is a number of habitat measurement taken at river mile r during bi-week w. DO

was measured to the closet 0.1 mg/L, and water temperature was measured to the

closest 0.1°C. Conductivity of water was measured in millisieman/cm at 25 °C (C25)

and it was used to compute salinity concentration (ppt) using equation (2) as (Texas

Instruments 1976)

Salinity = -100 ln (1-C25/178.5) (2)

SAV data were from the study by the collaboration among the Institute of

Ecosystem Studies, the Hudson River National Estuarine Research, New York State

Department of Environmental Conservation (NYSDEC), and the Cornell Institute for

Resource Information Systems. The SAV study was conducted from RE2 to RE12 in

August 1995 and July–August 1997. Photographic, digital, and grab sampling data

were used to classify and map bed areas of Trapa natans (water chestnut—rooted with

floating leaves) and Vallisneria americana (water celery).

Substrate data from the Utilities’BSS (1974–2001) and NYSDEC (1998–2003)

were combined to describe bottom condition along the study gradient. The Utilities’

substrate data were observed in shallow water areas (≤3 m deep) from RE1 to RE12

and its composition was classified into four types: mud, sand, gravel with <76 mm

grain size, and gravel with ≥76 mm grain size. The NYSDEC’s substrate data were

collected across the Hudson River estuary system, but they were not completely done

35

in shallow water <4 m deep (Bell et al. 2006). Both core and grab samples were

categorized with some guidance from the acoustic backscatter data (Nitsche et al.

2004) and the sediment profiler imagery data (Rhoads and Cande 1971, Rhoads and

Germano 1982). Based on grain size and percentage composition, the NYSDEC’s

substrates were classified into: mud (<0.062 mm grain size), sand (0.062–2 mm),

gravel (2–36 mm), and six mixed groups including sandy mud (mud with >10% sand),

gravelly mud (mud with >10% gravel), muddy sand (sand with >10% mud), gravelly

sand (sand with >10% gravel), muddy gravel (gravel with >10% mud), and sandy

gravel (gravel with >10% sand).

Ecotone and habitat analyses

Because the Utilities’ BSS was not conducted in RE0 prior to 1996, the ecotone

and habitat analyses were done from RE1 to RE12 on the tidal Hudson River gradient

based on the BSS fish and habitat data available from 1974 to 2001. Ecotone

formation was detected monthly in all years the fish data were available for all 12

regions (RE1–RE12). Turnover rate or beta diversity (Whittaker 1975, Gauch 1982,

Delcourt and Delcourt 1992, Winemiller and Leslie 1992, Wagner 1999, Hoeinghaus

et al. 2004) in fish species composition was used to detect ecotone formation on the

gradient. Detrended correspondence analysis (DCA) in CANOCO 4.0 software (ter

Braak and Smilauer 1998) was used to compute the beta diversity (hereafter called

beta A) based on both occurrence and percentage abundance by species of fish in each

region. In the computation, the sample scores on the DCA ordination are rescaled into

the average standard deviation (SD) units of species turnover (Hill 1979). A gradient

length of 1.0 SD represents approximately one-fourth of a complete turnover rate in

species composition on the gradient (Gauch 1982). Species appear, rise to their

modes, and disappear over a distance of about 4 SD. That is, if the beta A on the

36

DCA’s axis 1 (that accounts for the most variation) is ≥4.0 SD, it indicates that species

compositions at both extreme ends of a gradient represent completely different

communities (Gauch 1982). This indicates complete formation of at least one ecotone

on the gradient. The peak absolute difference (≥1 SD for this study) in the beta A (δ

beta A) indicates an ecotone periphery (a margin where a sharp transition between

adjacent habitats occurs).

Multiple functions of an ecotone (an ecotonal habitat plus its ecotone peripheries)

were interpreted from the relative observed abundance by species of fish among an

ecotonal habitat, an upper adjacent habitat, and a lower adjacent habitat. This study

proposed that an ecotone acts as an aggregator for a fish species with >60% abundance

observed in the ecotonal habitat and as a mediator for fish species with 30–60%

abundance observed in the ecotonal habitat. Concurrently, an ecotone acts as a soft

barrier for a fish species with 5–30% abundance observed in the ecotonal habitat.

Lastly, an ecotone acts as a hard barrier for a fish species with <5% abundance

observed in the transitional habitat. Forming as an ecotone periphery (ep) between

adjacent habitats (without ecotone width), the ep may be neutral allowing fish species

with >30% abundance to distribute between the adjacent habitats. Concurrently, an ep

may partly or completely inhibit distribution of a fish species from one habitat to the

other adjacent habitat causing low (5–30%) or lower-absent (0–5%) abundance,

respectively, of fish in the latter habitat. Each fish species with <5 individuals

observed along the gradient in each analyzed time period (i.e., a certain month in a

certain year) was not included in the comparison because of low catch.

General linear model with multivariate analysis of variance (GLM MANOVA)

procedure in SPSS version 14.0 software was used to compare mean values of DO,

salinity, water temperature, and their associated delta values (i.e., absolute difference

between minimum and maximum values) between two periods of ecotone analysis

37

(i.e., non-detected period, ECO = 0 and detected period, ECO = 1) and due to the

interactions of ECO × RE, ECO × Month, ECO × Year, ECO × RE × Month, ECO ×

RE × Year, and ECO × Month × Year. Tukey’s honestly significant differences

(HSD) test was used to monthly compare mean values of each habitat variable

between paired regions when the significant difference of the variable was detected.

Bed areas (km2) of two SAV species (Trapa natans and Vallisneria americana) from

the SAV shapefile were estimated by region along the Hudson River estuary gradient

using ArcGIS 9.1.

Area (km2) of each substrate type including mud (<0.062 mm grain size), sand

(0.062–2 mm), gravel (2–76 mm), cobble (>76 mm), and the six mixed groups from

the NYSDEC and Utilities’ datasets were combined and re-calculated. Because the

NYSDEC’s substrate data were less completed along the Hudson inshore zone, only

offshore area for each substrate type was retrieved from this dataset and summed up

with the inshore area of the associated substrate type estimated from the Utilities’ BSS

for each associated region. That is, after the area for each substrate type from the

NYSDEC’s shapfile was estimated by region using ArcGIS 9.1, the offshore area for

each substrate type in each associated region was re-computed using equation (3) as

follows:

( )Utilities(shore,DEC)(total,DEC)(total,

DEC),(sub,offshore)(sub AreaArea

Area

AreaArea ii

i

jiji −×= , (3)

where Areai(sub j, offshore) is the estimated area of substrate type j at region i (i = 1, 2,

3,…, 12) in the offshore zone, Areai(sub j, DEC) is the estimated area of substrate type j at

region i based on the NYSDEC’s data, Areai(total, DEC) is the estimated area for all

substrate types at region i based on the NYSDEC’s data, and Areai(shore, Utilities) is the

estimated area in the inshore zone at region i based on the Utilities’ data.

38

RESULTS

A total of 127 fish species (excluded 14 unidentified species) from 47 families was

recorded along the Hudson River estuary gradient (RE1–RE12) from the BSS during

1974–2001. Of the 47 families, four families: Cyprinidae, Centrarchidae, Clupeidae,

and Sciaenidae had the highest species richness. Blueback herring (Alosa aestivalis)

was the most abundant (34%) species and distributed throughout the Hudson River

estuary. The latter four fish species with >8% observed abundance on the gradient

were bay anchovy (Anchoa mitchilli, 9%), white perch (Morone americana, 9%),

unidentified clupeid (9%), and American shad (Alosa sapidissima, 8%).

Distribution patterns of fish along the Hudson River estuary gradient are

apparently shaped by fish migratory behaviors (anadromous and catadromous) and

salinity tolerance across marine-estuarine and estuarine-freshwater transitions. Based

on these criteria, the Hudson fish species were classified into five assemblages,

comprising anadromous, catadromous, estuarine, freshwater, and marine (Smith

1985). Of the total 127 identifiable fish species recorded along the gradient, nine

species were anadromous fish, one species (American eel—Auguilla rostrata) was

catadromus, 13 species were estuarine fish, 59 species were freshwater fish, and 45

species were marine fish (Appendix 2.1).

Ecotone formation and function

The ecotones were detected in 116 time periods (76%) of the total 152 time

periods the fish data were available for the analysis during 1974–2001 (Figure 2.2).

39

Figure 2.2. Frequency of ecotones detected monthly based on the data collection along the Hudson River estuary gradient during 1974–2001.

40

The ecotone was detected with the highest frequency (18 times) at RE2, while the

ecotone peripheries alone were highly detected between RE1 and RE2 (18 times) and

between RE11 and RE12 (30 times) on the Hudson River estuary gradient (Figure

2.3). Ecotones and ecotone peripheries were less detected at the remaining portions on

the gradient (Figure 2.3).

Figure 2.3. Ecotones (thick bars) and ecotone peripheries (thick lines) detected during 1974–2001on the Hudson River estuary gradient covering 228 river kilometers (RKM) from region (RE) 1 to RE12 are grouped into each frequency class. Numbers of times of the formation of an ecotone and an ecotone periphery in each location are shown in brackets in each frequency class on the right side of the graph.

With the highest frequency (18 time periods) of detection at RE2, this ecotone

acted as a hard barrier for most (>35%) of the total fish species observed along the

41

gradient in the associated time periods (Table 2.1). Similarly, the ecotones detected at

other locations (except at RE2–RE4) acted as a hard barrier for most of the Hudson

fish species (Table 2.1, see ecotone functions from all detected frequencies in

Appendix 2.2). Detecting 10 times at RE2–RE4, this ecotone acted as an aggregator

for most of the fish species observed from the six time periods (July 1986, August

1979, September 1982 and 1988, October 1975, and November 1986), but it acted as a

mediator, soft barrier, or hard barrier for most of the fish species observed in the rest

time periods (Table 2.1). Consistent with most of the ecotones functioning as hard

barriers, the ecotone peripheries detected alone between particular adjacent regions on

the gradient completely inhibited the distribution between the adjacent regions of most

(>35%) fish species in the analyzed time periods (Table 2.2, see ecotone periphery

functions from detected frequencies in Appendix 2.3).

Table 2.1. Functions of the ecotones detected ≥5 times at certain regions (RE) and time periods on Hudson fish species assemblages.

Ecotone No. of selected1

location Aggregator Mediator Soft barrier Hard barrier species/gradient

RE2May-75 3.1 21.9 28.1 46.9 32May-76 6.7 26.7 26.7 39.9 30May-78 15.6 12.5 21.9 50.0 32Jun-74 8.8 11.8 20.6 58.8 34Jun-75 10.9 16.2 24.3 48.6 37Jun-76 12.2 9.8 34.1 43.9 41Jun-77 16.7 13.9 25.0 44.4 37Jul-78 7.0 18.6 14.0 60.4 43Jul-79 26.2 28.6 11.9 33.3 42Jul-84 31.0 21.4 9.5 38.1 42Jul-90 32.3 19.4 12.9 35.4 31Aug-78 10.2 18.4 20.4 51.0 49Aug-84 36.6 14.6 22.0 26.8 41Aug-96 21.4 10.7 17.9 50.0 28Sep-84 31.3 12.5 15.6 40.6 32Nov-76 18.2 13.6 22.7 45.5 22Nov-79 37.5 17.5 12.5 32.5 40Dec-78 0.0 11.1 27.8 61.1 18

Function (%) for the total selected species

42

Table 2.1 (Continued).

Ecotone No. of selected1

location Aggregator Mediator Soft barrier Hard barrier species/gradient

RE2-RE3Jun-91 19.0 23.9 21.4 35.7 42Jul-00 20.0 17.1 8.6 54.3 35Sep-79 42.9 34.3 11.4 11.4 35Sep-81 27.9 9.3 23.3 39.5 43Sep-87 46.0 16.2 18.9 18.9 37Sep-99 12.5 6.3 18.8 62.4 32Oct-83 36.8 5.3 21.1 36.8 19Nov-78 16.7 33.3 13.3 36.7 30

RE2-RE4Jul-86 42.1 15.8 18.4 23.7 38Aug-74 22.0 30.0 22.0 26.0 50Aug-79 36.4 25.0 18.1 20.5 44Sep-82 38.5 25.6 18.0 17.9 39Sep-88 51.5 9.1 15.2 24.2 33Sep-96 30.0 20.0 30.0 20.0 20Oct-75 38.9 22.2 22.2 16.7 36Oct-98 22.6 19.4 6.5 51.5 31Nov-80 23.5 29.5 23.5 23.5 17Nov-86 36.0 28.0 12.0 24.0 25

RE3-RE4Apr-74 50.0 11.1 16.7 22.2 18May-78 37.5 37.5 15.6 9.4 32Jun-75 29.8 40.5 10.8 18.9 37Jul-75 25.6 30.2 23.3 20.9 43Aug-81 18.0 35.9 28.2 17.9 39Aug-97 3.6 25.0 42.8 28.6 28Sep-01 3.5 13.8 24.1 58.6 29Oct-84 20.1 23.3 33.3 23.3 30Dec-78 38.9 27.8 11.1 22.2 18

RE3-RE5Aug-98 15.2 18.2 33.3 33.3 33Sep-98 7.0 10.3 37.9 44.8 29Oct-85 20.0 20.0 50.0 10.0 30Oct-92 21.4 25.0 14.3 39.3 28Nov-87 14.3 7.1 35.7 42.9 14

RE4-RE5Jul-90 3.2 9.7 12.9 74.2 31Aug-99 8.2 10.8 32.4 48.6 37Oct-83 15.8 26.3 31.6 26.3 19Oct-93 13.3 20.1 33.3 33.3 30Sep-81 4.7 9.3 44.2 41.8 43

RE5Jun-91 0.0 4.8 26.2 69.0 42Jun-92 0.0 8.3 25.0 66.7 24Jul-93 0.0 13.9 22.2 63.9 36

Function (%) for the total selected species

43

Table 2.1 (Continued).

Ecotone No. of selected1

location Aggregator Mediator Soft barrier Hard barrier species/gradient

RE5 (Continued)Sep-93 7.0 3.4 31.0 58.6 29Sep-96 15.0 10.0 30.0 45.0 20Dec-78 50.0 16.7 5.6 27.7 18

RE6Jun-91 0.0 7.1 16.7 76.2 42Jun-92 0.0 16.7 20.8 62.5 24Jul-90 3.2 3.2 13.0 80.6 31Aug-88 3.2 15.6 28.1 53.1 32Dec-78 5.6 16.7 11.0 66.7 18

RE7-RE11Aug-87 21.2 27.3 18.2 33.3 33Aug-97 29.4 14.7 20.6 35.3 34Sep-87 19.4 22.3 19.4 38.9 36Sep-90 16.6 26.7 26.7 30.0 30Oct-88 27.6 6.9 31.0 34.5 29

Function (%) for the total selected species

1A fish species with less than 5 individuals collected along the Hudson River estuary gradient in each of the 116 time periods was not included in the analysis.

Table 2.2. Functions of the ecotone peripheries detected ≥5 times at certain regions (RE) and time periods on Hudson fish species assemblages.

Ecotone periphery No. of selected1

location between Neutral Partly inhibit Completely inhibit species/gradient

RE1-RE2Apr-78 8.0 20.0 72.0 25Apr-79 12.5 29.2 58.3 24May-74 26.1 4.3 69.6 23May-77 12.2 24.2 63.6 33Jun-78 12.5 20.0 67.5 40Jul-76 13.3 15.6 71.1 45Aug-80 12.2 9.8 78.0 41Aug-82 8.9 23.5 67.6 34Aug-89 10.0 10.0 80.0 30Aug-93 14.7 2.9 82.4 34Sep-75 8.3 22.9 68.8 48Sep-83 6.3 18.7 75.0 32Sep-85 11.8 23.5 64.7 34Sep-86 3.0 9.1 87.9 33Oct-74 11.8 17.6 70.6 34

Function (%) for the total selected species

44

Table 2.2 (Continued).

Ecotone periphery No. of selected1

location between Neutral Partly inhibit Completely inhibit species/gradient

RE1-RE2 (Continued)Oct-77 15.2 27.3 57.5 33Oct-81 16.7 13.9 69.4 36Nov-77 14.7 23.5 61.8 34

RE3-RE4Jun-88 17.1 17.1 65.8 35Jul-96 18.0 25.6 56.4 39Sep-76 18.0 41.0 41.0 36Sep-80 11.1 33.3 55.6 36Sep-94 8.6 17.1 74.3 35Oct-78 25.7 34.3 40.0 35Oct-80 15.4 26.9 57.7 26Oct-86 25.8 29.0 45.2 31Oct-95 13.3 30.0 56.7 30

RE10-RE11Jun-74 35.3 35.3 29.4 34Jun-78 22.5 35.0 42.5 40Jul-75 30.2 30.2 39.6 43Aug-79 29.5 29.5 41.0 44Sep-75 22.9 14.6 62.5 48Sep-80 30.6 19.4 50.0 36Oct-74 14.7 38.2 47.1 34Nov-80 11.8 5.9 82.3 17Nov-86 24.0 24.0 52.0 25

RE11-RE12May-74 13.1 21.7 65.2 23May-77 9.1 30.3 60.6 33May-78 21.9 18.7 59.4 32May-80 16.7 13.9 69.4 36Jun-79 13.9 16.7 69.4 36Jun-80 20.0 32.5 47.5 40Jun-95 9.6 33.3 57.1 21Jul-76 19.5 31.7 48.8 41Jul-77 14.0 25.5 60.5 43Jul-78 28.6 26.2 45.2 42Jul-97 8.6 20.0 71.4 35Jul-00 20.6 32.4 47.0 34Aug-75 20.0 28.9 51.1 45Aug-76 29.2 20.8 50.0 48Aug-78 28.6 22.4 49.0 49Aug-80 26.8 19.5 53.7 41Sep-76 25.6 18.0 56.4 39Sep-78 28.9 15.8 55.3 38Sep-79 22.9 22.9 54.2 35Sep-82 20.5 18.0 61.5 39Sep-94 20.0 20.0 60.0 35Sep-99 18.8 43.8 37.4 32

Function (%) for the total selected species

45

Table 2.2 (Continued).

Ecotone periphery No. of selected1

location between Neutral Partly inhibit Completely inhibit species/gradient

RE11-RE12 (Continued)Sep-01 16.7 13.3 70.0 30Oct-76 10.3 25.6 64.1 39Oct-78 20.0 11.4 68.6 35Oct-79 22.3 19.4 58.3 36Oct-80 15.4 26.9 57.7 26Oct-93 10.3 13.8 75.9 29Nov-78 16.7 10.0 73.3 30Nov-79 13.4 23.3 63.3 30

Function (%) for the total selected species

1A fish species with less than 5 individuals collected along the Hudson River estuary gradient in each of the 116 time periods was not included in the analysis

Changing habitat related to ecotone formation

Only mean values of DO were significantly different (GLM MANOVA: p≤0.050)

between the non-detected (ECO = 0) and detected (ECO = 1) ecotone periods, whereas

mean values of DO, delta salinity, and delta water temperature were significantly

differences due to the interactions of ECO × RE (Table 2.3). Mean values of all

habitat variables and their associated deltas showed significant differences due to the

interactions of ECO × Month, ECO × Year, ECO × RE × Month, ECO × RE × Year,

and ECO × Month × Year (Table 2.3). Mean values of DO, salinity, and water

temperature were then further examined over space (RE) and time (Month) based on

the results from Turkey’s HSD post hoc.

The DO along the Hudson River estuary gradient ranged from 11 to 13 mg/L in

April, but decreased to 9–11 mg/L in May, and 7–10 mg/L during June–September.

The DO increased again to 8–10 mg/L in October and 10–14 mg/L in December

(Figure 2.4).

46

Table 2.3. Results from multivariate analysis of variance for each habitat variable and its delta (absolute difference between minimum and maximum values) between non-detected (ECO = 0) and detected (ECO = 1) ecotone periods and due to the interactions of ECO × RE (river region), ECO × Month, ECO × Year, ECO × RE × Month, ECO × RE × Year, and ECO × Month × Year.

Dependent Mean Variable Square

Dissolved oxygen 1 2.45 4.71 0.030Delta dissolved oxygen 1 0.02 0.01 0.921Salinity 1 0.03 0.06 0.804Delta salinity 1 1.27 1.03 0.311Water temperature 1 0.07 0.12 0.730Delta water temperature 1 2.01 0.60 0.440

Dissolved oxygen 11 1.08 2.07 0.020Delta dissolved oxygen 11 2.70 1.13 0.336Salinity 11 0.75 1.72 0.064Delta salinity 11 3.60 2.90 0.001Water temperature 11 0.97 1.78 0.053Delta water temperature 11 9.41 2.78 0.001

Dissolved oxygen 6 3.06 5.89 0.000Delta dissolved oxygen 6 8.59 3.59 0.002Salinity 6 3.59 8.21 0.000Delta salinity 6 8.95 7.23 0.000Water temperature 6 11.90 21.83 0.000Delta water temperature 6 23.04 6.81 0.000

Dissolved oxygen 18 3.53 6.79 0.000Delta dissolved oxygen 18 12.97 5.42 0.000Salinity 18 3.63 8.30 0.000Delta salinity 18 5.81 4.69 0.000Water temperature 18 9.73 17.86 0.000Delta water temperature 18 30.19 8.93 0.000

Dissolved oxygen 154 1.51 2.90 0.000Delta dissolved oxygen 154 2.94 1.23 0.043Salinity 154 2.11 4.84 0.000Delta salinity 154 2.88 2.33 0.000Water temperature 154 2.59 4.75 0.000Delta water temperature 154 7.37 2.18 0.000

Dissolved oxygen 493 1.27 2.44 0.000Delta dissolved oxygen 493 3.55 1.48 0.000Salinity 493 1.77 4.04 0.000Delta salinity 493 2.86 2.31 0.000Water temperature 493 0.63 1.16 0.026Delta water temperature 493 6.99 2.07 0.000

ECO × RE × Month

Source p-value

ECO × RE × Year

ECO

ECO × RE

ECO × Month

ECO × Year

df F

47

Mean values of DO between the detected and non-detected ecotone periods were

significantly different at particular region(s) in all eight months included in the

analysis, however, inconsistent patterns were shown over time (Figure 2.4).

Figure 2.4. Monthly mean values of dissolved oxygen observed along the Hudson River estuary gradient in the detected ecotone (line with solid dot markers) and non-detected ecotone (line with open dot markers) periods.

48

Focusing on mean values of DO in the detected ecotone period, the abrupt changes

of DO from RE1 to RE4 and between RE11 and RE12 in most of the analyzed months

(Figure 2.4) were parallel with high frequencies of detecting ecotones and ecotone

peripheries within this gradient range.

Unlike DO, the salinity showed consistent patterns of change over time along the

Hudson River estuary gradient. Mean values of salinity along the gradient were lower

in April and May, but higher in June, October, and December, and reached the highest

during July–September (Figure 2.5). In general, the salinity was the highest at RE 1

that was influenced by the marine conditions of New York Bay and it abruptly

decreased over distance until became tidal fresh at the upper zone of the gradient

(Figure 2.5).

The mean values from the non-detected ecotone period were significantly higher

than those from the detected ecotone period at RE1–RE2 in May, but they were

significantly lower at RE6 in April, RE1–RE2 in October, and RE1–RE4 in December

(Figure 2.5). The salinity along the gradient was similar between the two periods in

the remaining months (Figure 2.5).

49

Figure 2.5. Monthly mean values of salinity observed along the Hudson River estuary gradient in the detected ecotone (line with solid dot markers) and non-detected ecotone (line with open dot markers) periods.

Water temperature along the Hudson River estuary gradient was 7–10°C in April

and it increased to 14–17°C in May. The water temperature was higher during June–

August (22–27°C), but lower in September (20–24°C) and October (14–18°C), and the

lowest in December (2–7°C; Figure 2.6).

50

Figure 2.6. Monthly mean values of water temperature observed along the Hudson River estuary gradient in the detected ecotone (line with solid dot markers) and non-detected ecotone (line with open dot markers) periods.

Mean values of the water temperature in the detected ecotone period were

significantly lower than those in the non-detected ecotone period at most regions in

May and June, RE1–RE6 in July, and RE7–RE12 in December (Figure 2.6). In April,

51

August, and September, the mean values of water temperatures from the non-detected

ecotone period was significantly higher at some particular regions, while there was no

significant difference of the mean values along the gradient between the two periods in

October (Figure 2.6).

Two SAV species (Trapa natans and Vallisneria americana) and their bed areas

varied along the tidal river gradient (Figure 2.7) above RE1, where the SAV data were

unavailable. Trapa natans beds were not observed from RE2 to RE4, where the

ecotones were highly detected within this range. The estimated bed areas of Trapa

natans along the remaining portion on the gradient varied from 0.1 km2 at RE5 and

RE12 to the highest (1.8 km2) at RE11 (Figure 2.7). Bed areas of Vallisneria

americana were observed from RE2 to RE12 and ranged from the lowest (0.2 km2) at

RE2 and RE8 to the highest (5.3 km2) at RE10 (Figure 2.7).

Abrupt changes in bed areas of the two SAV species between RE11 and RE12

(Figure 2.7) were congruent with the highest frequency of the ecotone periphery

detected between these two regions. Bottom of water along the study gradient was

mainly a combination of mud and sandy mud from RE1 to RE9, and a combination of

sand and muddy sand from RE10 to RE12 (Figure 2.7). Four substrate types including

gravelly sand, gravel, sandy gravel, and cobble were observed at low percentage

compositions along the gradient (Figure 2.7). Gravelly mud and muddy gavel were

observed <2% and at RE1 only.

52

Figure 2.7. Values of dissolved oxygen (DO), salinity, and water temperature were averaged by region over the 116 time periods that the ecotones were detected. Bed areas of two submerged aquatic vegetation (SAV) species (Trapa natans and Vallisneria american) vary along the gradient (data for region 1 were unavailable). Proportions of 10 substrate types were observed in the tidal river estuary. Two of them (gravelly mud and muddy gravel) not shown in the graph were observed <2% and at RE1 only.

53

DISCUSSION

Formations of the Hudson ecotones were highly dynamic showing both shift in

location and change in structure over the detection periods (1974–2001). The ecotone

formations were related with the changes in habitat conditions over space and time.

Changeovers in fish species composition influencing by the transitions in

physicochemical conditions (Winemiller and Leslie 1992, Marshall and Elliott 1998,

Whitfield 1999, Martino and Able 2003) indicated the ecotone formations at the

particular portions on the Hudson River estuary gradient. Consistent with the study by

Attrill and Rundle (2002), the tidal Hudson River represents two important transitional

conditions between brackish and estuarine waters (from the lowest zone to the middle

zone of the gradient) and between estuarine and freshwater (from the middle zone to

the upper zone).Although the ecotone analysis was not determined at the lowest zone

(RE0) because of the unavailable fish data at this region prior to 1996, it is speculated

that the ecotone periphery between RE0 and RE1 is most likely to exist due to the

sharp transition between polyhaline and mesohaline waters (Weinstein et al. 1980)

between the two regions. At the lower zone, the sharp changes in DO and water

salinity between RE1 and RE2 apparently contributed to the formation of the ecotone

periphery (the narrow transition) that was highly detected between these two regions.

Salinity was not a major factor influencing formations of the ecotone peripheries at the

upper zone of the gradient where water condition becomes the tidal fresh (salinity

<0.5ppt; Winemiller and Leslie 1992).

At the upper gradient portion, the ecotone periphery frequently detected between

RE11 and RE12 was most likely influenced by the abrupt declines in DO and water

temperature from RE11 to RE12 in most months, the decreased abundance of Trapa

natans and Vallisneria americana from RE11 to RE12, and the dam constructions

54

above RE12 and in close proximity. Along the gradient, SAV occupied approximately

8% of the river bottom area, and 18% of the shallow water areas (≤3 m deep) was

available for SAV to colonize (Nieder et al. 2004). Below RE12, higher abundance of

the two SAV species, especially from RE9 to RE11, created both food sources and

shelters for inshore fish and other benthic animals in the shallow habitats (Abood and

Metzger 1996, Noordhuis et al. 2002, Strayer et al. 2004). In contrast, the dams

existing around and above RE12 alter flow regimes by reducing peak discharged

downstream and meting out flow over time to maintain minimum depths for

navigation (Lumia 1998). These consequences tend to affect distributions of fish

assemblages to the downriver portion (Abood and Metzger 1996, Schmidt and Cooper

1996).

While abrupt changes in both fish species composition and habitat conditions

concomitantly reflected the formations of ecotone peripheries between the particular

regions in the lower and upper zones, the ecotones were mainly detected within the

lower-middle zone of the gradient. The shift in locations of these ecotones, but still

within the gradient range from RE2 to RE5, tended to relate with the changes in

salinity regime, tidal current, and freshwater runoff seasonally or yearly (Weinstein et

al. 1980, Stanne et al. 1996, Attrill and Rundle 2002, Jaureguizar et al. 2003). Most of

the ecotones detected in the lower-middle zone had the breadth covering one or two

important nursery areas (i.e., Tappan Zee—RE2 and Haverstraw Bay—RE3) with low

salinity (1.0–5.1 ppt) serving diverse fish species (Stanne et al. 1996, Daniels et al.

2005) from marine (e.g., bay anchovy—Anchoa mitchilli and Atlantic menhaden—

Brevoortia tyrannus), estuarine (e.g., hogchoker—Trinectes maculates and banded

killifish—Fundulus diaphanus diaphanus), and freshwater (e.g., green sunfish—

Lepomis cyanellus and brown bullhead—Ameiurus nebulosus) assemblages.

55

However, in most time periods of the study, only one ecotone with the breadth

from RE2 to RE4 acted as an aggregator for most fish species, while the other

ecotones acted as soft or hard barriers for most of the Hudson fish (Table 2.1).

Furthermore, functional patterns of these ecotones on fish species assemblages were

not clearly different among seasons. From the comparison, while certain fish species

from marine, estuarine, and freshwater assemblages were usually observed in most

ecotones, total numbers of the fish species from the mixed assemblages in these

ecotones were intermediate or lower than those from the marine-estuarine or

freshwater assemblage observed in the adjacent habitat downstream or upstream,

respectively. Additionally, the abundance (numbers of individuals) of most fish

species was relatively low in the ecotones. These reflected complex distributions of

the Hudson fish assemblages and their variety of life histories (Hurst et al. 2004,

Daniels et al. 2005) responding to changing habitat conditions in and around the

ecotones on the gradient (Kupschus and Tremain 2001, Able 2005).

In an estuary, distributions of fish assemblages highly vary over short and long-

term periods due to: 1) complex life histories of fish and their habitat preferences that

may widely vary among life history stages (Able and Fahay 1998), 2) influences of the

recruitment dynamics of individual species (Hurst et al. 2004) and their differential

utilization of specific portions of the estuary (Weinstein et al. 1980), and 3) the

behaviors of fish that are inadequately understood (Able 2005). Along the Hudson

River estuary gradient, distributions of some fish species are restricted by their narrow

salinity tolerances, whereas distributions of other fish species with broader salinity

tolerances may be limited by other habitat variables that are correlated with salinity

concentration (Winemiller and Leslie 1992), e.g., tide, freshwater flow, and water

temperature (Vernberg and Vernberg 1976).

56

Influencing by river discharge and salinity concentration, young marine fish in the

Hudson system generally move upriver on the incoming flow of dense saline water

near the river bottom, whereas the young anadromous fish travel downriver on the

outgoing freshwater nearer the surface (Stanne et al. 1996). These fish assemblages

generally seek nursery areas with low salinity in the lower-middle zone of the tidal

river to temporarily spend their first year of life here and eventually migrate to the

Atlantic Ocean (Stanne et al. 1996, Able and Fahay 1998) as juveniles or adults

(Weinstein et al. 1980). For freshwater fish, some species may periodically move

farther downstream during high freshwater runoff with low salinity, while most

freshwater species are residents in the tidal fresh regions. Conclusively, patterns of

fish use of estuarine habitats in the lower-middle zone of Hudson is apparently a

continuum (Attrill and Rundle 2002, Able 2005); some species have obligated life

history stages (residents or diadromous) in the estuary, others are estuarine

opportunists, and still others are simply strays that occasionally find their ways into

the estuary (Able 2005).

The variations in freshwater flow, DO, salinity, and water temperature from month

to month also affected fish distributions (e.g., Gilmore 1995, Marshall and Elliott

1998, Whitfield 1999, Martino and Able 2003) in and around the Hudson ecotones.

During low freshwater flow with the increase salinity along the gradient, distribution

ranges of some marine and estuarine fish species show spatial change as they migrate

farther upstream (Attrill and Rundle 2002) beyond the brackish-estuarine ecotones (at

the lower-middle zone) leading to low species richness and abundance of fish in these

ecotones. In addition, both species richness and abundance of fish are mostly likely to

decrease at the freshwater-estuarine ecotones (at the middle-upper zone) of the

gradient as freshwater species became intolerant of increased salinity, while only a

few estuarine species can tolerate low salinity (Rundle at al. 1998).

57

Focusing at the lower-middle zone from RE2 to RE5 (where the ecotones were

highly detected), DO and water temperature showed gradual changes within the range

from RE2 to RE5 in a few months, but sharply fluctuated among these regions in most

months. Wells and Young (1992) found that the seasonal cycle accounted for most

(>99%) of the variation in water temperature in the Hudson system, whereas patterns

of changes in DO reflected an inverse relationship with water temperature (Mancroni

et al. 1992). Unlike DO and water temperature, salinity showed consistent pattern of

abrupt decrease from RE2 to RE5 in all study months, except in April. The synergy of

both gradual and sharp changes among these habitat variables plays important role in

structuring fish species assemblages (Whitfield 1999, Marshall and Elliott 1998) in

and around the Hudson ecotones. The bottom substrates are another factor influencing

distributions of fish species assemblages in an estuary at some extent (Weinstein et al.

1980). However, this habitat variable may play less influence on the distributions of

fish assemblages along the tidal Hudson River where the bottom of water is relatively

homogeneous. More than half of the gradient length (from RE1 to RE9) comprises a

major combination of mud and sandy mud, while sand and muddy sand were major

combined substrates at the remaining gradient length (RE10–RE12).

Since 1991, the zebra mussel invasion has been observed in the Hudson system

(Strayer et al. 1996) and its effect might be an additional factor shaping ecotone

functions on the Hudson fish species assemblages afterward. With their high filtration

rate of 6 m3/m2.day (Strayer et al. 1996), the zebra mussel populations are suspected to

bring about a selective decline in the density of phytoplankton and small zooplankton

(rotifers, tintinnids, and copepod nauplii) and a loss of consumers that depended on

phytoplankton and other edible particles in deep water of the Hudson (Pace et al.

1998, Strayer et al. 1999). In contrast, the filtration rates of the zebra mussel

populations increased sunlight penetration through surface of shallow water of the

58

Hudson enhancing the growth of aquatic plants and the invertebrates associated with

them (Strayer et al. 1999). These consequences interfered distribution patterns of the

Hudson fish assemblages and caused 28% of the median decrease in abundance of

open water fish species (e.g., Also spp.) due to the decline of food source in deep

water, but caused 97% of the median increase in abundance of littoral fish (e.g.,

sunfishes) due to the increase of food source in shallow water (Strayer et al. 2004). As

a result, both open-water and inshore fish assemblages move downstream and

upstream, respectively, away from the middle portion of the Hudson River estuary

gradient.

In conclusion, the formations of Hudson ecotones showed both changes in location

and structure spatially and temporally. In some time periods of detection, the ecotones

changed in their breadth (from broader to narrower or vice versa) or considerably

declined to be just ecotone peripheries. The ecotones were highly detected within the

lower–middle zone (RE2–RE5) of the Hudson River estuary gradient, whereas the

ecotone peripheries alone were highly detected at the lower zone (between RE1 and

RE2) and the upper zone (between RE11 and RE12) of the gradient. Abrupt changes

in one or more habitat variables and the peak turnover rates in fish species

composition were coincidently observed at the ecotone peripheries. Functional

patterns of the Hudson ecotones were not distinctly different across seasons. The

ecotone with the breadth from RE2 to RE4 was most likely to be the optimal zone for

distributions of diverse fish species assemblages, whereas the other ecotones acted as

barriers for most of the Hudson fish species. Likewise, the ecotone peripheries on the

gradient mainly inhibited distributions of most fish species between adjacent regions

with high contrast of habitat changes. Relatively low abundances of most fish species

observed in the Hudson ecotones were apparently related with complex migratory

behaviors and life histories of fish responding to variations in the estuarine conditions

59

in space and time. The effect of zebra mussel invasion in the Hudson system might be

an additional factor shaping the ecotone functions on distribution patterns of the

Hudson fish assemblages.

To explore influences of changing habitat in and around aquatic ecotones on

distributions of fish species assemblages in space and time, an explicitly abundance

exchange model was developed to quantify distribution patterns of target fish species

in association with their habitat preferences across lentic-lotic transitions on the two

freshwater gradients located on the southeastern coast of Lake Ontario. The

methodology of study, results, and discussion are presented in Chapter 3.

60

*CHAPTER THREE

An abundance exchange model of fish assemblage response to changing

habitat along embayment-stream gradients of Lake Ontario, New York

ABSTRACT

A spatially explicit abundance exchange model (AEM) was developed to predict

distribution patterns of five fish species in relation to their population characteristics

and habitat preferences along two embayment-stream gradients associated with Lake

Ontario, New York. The five fish species were yellow perch (Perca flavescens),

bluegill (Lepomis macrochirus), logperch (Percina caprodes), bluntnose minnow

(Pimephales notatus), and fantail darter (Etheostoma flabellare). Preference indexes

of each target fish species for water depth, water temperature, current velocity, cover

types (aquatic plants, algae, and woody debris), and bottom substrates (mud, sand,

gravel-cobble, and rock-bedrock) were estimated from the field observations, and

these were used to compute habitat preference (HP) of the associated fish species.

Fish HP was a key variable in the AEM to quantify abundance exchange of an

associated fish species among habitats on each study gradient. According to the

results, the AEM efficiently determined local distribution ranges of the fish species on

one study gradient. Results from the model validation showed that the AEM with its

estimated parameters was able to quantify most of the fish species distributions on the

second gradient. Overall, the AEM is rigorous for quantifying the distribution patterns

of the target species along the changing habitat gradients.

*Singkran, N., An abundance exchange model of fish assemblage response to changing habitat along embayment–stream gradients of Lake Ontario, New York, Ecol. Model. (2006), doi:10.1016/j.ecolmodel.2006.10.012

61

With its flexible structure that is applicable for array functions and differential

equations from both static and dynamic components, the AEM can be modified to

determine patterns of organism distribution in complex systems with different

environments and geography.

INTRODUCTION

The influence of changing landscapes and environments on organism distribution

across ecotones (transition zones) of contrasting habitats (e.g., grassland-forest) has

been increasingly studied in terrestrial ecosystems (e.g., Clements 1905, Leopold

1933, Odum 1971, Risser 1990, Johnson et al. 1992, Lidicker 1999, Fortin et al. 2000,

Harding 2002, Cadenasso et al. 2003, Ries et al. 2004), whereas fewer studies of this

similar research type have been conducted in aquatic ecosystems (Winemiller and

Leslie 1992, Willis and Magnuson 2000, Martino and Able 2003, Miller and Sadro

2003). Additionally, although distribution of organisms is a key mechanism

underlying the population dynamics (Nisbet and Gurney 1982, Tilman and Kareiva

1997), few empirical studies (Reyes et al. 1994, Gaff et al. 2000, Kupschus 2003) have

quantitatively determined organism distribution in association with both changes in

habitat and population characteristics (birth, death, and migration) over space and

time.

In this study, an abundance exchange model (AEM) was built to quantify the

distribution of fish species assemblages in relation to both population characteristics

and habitat preferences of fish across lentic-lotic transitions on two embayment-stream

gradients associated with Lake Ontario, New York. For purposes of this study, the

term “transition” referred to both abrupt and gradual changes in spatial habitat

heterogeneity between lentic (embayment) and lotic (stream) systems on the study

62

gradients. As a result, this transition is often distinctly observed at the lentic-lotic

periphery, and it may gradually or abruptly occur across the zone of transition between

the two systems.

The objectives of developing the spatially explicit AEM were to determine how

changing habitat influences distribution patterns of fish species assemblages along the

embayment-stream gradients and how population characteristics shape the size of fish

population over time. To test if the AEM could efficiently capture distribution

patterns of fish across different habitat conditions, five fish species, yellow perch

(Perca flavescens), bluegill (Lepomis macrochirus), logperch (Percina caprodes),

bluntnose minnow (Pimephales notatus), and fantail darter (Etheostoma flabellare)

were selected for modeling. Yellow perch and bluegill commonly distributed in the

embayment-downstream parts on both study gradients. Logperch was mainly

observed at the downstream-middle part on one study gradient, but this species was

not observed on the second gradient. Bluntnose minnow commonly distributed

throughout both study gradients, while fantail darter mainly distributed at the upstream

parts of both study gradients.

In aquatic systems, high variability in hydrology (Wiens 2002) and climate-linked

seasonal changes can alter other aquatic habitat variables (e.g., water temperature and

growth of aquatic plants and algae). As a result, the migration rate of fish along a

gradient is most likely to reflect fish habitat preference within a dynamic template.

Additionally, it has been considered that each fish species has a specific range of

distribution (e.g., Morrison et al. 1985, Guay 2000), the species stays within its range

but will variably distribute among habitats therein. This conforms to nonlinear

patterns of organism distribution along changing habitat gradients (Whittaker 1975,

Lewis 1997). In this study, both dynamic habitat variables (water depth, water

temperature, current velocity, aquatic plants, algae, and woody debris) and static

63

habitat variables (the bottom substrates, i.e., mud, sand, gravel-cobble, and rock-

bedrock) were observed along the study gradients and used to estimate fish habitat

preferences.

To test the hypothesis that distributions of fish across lentic-lotic transitions on the

study gradients are based on fish preferences for changing habitats within their local

ranges, habitat preferences of the target fish species at two life stages (0+ years and 1+

years) were the key variable in the AEM to quantify distribution patterns of the

associated fish species along each study gradient. STELLA® 7.0.3 Research (High

Performance Systems, Inc. 2002) was used to run the AEM. Since it was developed

by Barry Richmond in 1985, STELLA has been widely used by scientists across fields

of study. This software is capable of simulating many components and interactions of

complex systems, and is fast and efficient at calculating a diversity of array functions

and differential equations (e.g., Reyes et al. 1994, Pan and Raynal 1995, Marín 1997,

Gottlieb 1998, Ford 1999, Krivtsov et al. 2000, Dew 2001, Gertseva et al. 2003,

Gertseva et al. 2004, Kilpatrick et al. 2004, McCausland et al. 2006).

Besides model development and prediction, model validation, an important

modeling process (Haefner 1996, Olden et al. 2002, Ottaviani et al. 2004), was the

other objective of this study. Although numerous modeling approaches (e.g.,

statistics, system dynamics, and geographic information system (GIS)) have been used

for modeling species distribution, few of the models were further validated after the

predictions were completed (Olden et al. 2002). Model validation is crucial for

evaluating the accuracy of the model’s predictions and future applicability (Olden et

al. 2002, Ottaviani et al. 2004). In addition, proper ways of validating the model must

be implemented. To verify that the model prediction is robust and unbiased and has

minimal errors (i.e., over-fitting data, overrating model performance, or

underestimating the error for future application), one must validate the model with an

64

independent data set that has not been previously used to estimate the model’s

parameters (Guisan and Zimmermann 2000, Elith and Burgman 2002, Fielding 2002,

Olden et al. 2002, Ottaviani et al. 2004).

In this study, to evaluate if the AEM is robust and efficient for predicting the

distribution of the target fish species in similar geographic and hydrologic systems, the

model with its estimated parameters to quantify the distribution pattern of each target

species on one study gradient was validated against the observed data of the same fish

species on the second study gradient located in close proximity. Accuracy of the

model prediction was determined using the chi-square (χ2) statistic to test the

goodness of fit between predicted and observed percentage abundances of fish along

each study gradient. The important modeling processes and the model performance

presented in this study may contribute to applications for better predicting the

distribution of fish species assemblages across different habitats. This knowledge will

help resource managers evaluate and maximize healthy habitats among different

aquatic ecosystems to support diverse fish assemblages at both local and regional

scales.

STUDY GRADIENTS AND FIELD OBSERVATIONS

The empirical studies were conducted across lentic-lotic transitions on two

embayment-stream gradients, Floodwood (FL) and Sterling (ST) gradients, located on

the southeastern coast of Lake Ontario, New York (Figure 3.1). The FL gradient

comprises Floodwood Pond and Sandy Creek. The total watershed area for the FL

gradient was 671.82 km2, and the surface area of Floodwood Pond alone was 0.078

km2. Sandy Creek ranges in width from 6 m (upstream) to more than 50 m (entering

Floodwood Pond) and has a mean annual discharge of 7.8 m3/s (37-year record—

65

USGS gage 04250750 above study area). The ST gradient includes Sterling Pond and

Sterling Creek. The total watershed area associated with the ST gradient was 210.76

km2, and the surface area of Sterling Pond alone was 0.38 km2. Sterling Creek ranges

in width from 3 m (upstream) to more than 20 m (entering Sterling Pond) with a mean

annual discharge of 1.9 m3/s (37-year record—USGS gage 04232100 above study

area).

Figure 3.1. The Floodwood (FL) and Sterling (ST) gradients with eight sampling stations (stars, FL; circles, ST) for monthly collecting of target fish and habitat variables during June–August 2002. The fish and habitat collections were repeated at 12 stations (circles) along the FL gradient during June–August 2004.

A Garmin Etrex Venture global positioning system was used to mark locations of

sampling stations on both gradients. In the summer months (June–August) of 2002,

eight transect stations on the FL gradient (Figure 3.1: stars) were selected (between

43°43'16" N, 76°12'10" W and 43°48'47" N, 76°4'31" W) using a stratified random

sampling method to cover all habitat types. The length of the entire sampling gradient

was about 18.6 km. The same stratified random method was used to select eight

transect stations (between 43°20'34.5" N, 76°41'54.7" W and 43°16'51.6" N,

66

76°37'37.6" W) on the ST gradient (Figure 3.1). The length of the entire sampling

gradient was about 16.3 km.

Abundance of the five selected fish species (yellow perch, bluegill, logperch,

bluntnose minnow, and fantail darter) was observed once a month along the FL and

ST gradients during June–August 2002. Fish were randomly sampled for 15 minutes

per station at all FL and ST stations using electrofishing gear (the power supply to

generate electricity was a 3500-W generator and a transformer with DC voltage output

controller of 250–350 V). Juveniles and adults of each target species caught at each

station were enumerated and their total length was measured to the nearest millimeter.

In the same summer months, the habitat variables were measured once a month in

all stations where the fish species were collected as follows. Water depth (m) and

water temperature (°C) were randomly measured five times in each sampling station.

Current velocity (m/s) was randomly measured five times within each station using a

Marsh-McBirney 2000-model (Bovee 1982). Bottom substrates (%) were visually

observed once a month in all sampling fish stations on each study gradient; at the first

observation, substrate grain sizes (diameter) were measured and classified into four

types modified from Pusey et al. (1993): mud (<1 mm diameter), sand (1–16 mm

diameter), gravel-cobble (16–128 mm diameter), and rock-bedrock (>128 mm

diameter). The average percentage of each substrate type and average abundance of

each cover type was calculated from their 10 random observations in each station.

The abundance of aquatic plants, algae, and woody debris was coded on a scale of 0–

3: 0 = absent, 1 = low, 2 = common, and 3 = high.

In the summer months (June–August) of 2004, the fish and habitat collections

were repeated at 12 sampling stations (Figure 3.1: circles) along the FL gradient. The

12 sampling stations were located between 43°43'20.6" N, 76°12'13.3" W and

43°51'11.7" N, 75°56'44" W, the total sampling gradient length was 28.2 km. The

67

target fish species were sampled monthly at a total of 60 locations within the 12

stations along the FL gradient. With a fine scale of sampling, juveniles and adults of

each target species were collected for 10 minutes in each of the 60 locations using the

same fishing gear and sampling protocols as in the summer of 2002. Individuals of

each target fish species collected from all locations within each station were

accumulated, enumerated, and measured for total length to the nearest millimeter.

In the same summer months of 2004, the same habitat variables were sampled

monthly five times in each of the 60 locations within the 12 stations where the target

fish species were collected using the same methods and instruments as in the summer

of 2002. As a result, 900 samples for each habitat variable were collected along the

FL gradient during the three summer months of 2004. Mean values of each habitat

variable for the 180 samples along the FL gradient were then obtained by dividing the

total 900 samples by the five times of collection in each location (i.e., 900/5 = 180) in

summer 2004. The mean habitat variables and abundance data for the target fish

species collected on the FL gradient in the 2004 summer were used to build the AEM

and to estimate and calibrate the model’s parameters (i.e., a preference index of the

target fish species for each of the habitat variables). The model with its estimated

parameters was then validated against the fish and habitat data collected on the ST

gradient over the three summer months of 2002 (see detail in next section).

MODELING FISH DISTRIBUTIONS

ALONG EMBAYMENT-STREAM GRADIENTS

Applications of diffusion and random walk theories have been accepted as

powerful methods for spatially modeling organism dispersion (Nisbet and Gurney

1982, Reyes et al. 1994, Turchin 1998, Fagan et al. 1999, Kot 2001, Okubo and Levin

68

2001, Sparrevohn et al. 2002). In homogeneous habitat conditions, population

distribution of a given species in relation to certain habitat variables on a one-

dimensional gradient can be described as a Gaussian curve (Whittaker 1975, Lewis

1997). This normal distribution is the main idea proposed in Fick’s law of

particle/organism dispersion on a one-dimensional system expressed as (Okubo and

Levin 2001)

,22

2exp

2π0),( ⎟

⎟

⎠

⎞

⎜⎜

⎝

⎛−=

σ

xσ

ntxP (1)

where P(x, t) is the total individuals of a given species at location x and time t, n0 is the

initial number of individuals of a given species, and σ2 is the variance of the normal

distribution of a given species. It has been statistically proven that σ equals Dt2

(e.g., Fischer et al. 1979), where D is a constant dispersion rate (distance2/time) of the

population, and x is distance. Equation (1) after being rewritten becomes

⎟⎟

⎠

⎞

⎜⎜

⎝

⎛−

π=

Dtx

Dtn

txP4

2exp

40),( (2)

However, using a constant D in equation (2) may be unrealistic for describing

species distribution across heterogeneous habitat conditions (Johnson et al. 1992,

Fortin et al. 2000, Okubo and Levin 2001). In this study, it was assumed that the

migration rate of each target fish species varied depending on the fish preference for

changing habitat conditions across the lentic-lotic transition on each study gradient.

This assumption is based on the non-Fick’s law of dispersion (Johnson et al. 1992),

69

stating that although a given species acts as a random walker, it displays biased

decision to distribute to its optimal habitat.

Abundance exchange model development

A system dynamics approach, a rigorous modeling method that enables users to

build formal computer simulations of complex dynamic systems (Ford 1999, Sterman

2000) was used for building an abundance exchange model (AEM) to determine

distribution patterns of the target fish species along the two embayment-stream

gradients (Figure 3.1). The AEM includes both within-habitat and across-habitat

feedbacks to integrate population characteristics and habitat changes. The model

parameters were estimated and calibrated based on the fish and habitat data collected

along the FL gradient over the three summer months (June–August) of 2004. To

construct the AEM for each target fish species with respect to the species preference

for changing habitat, two major assumptions were made. First, distribution of the

target fish species at two life stages (0+ years and 1+ years) within and across habitats

on each study gradient was the result of functional migrations (Lucas and Baras 2001)

of fish to seek preferred areas for refuge in winter (November–March) and growth

(feeding and spawning) in growing season (April–October). Second, since small and

shallow waters can be conceptualized as one-dimensional habitats (Skalski and

Gilliam 2000), the distribution of each target fish species along the shallow gradients

(<6 m deep) was considered on one dimension only (i.e., in a horizontal direction from

downstream to upstream or vice versa).

The conceptual diagram (Figure 3.2) shows how the distribution pattern of each

target fish species within and across habitats was quantified along each study gradient.

From the diagram, fish that successfully overwinter in habitat n will migrate from

wintering areas to growing areas for breeding, feeding, or refuge in the growing

70

season. Mature fish will spawn at a certain period of time and have offspring in the

same season. In the late growing season, young fish will grow up and aggregate with

the adults to overwinter in the wintering areas. The same pattern of fish migration

within habitats is assumed to occur in habitat n–1 and n+1. Concurrently, migration of

a target fish species at both life stages also occurs among adjacent habitats.

Figure 3.2. A conceptual diagram of fish distribution within and among habitats. Migration rates of fish at two life stages in the growing season (yk, 0+ yr old; ak, 1+ yr old) and for all fish in winter (k) among habitats n, n–1, and n+1 are dependent on fish habitat preference in the growing season (yHP, 0+ yr old; aHP, 1+ yr old) and in winter (HP), respectively. The three graphs below show the estimated preference indexes (Pi) of the three example habitat variables that were used to compute the geometric mean HP for the target fish species at each life stage.

The migration rate (k) of each fish species varies in relation to the species habitat

preference (HP) along each study gradient. For instance, if a fish species preferred

71

habitat n–1 to habitat n, more individuals of that species would migrate to habitat n–1,

rather than stay in habitat n. In contrast, if suitable habitat conditions are not available

in habitat n–1 and n+1, the fish species’ populations are most likely to distribute only

within habitat n. In the AEM, the k of each target fish species among adjacent habitats

was derived from the fish species HP, such that k = (1–HP) over time t of the

simulation. The HP of each target species at each life stage is estimated using a

geometric mean model, equation (3), as follows.

NN

xx

/1

1PiHP ⎟⎟

⎠

⎞⎜⎜⎝

⎛= ∏

=, (3)

where Pix is a fish preference index (Degraaf and Bain, 1986) for each habitat variable

x. The model in equation (3) was employed from the habitat suitability index method

of Bovee (1986), which has been widely applied for determining health of fish habitats

(e.g., Pajak and Neves 1987, Reyes et al. 1994, Guay et al. 2000) and characterizing

species distributions (O’Connor 2002). In equation (3), Pix, is expressed as

)/()/(Pi

EtEjFtFj

x = (4)

where Fj is the number of individuals of a target fish species observed at an intensity

interval j of a habitat variable x, Ft is the total number of individuals of a target fish

species observed from all intensity intervals j of a habitat variable x, Ej is the number

of observations for a habitat variable x at an intensity interval j, and Et is the total

number of observations for a habitat variable x from all intensity intervals j.

To obtain the index between 0 and 1, the Pix on each interval j is normalized (i.e.,

Pix,j/Pix,jmax). Examples functions of the Piwater depth, Picurrent velocity, and Piwater temperature of

72

bluntnose minnow at both life stages in the growing season in habitat n–1 were shown

in Figure 3.2. The fish HP is then computed as a product of the fish species Pix for all

habitat variables N. To derive the HP value between 0 and 1 over time t of the

simulation, the HP, equation (3), is normalized with the power of 1/N. A summarized

equation incorporating both population characteristics and HP to simulate the AEM

for each target fish species along each study gradient is

{ } { }

),()(HPHP

HP)(HPHP

HP

)(HPHP

HP)(HPHP

HP

)()()()()(

n11

1

11

1

11

1

11

1

11111111

PrYykzyy

yAakzaa

a

Yykzyy

yAakzaa

a

YfYykAakYykAakdt

dP

dnnnn

nnn

nn

n

nnnn

nnn

nn

n

nn,nnn,nnn,nnn,nnn

×−⎭⎬⎫

⎩⎨⎧

××⎟⎟⎠

⎞⎜⎜⎝

⎛++

+××⎟⎟⎠

⎞⎜⎜⎝

⎛++

−

⎭⎬⎫

⎩⎨⎧

××⎟⎟⎠

⎞⎜⎜⎝

⎛++

+××⎟⎟⎠

⎞⎜⎜⎝

⎛++

−

+×+×+×+×=

+−

+

+−

+

+−

−

+−

−

++++−−−−

(5)

where Pn is the population of a target fish species in habitat n, and akn and ykn are

migration rates of 1+ year-old fish (A) and 0+ year-old fish (Y), respectively, out of

habitat n. Similarly, akn–1,n and akn+1,n are migration rates of A from habitat n–1 and

n+1, respectively, to habitat n, and ykn–1,n and ykn+1,n are migration rates of Y from

habitat n–1 and n+1, respectively, to habitat n. In this equation, akn equals (1–aHPn)

over time t. The rest of the migration rates are computed in the same way. The aHPn,

aHPn–1, and aHPn+1 are the HPs of A in habitats n, n–1, and n+1, respectively.

Likewise, the yHPn, yHPn–1, and yHPn+1 are the HPs of Y in habitats n, n–1, and n+1,

respectively.

A small number z (e.g., z ≤ 10–6) is added to the model to avoid zero denominators

in case the species HP in any habitat becomes zero at a certain time t of simulation.

The birth of a target species in habitat n, for example, is represented by a logistic

growth model, f(Yn), where f(Yn) equals rb × An × (1–Yn/Cy). That is, the birth of Yn

depends on An abundance, birth fraction rb, and carrying capacity Cy of Yn. The

73

natural mortality of An in the growing season is ignored, but it is considered in winter,

and it is computed by multiplying the death fraction rd of fish over the winter months

by the fish population that survives winter in habitat n.

Model simulation and validation

The model was initiated by assigning individuals of all target fish species to

certain FL stations (Table 3.1) where the fish species were observed in high

abundance (>70% of the total observed individuals along the FL gradient) in August

2004. Because of insufficient information to estimate carrying capacity Cy for each

target species at each station, the same value of Cy for each target species was assigned

to all stations in the growing season (Table 3.1). Ranges of birth fraction rb in the

growing season and death fraction rd in winter for each target species were assigned to

all stations (Table 3.1).

Table 3.1. Initial values of the population variables (Cy, rb, and rd) and individuals of each target fish species for simulating the abundance exchange model.

Yellow perch 25,000 FL1, ST1 6,000 0.40–0.60 0.05–0.10 (Perca flavescens )Bluegill 55,000 FL1, ST1 8,000 0.40–0.60 0.05–0.10 (Lepomis macrochirus )Logperch5 6,000 FL5 2,000 0.22–0.24 0.08–0.10 (Percina caprodes )Bluntnose minnow 3,000 FL5, FL12, 2,500 0.30–0.35 0.12–0.15 (Pimephales notatus ) ST7Fantail darter 8,000 FL12, ST8 2,500 0.30–0.40 0.08–0.09 (Etheostoma flabellare )

r b3 r d

4Fish species Initial no. of individuals Station1 C y

2

1The initial number of individuals of each target species was assigned to certain stations on the Floodwood (FL) and Sterling (ST) gradients to start the simulation on December 1st. Unassigned stations had zero individuals of fish at the starting time of the simulation. 2The same value of carrying capacity Cy (individuals) for each target species was assigned to all FL and ST stations in the growing season.

74

Table 3.1 (Continued).

3The same range of birth fraction rb (month-1) for each target species was assigned to all FL and ST stations to generate series of uniformly distributed random values of rb in the growing season. 4The same range of death fraction rd (month-1) for each target species was assigned to all FL and ST stations to generate series of uniformly distributed random values of rd in winter. 5Distribution pattern of logperch was not modeled on the ST gradient.

From these initial conditions, the model generates series of uniformly distributed

random values of population variables for each fish species in the associated seasons.

To explore how the model would behave in a system in which a sharp change in each

population variable might occur for some reason (e.g., food increase/decline,

predator), ±50% changes in value of each population variable were assigned one at a

time at all FL stations to rerun the model.

Mean values of each habitat variable from the 180 samples along the FL gradient

were classified into intervals using natural breaks (Jenks), one of the classification

types in ArcGIS 9.1. This method is useful for a data set that contains relatively big

jumps in the data values. Thus, for each habitat variable, similar data values were

grouped together in the same interval, and the differences between intervals were

maximized. A fish species Pi for each habitat variable at each classified interval was

then estimated using equation (4) (Table 3.2).

Series of IF-THEN-ELSE statements (logical functions) were used to incorporate

the species Pi at each interval for all habitat variables (Table 3.2) into the AEM for

computing the species HP, equation (3). The species’ migration rate k in the model

was then converted from the species HP and used to quantify abundance exchange of

that species among FL stations in the growing season.

75

Table 3.2. Preference index (Pi) for each habitat variable of five modeled fish species at two life stages (0+yr and 1+yr old) estimated from the fish and habitat data observed along the Floodwood gradient over the three summer months (June–August) of 2004.

0+yr 1+yr 0+yr 1+yr 0+yr 1+yr 0+yr 1+yr 0+yr 1+yr

Water depth (m)0.10-0.3 0.10 0.16 0.17 0.10 1.00 1.00 0.62 1.00 1.00 0.730.31-0.6 0.03 0.00 0.17 0.02 0.17 0.29 0.75 0.43 0.73 1.000.61-1.0 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.61 0.49 0.771.01-4.0 1.00 1.00 1.00 1.00 0.37 0.09 0.08 0.22 0.00 0.00

Current velocity (m/s)0.03-0.06 1.00 1.00 1.00 1.00 0.19 0.00 0.05 0.12 0.00 0.000.07-0.12 0.18 0.46 0.59 0.90 1.00 1.00 1.00 1.00 0.92 0.680.13-0.24 0.04 0.00 0.07 0.06 0.66 0.91 0.48 0.41 0.37 0.700.25-0.36 0.04 0.00 0.00 0.06 0.76 0.86 0.64 0.60 1.00 1.000.37-0.48 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.13 0.42

Water temperature (°C)16.1-18 0.71 0.68 0.90 0.57 0.51 0.58 1.00 0.45 0.82 0.7418.1-20 1.00 0.45 1.00 0.27 1.00 1.00 0.59 0.73 1.00 1.0020.1-22 0.27 0.00 0.63 0.32 0.00 0.00 0.49 0.27 0.18 0.6622.1-24 0.30 0.27 0.00 0.28 0.51 0.00 0.19 0.10 0.05 0.1724.1-27 0.78 1.00 0.90 1.00 0.57 0.78 0.56 1.00 0.02 0.16

Aquatic plants1

0.01-0.3 0.00 0.00 0.00 0.00 0.00 0.00 0.65 0.64 1.00 1.000.31-0.6 0.01 0.00 0.04 0.08 0.11 0.12 1.00 0.39 0.36 0.680.61-1.0 0.54 0.94 0.31 0.82 1.00 1.00 0.59 1.00 0.01 0.031.01-1.5 0.33 0.25 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.001.51-3.0 1.00 1.00 1.00 0.99 0.00 0.00 0.00 0.31 0.00 0.00

Algae1

0.00-0.2 1.00 1.00 0.92 1.00 0.33 0.09 0.15 0.30 0.00 0.000.21-0.6 0.40 0.26 0.00 0.47 0.23 0.00 0.04 0.22 0.08 0.050.61-1.2 0.09 0.00 0.00 0.07 1.00 1.00 1.00 0.37 0.01 0.101.21-2.0 0.28 0.14 1.00 0.11 0.00 0.00 0.13 0.15 0.06 0.262.01-3.0 0.06 0.08 0.24 0.05 0.42 0.33 0.87 1.00 1.00 1.00

Woody debris1

0.01-0.3 0.08 0.00 0.00 0.08 1.00 1.00 0.35 0.93 1.00 1.000.31-0.6 0.22 0.22 0.54 0.33 0.46 0.62 1.00 1.00 0.48 0.910.61-1.0 0.88 0.52 0.17 0.58 0.28 0.00 0.28 0.23 0.02 0.181.01-1.5 1.00 1.00 1.00 1.00 0.00 0.00 0.08 0.17 0.00 0.00

Mud (%)0.0-5 0.04 0.08 0.11 0.04 0.51 0.52 0.62 1.00 1.00 1.005.1-15 0.08 0.00 0.24 0.06 0.56 1.00 1.00 0.34 0.01 0.07

15.1-30 0.25 0.47 0.29 1.00 1.00 0.24 0.13 0.41 0.00 0.0070.1-100 1.00 1.00 1.00 0.54 0.00 0.00 0.05 0.21 0.00 0.00

Fantail darterPi

Yellow perch Bluegill Logperch Bluntnose minnowHabitat variable

76

Table 3.2 (Continued).

0+yr 1+yr 0+yr 1+yr 0+yr 1+yr 0+yr 1+yr 0+yr 1+yr

Sand (%)0.0-5 0.00 0.00 0.00 0.00 0.00 0.00 0.40 0.29 0.06 0.29

5.1-15 0.19 0.28 0.33 0.17 0.80 1.00 1.00 1.00 1.00 1.0015.1-30 1.00 1.00 1.00 0.30 0.00 0.00 0.36 0.30 0.12 0.2030.1-50 0.27 0.58 0.32 1.00 1.00 0.26 0.16 0.35 0.00 0.00

Gravel-cobble (%)0.0-5 1.00 0.99 1.00 0.38 0.00 0.00 0.22 0.16 0.03 0.19

5.1-15 0.56 0.77 0.86 0.68 0.00 0.00 0.26 0.14 0.07 0.8215.1-30 0.28 0.42 0.47 1.00 0.22 0.19 0.98 0.44 1.00 1.0030.1-50 0.52 1.00 0.86 0.54 1.00 1.00 1.00 1.00 0.07 0.17

Rock-bedrock (%)0.0-5 1.00 1.00 1.00 0.54 0.00 0.00 0.05 0.12 0.00 0.00

5.1-15 0.25 0.47 0.29 1.00 0.77 0.16 0.11 0.24 0.00 0.0030.1-50 0.00 0.00 0.00 0.00 0.00 0.00 0.55 0.25 0.21 0.5450.1-75 0.12 0.16 0.24 0.10 1.00 1.00 1.00 1.00 1.00 1.0075.1-100 0.00 0.00 0.06 0.00 0.00 0.00 0.24 0.21 0.07 0.67

Habitat variable

PiYellow perch Bluegill Logperch Bluntnose minnow Fantail darter

1Abundance of each cover type was numerically coded, 0 = absent, 1 = low, 2 = common, and 3 = high.

To simulate habitat change along the FL gradient in the growing months, series of

uniformly distributed random values of all dynamic habitat variables (except

substrates) were generated based on the observed values from the summer of 2004

(Figure 3.3). The distributed random values of each habitat variable along the FL

gradient in April and May were generated from the observed values in June. The

distributed random values of the habitat variables in September and October were

generated from the observed values in August. The substrate composition observed at

each station along the FL gradient (Figure 3.3) was assumed to be constant across

seasons (Hatzenbeler et al. 2000). Because the target fish species and habitat data

were not collected during the winter, the HPwinter of each target species was assumed to

77

be the same as the species HP in the late growing season (October) estimated by the

model.

Figure 3.3. Categorized mean values of all habitat variables (except substrate composition) observed monthly during June–August 2002 along the Floodwood gradient. Abundance of aquatic plants, algae, and woody debris was rated from absent (0) to high (3). Percentage composition of four substrate types was averaged over the three months along the gradient.

78

The model was validated against the fish and habitat data collected on the ST

gradient in the summer of 2002. The AEM with all estimated Pi for each species from

the FL gradient was used to predict the distribution pattern of the same species (except

logperch) along the ST gradient. Logperch was not modeled on the ST gradient

because this species was not observed at all ST stations over the three summer months

(June–August 2002). To simulate the AEM on the ST gradient, the same individuals

of the target species assigned to the FL stations were assigned to the particular ST

stations (where the target fish species accounted for more than 70% of the total

observed individuals in August 2002) (Table 3.1). The same ranges of Cy, rb, and rd of

each modeled species assigned to the FL stations were also applied to all ST stations

(Table 3.1). Series of uniformly distributed values of the habitat variables along the

ST gradient in the growing months were generated from the values observed on this

gradient during the summer of 2002 (Figure 3.4) based on the same assumption made

for those on the FL gradient.

STELLA® 7.0.3 Research (High Performance Systems, Inc. 2002) was used to run

the AEM for a long-term prediction (100 hundred years) on each study gradient

starting with month 0 on December 1st and ending at 1,200 months on November 30th.

The interval of time between calculations (dt) was set to a small value of 0.0625

months to avoid artifactual delays and dynamics during the software calculation. The

artifactual delays result from the fundamental conceptual distinction between stocks

(state variables) and flows. Stocks exist at a point in time, whereas flows exist

between points in time. Consequently, the two components cannot coexist at the same

instant in time. For example, during migration of fish from one habitat to another

habitat, it will take an “instant” for fish from the first habitat to actually arrive in the

second habitat.

79

Figure 3.4. Categorized mean values of all habitat variables (except substrate composition) observed monthly during June–August 2002 along the Sterling gradient. Abundance of aquatic plants, algae, and woody debris was rated from absent (0) to high (3). Percentage composition of four substrate types was averaged over the three months along the gradient.

80

The “instant” constitutes the software’s artifactual delay. As long as dt is defined

as a relatively small value, the artifactual delays are usually insignificant (High

Performance Systems, Inc. 2002). Avoiding artifactual dynamics is the second reason

for choosing a small dt value for running the model that generates a dynamic pattern

of behavior caused by the way the software is performing its calculations. If the dt

value is too large, the flow volumes will be too large, and pathological results are

generated. However, the two problems are unlikely to occur when running the model

with the dt at or below 0.25 (High Performance Systems, Inc. 2002).

The chi-square (χ2) statistic was used to test the goodness of fit between the mean

predicted abundance (%) on August 31st from the 100-year simulation and the

observed abundance (%) of the target fish species in the same month on each study

gradient.

RESULTS

Model prediction

Populations of the five modeled fish species showed seasonal patterns of

distribution along the FL gradient over the 100-year simulation following results for

yellow perch (Figure 3.5). The fish populations declined in winter as a result of winter

mortality and increased in the growing season owning to birth and growth of

offspring. Populations were relatively stable for a certain time period in the growing

season after the spawning period when no offspring were born. After each growing

season, the fish populations rapidly declined during the subsequent winter. As

expected, the fish population size increased when higher value of either Cy (Figure

3.5B) or rb (Figure 3.5C), or lower value of rd (Figure 3.5D) was assigned one at a

time for simulating the model.

81

Figure 3.5: Seasonal distribution pattern of yellow perch over the 100-year simulation (A). The fish population size exponentially decreased or increased for the first 10-year period of the simulation when changes of 50% of carrying capacity Cy (B), birth fraction rb (C), or death fraction rd (D) were assigned to the abundance exchange model, but the population size returned to the seasonal pattern over the long-term prediction.

82

In contrast, the fish population size declined when lower value of either Cy (Figure

3.5B) or rb (Figure 3.5C), or higher value of rd (Figure 3.5D) was used one at a time to

run the model. Results from the model simulation showed that the fish population

sizes exponentially increased or decreased for short-term predictions (e.g., around the

first 10 years of the simulation) when any one of the three population variables was

disturbed. The fish population sizes returned to their seasonal patterns of distribution

(Figure 3.5B–D) when the population growth reached the carrying capacity of the

system.

Across both temporal and spatial scales, the mean predicted percentage abundance

of each target species in August from the 100-year simulation agreed with the

observed percentage abundance in August 2004 (χ2predicted < 19.68, p>0.05, d.f. = 11)

on the FL gradient (Figure 3.6). Although the yellow perch abundance was somewhat

underestimated at FL1 and overestimated at FL3 (Figure 3.6), the prediction was, in

general, acceptable (χ2predicted = 12.15, i.e., < 19.68, p>0.05, d.f = 11) compared to the

observation.

The AEM could efficiently quantify the distribution pattern of bluegill (χ2predicted =

3.84) and logperch (χ2predicted = 9.35) on the FL gradient with less underestimation or

overestimation (Figure 3.6). The predicted abundance for bluntnose minnow and

fantail darter was also acceptable (χ2predicted = 16.37 for bluntnose minnow, χ2

predicted =

15.33 for fantail darter), although the overestimation or underestimation occurred at

some FL stations (Figure 3.6).

83

Figure 3.6. Observed abundances of yellow perch (YP), bluegill (BG), bluntnose minnow (BM), logperch (LP), and fantail darter (FTD) along the Floodwood gradient in August 2004 were compared with the mean predicted abundance ± standard deviation of the prediction of the associated fish species in August over the 100-year simulation.

84

The zero Pi of yellow perch (Table 3.2) for habitats with moderate-to-high

proportions of rock-bedrock above FL5 (Figure 3.3) limited the distribution of this

species to within FL1–FL5 as predicted by the model over the 100-year simulation

(Figure 3.7). The yellow perch population mainly aggregated at FL1 during winter

and early spring and migrated to FL2 in May and July, although a portion migrated to

FL4 and FL5. In the late growing season, most yellow perch moved downstream to

FL1 (Figure 3.7). Like yellow perch, the abundance exchange of bluegill occurred

between FL1 and FL5. The bluegill population mainly aggregated at FL1 during

winter and early spring. In June and July, bluegill were most abundant at FL2, but the

population spread to FL1 and FL3 in August, and most moved downstream to FL1 in

the late growing season (Figure 3.7).

The zero Pi of logperch for habitats containing high proportions of mud, such as

FL1–FL2, or moderate-to-high rock-bedrock content, such as FL6–FL12, defined the

distribution range of this species within FL3–FL5 (Figure 3.7). Logperch mainly

aggregated at FL4 and FL5 in the growing season, but at FL3 and FL4 in November

(Figure 3.7). Most fantail darter aggregated at the upstream portion on the FL gradient

across a year. Relatively few of them distributed to FL4 and FL5 in the growing

season (Figure 3.7). The distribution of fantail darter below FL4 (Figure 3.7) was

restricted by its zero Pi for habitats with little-to-no rock-bedrock content at FL1–FL3

(Figure 3.3). Unlike all other target species, bluntnose minnow distributed throughout

the FL gradient. In general, most bluntnose minnow aggregated at FL4, FL5, and

FL12 across a year (Figure 3.7).

85

Figure 3.7. Mean predicted abundance of yellow perch (YP), bluegill (BG), bluntnose minnow (BM), logperch (LP), and fantail darter (FTD) among months and stations on the Floodwood gradient over the 100-year simulation.

86

Model validation

The AEM with the estimated Pi for each target species on the FL gradient was

validated against the observed abundance of the same target species on the ST gradient

in the summer of 2002. The AEM predictions for yellow perch and fantail darter

agreed with the observations (Figure 3.8) along the ST gradient (χ2predicted < 14.07,

p>0.05, d.f. = 7).

Figure 3.8. Observed abundance of yellow perch (YP), bluegill (BG), bluntnose minnow (BM), and fantail darter (FTD) along the Sterling gradient in August 2002 were compared with the mean predicted abundance ± standard deviation of the prediction of the associated fish species in August over the 100-year simulation.

87

However, the model prediction was inconsistent with the observations of bluegill

and bluntnose minnow at some stations on the ST gradient. Significant differences

between predicted and observed abundances of bluegill occurred at ST7 and ST8

(χ2predicted > 3.84, p≤0.05, d.f. = 1). During the field sampling in August 2002, few

(6.3%) bluegill were observed at ST7 and fewer (3.1%) at ST8, while the average

abundance of bluegill over the 100-year simulation in the same summer month was

overestimated (16.2%) at ST7 and underestimated (0.3%) at ST8 (Figure 3.8).

Significant differences between predicted and observed abundances of bluntnose

minnow occurred at ST1 and ST3 (Figure 3.8). Few (3.5%) bluntnose minnow were

observed at ST1 and none at ST3 in August 2002, while the prediction of this species

was underestimated (0.3%) at ST1 and overestimated (4.7%) at ST3 (Figure 3.8).

According to the distribution patterns of the target species among months and ST

stations over the 100-year simulation (Figure 3.9), yellow perch distributed within the

range between ST1 and ST6, whereas bluegill distributed throughout the ST gradient

(Figure 3.9). Most yellow perch and bluegill aggregated at ST1 in winter and early

spring and migrated between ST1 and ST2 in the growing season. As on the FL

gradient, bluntnose minnow distributed throughout the ST gradient in the growing

season. Most of them aggregated at ST7 across a year (Figure 3.9). Predicted

distribution of fantail darter was between ST7 and ST8 across a year (Figure 3.9). In

general, the predicted abundance patterns of the modeled species in the summer

months were consistent with those from the field observations in the same time period.

88

Figure 3.9. Mean predicted abundance of yellow perch (YP), bluegill (BG), bluntnose minnow (BM), and fantail darter (FTD) among months and stations on the Sterling gradient over the 100-year simulation.

89

DISCUSSION

The modeled fish species showed relatively stable seasonal patterns of distribution

over the long-term prediction, and this was consistent with the findings of Reyes et al.

(1994), which revealed a stable seasonal pattern of pigfish abundance in Laguna de

Terminos, Mexico. The seasonal pattern of fish distribution can be explained in terms

of the negative and positive feedback loops (Ford 1999, Sterman 2000). The negative

feedback loop represents the contradictory conditions affecting fish distribution within

habitats between winter and the growing season, i.e., if more fish rapidly decline over

winter in these habitats, fewer fish will survive to have offspring or grow up in the

successive growing season. The positive feedback loop represents the supportive

condition increasing fish distribution within and across habitats between the two

seasons, i.e., if more young fish are born and more fish migrate from the other

habitats, a larger fish population in these habitats will be available to overwinter.

These consequences produced the seasonal patterns of fish distribution along Lake

Ontario’s coastal zone.

While the population variables (Cy, rb, and rd) shaped the population size of fish in

time, the model’s key variable, HP, accounted for fish distribution in space. As the

geometric mean of a species Pi for both static and dynamic habitat variables, HP could

capture the local range of species distribution and reflect the influence of habitat

changes (Morrison et al. 1985, Guay et al. 2000) on the abundance of a species within

its local range. Specifically, the estimated Pi (Table 3.2) for the static variable (i.e.,

substrate composition) defined the local distribution ranges of all modeled species on

the FL gradient and most of the species (except bluegill) on the ST gradient. Changes

in major substrate type(s) from a combination of mud and sand at FL3 to a

combination of gravel-cobble and rock-bedrock at FL4–FL5, and rock-bedrock at FL6

90

(Figure 3.3) marked these stations as the zone of transition in grain and extent

(Morrison and Hall, 2002) of the bottom landscape between lentic (embayment) and

lotic (stream) systems on the FL gradient. Similarly, changes in major substrate

type(s) from mud at ST6 to a combination of gravel-cobble and rock-bedrock at ST7

(Figure 3.4) indicated the distinct transition of the bottom landscape between the

lentic-lotic systems on the ST gradient.

In accordance with its local ranges of distribution in the upstream portions of

riffles and runs (Mundahl and Ingersoll 1983, Gray and Stauffer 1999) on both

gradients, fantail darter showed a high preference for shallow water, higher current

velocity, and rock-gravel streambeds (Table 3.2). In contrast, yellow perch and

bluegill represent embayment species that prefer deeper water habitat with common to

abundant aquatic plants and low current velocity (e.g., Harrington 1947, Wells 1968,

Helfman 1979, Hatzenbeler et al. 2000, Paukert and Willis 2002). Logperch

distributed within the lentic-lotic transition zone (FL3–FL5), habitats with moderate

current velocities, shallow water, and a combination of gravel-cobble and rock-

bedrock were preferred by this species (Smith 1985). Bluntnose minnow distributed

in a wide variety of water habitats (Trautman 1957, Becker 1983), but this species

preferred hard-bottomed, sandy, or gravelly shallows of pools for spawning activity

(Houston 2001) in the growing season. Consistently, high numbers of 0+year-old

bluntnose minnow were both observed and predicted at FL5 and FL12 that provide

suitable nursery substrates for this species.

Results from the model validation on the ST gradient revealed high levels of

agreement between the model predictions and the field observations for yellow perch

and fantail darter, but the predictions were not consistent with the observations for

bluegill and bluntnose minnow. The disagreement between predicted and observed

abundances of bluegill and bluntnose minnow at certain portions on the ST gradient

91

might be due to several causes. For bluntnose minnow, the small sample size (i.e., one

sample/station/month) of fish along the ST gradient might not well represent the

distribution pattern of this species along the gradient. Furthermore, increasing growth

of aquatic plants at the downstream and midstream portions of the gradient in the late

summer month could impede the ability to catch small fish. These factors might

explain why bluntnose minnow were not collected at most downstream-midstream

stations (ST3–ST6) on the ST gradient. Based on the broad tolerance of bluntnose

minnow for a wide variety of habitats, except the deeper waters of lakes and rivers

(Trautman 1957, Becker 1983), the model prediction revealed the distribution pattern

of this species throughout the ST gradient as being the same as that on the FL gradient.

The model parameters (i.e., Pi) for bluegill need some improvement to enhance the

ability of prediction on the ST gradient. Whereas this species was observed

throughout the ST gradient, results from the model simulation showed low efficiency

in quantifying abundance exchange of this species in the upstream portion of the

gradient. Pajak and Neves (1987) proposed three primary concerns that are generally

implicit in model validation procedures based on fish species Pi, one has to ensure that

(1) the fish species data to be validated are accurately collected, (2) the habitat

condition is accurately measured, and it is limited primarily by the habitat variables

included in the model, and (3) the sampling sites are representative sub-samples of the

target fish species’ habitats.

In this study, it is believed that methods of collecting both fish and habitat data

were appropriate. Nevertheless, it was noticeable that the sampling sites on the FL

gradient might not well represent all kinds of utilized substrates for bluegill on the ST

gradient. While the Pi for the rest of the habitat variables fell within the observed

ranges for bluegill on both study gradients, the Pi for substrate variables showed some

gaps between the quantified intervals (Table 3.2). This observation indicates that

92

although the substrates along the FL gradient were collected at a fine scale (900

samples), they were still unable to represent diverse proportions of the bottom

substrates occupied by bluegill on the ST gradient. Thus, the Pi for substrate variables

for this species may need to be improved for better model prediction. Similar gaps in

sampling coverage did not occur for the other modeled species.

Results from this study suggest that the species Pi for a given habitat variable

estimated from one study area may not completely represent the species Pi in another

study area (Pajak and Neves 1987, O’Connor 2002). Thus, to apply the AEM for

predicting fish species distribution in other water systems, one should crosscheck the

fish species Pi from existing data at both local and regional scales before incorporating

them into the AEM for computing the species HP. This is particularly necessary for

fish species with broad tolerances on the habitat variables that can powerfully define

local ranges of fish distribution (Hatzenbeler et al. 2000) or control distribution of fish

species from cell to cell (e.g., substrate types; Reyes et al. 1994). For different water

systems with different environmental conditions, influential habitat variables should

be considered for developing fish species Pi. Additionally, in large water systems

where fish movements in both vertical and horizontal directions are equally important,

two or more dimensions of fish distribution may need to be added in the model (e.g.,

Morrison et al. 1985, Reyes et al. 1994).

Overall, the AEM empirically developed from the fieldwork data performs rather

well for quantifying distributions of the coastal fish species assemblages along Lake

Ontario’s embayment-stream gradients. With its flexible model structure that can

accommodate array functions and multiple differential equations for both static and

dynamic components, the AEM can be modified to determine distribution patterns of

other types of organisms in different environmental and geographic systems.

93

Fish

spe

cies

RE

1R

E2

RE

3R

E4

RE

5R

E6

RE

7R

E8

RE

9R

E10

RE

11R

E12

Anad

rom

ous

Alew

ife24

6451

3258

5012

268

1353

014

429

7118

6885

4544

8302

7330

6382

(Alo

sa p

seud

ohar

engu

s)Am

eric

an s

had

1270

025

011

2908

633

150

3169

458

033

4465

729

278

2464

047

865

6385

845

563

(Alo

sa s

apid

issi

ma)

Atla

ntic

stu

rgeo

n0

70

00

30

014

717

30

0(A

cipe

nser

oxy

rinch

us

oxyr

inch

us)

Atla

ntic

tom

cod

1203

763

1415

2566

290

2110

70

10

4(M

icro

gadu

s to

mco

d)Bl

ueba

ck h

errin

g63

966

7261

145

225

1925

0118

8958

1661

1324

3664

2057

8014

9394

1437

8016

5875

1458

09(A

losa

aes

tival

is)

Hic

kory

sha

d3

90

90

00

00

10

2(A

losa

med

iocr

is)

Rai

nbow

sm

elt

1544

3027

1130

786

199

7412

310

664

1(O

smer

us m

orda

x m

orda

x)Se

a la

mpr

ey0

00

00

00

50

90

1(P

etro

myz

on m

arin

us)

Strip

ed b

ass

2725

960

760

5553

423

359

1501

611

946

7610

4969

6054

9021

1337

682

56(M

oron

e sa

xatil

is)

Cat

adro

mus

Amer

ican

eel

495

1919

1114

906

1038

1154

596

627

512

904

780

1824

(Ang

uilla

r ros

trata

)

App

endi

x 2.

1. C

umul

ativ

e ab

unda

nce

and

spec

ies r

ichn

ess o

f fis

h by

regi

on (R

E) a

long

the

Hud

son

Riv

er e

stua

ry

grad

ient

(RE1

–RE1

2) fr

om th

e fiv

e po

wer

pla

nt u

tiliti

es’ H

udso

n R

iver

est

uary

mon

itorin

g pr

ogra

m u

sing

bea

ch se

ine

surv

ey d

urin

g 19

74 to

200

1.

94

Fish

spe

cies

RE1

RE2

RE3

RE4

RE5

RE

6R

E7

RE

8R

E9

RE

10R

E11

RE

12

Est

uarin

eA

tlant

ic s

ilver

side

2047

395

912

6788

912

073

4011

1290

204

114

32

12

(Men

idia

men

idia

)B

ande

d ki

llifis

h95

7955

1642

131

796

4082

520

664

9930

8609

1677

418

514

3373

433

322

(Fun

dulu

s di

apha

nous

)Fa

t sle

eper

00

10

00

00

00

00

(Dor

mita

tor m

acul

atus

)Fo

ursp

ine

stic

kleb

ack

1516

4695

516

6826

9523

038

145

263

941

625

511

4(A

pelte

s qu

adra

cus)

Hog

chok

er24

1619

2314

3291

850

555

612

625

9512

936

525

(Trin

ecte

s m

acul

atus

)In

land

silv

ersi

de33

117

7417

7469

721

931

113

836

1089

4(M

enid

ia b

eryl

lina)

Mum

mic

hog

282

726

2782

2047

509

7030

1215

150

3050

1781

(Fun

dulu

s he

tero

clitu

she

tero

clitu

s)N

orth

ern

pipe

fish

996

4152

1607

283

305

00

00

06

(Syn

gnat

hus

fusc

us)

Sho

rtnos

e st

urge

on0

10

00

00

00

00

1(A

cipe

nser

bre

viro

stru

m)

Thre

espi

ne s

tickl

ebac

k11

3626

528

25

43

81

2(G

aste

rost

eus

acul

eatu

sac

ulea

tus)

Whi

te c

atfis

h14

1068

573

202

2639

3844

3546

2386

(Am

eiur

us c

atus

)

App

endi

x 2.

1 (C

ontin

ued)

.

95

Fish

spe

cies

RE1

RE2

RE3

RE4

RE5

RE

6R

E7

RE

8R

E9

RE

10R

E11

RE

12

Est

uarin

e (C

ontin

ued)

Whi

te m

ulle

t11

1311

210

03

00

20

0(M

ugil

cure

ma)

Whi

te p

erch

9677

7992

912

8023

4559

737

570

2709

922

055

2373

717

297

3874

041

222

1933

6(M

oron

e A

mer

ican

a)

Fres

hwat

erB

lack

bul

lhea

d0

01

00

00

01

10

0(A

mei

urus

mel

as)

Bla

ck c

rapp

ie0

1931

9028

4734

4430

5940

713

1(P

omox

is n

igro

mac

ulat

us)

Bla

ckno

se d

ace

1212

86

3920

40

00

254

(Rhi

nich

thys

atra

tulu

s)Bl

uegi

ll69

160

526

1047

810

1094

835

493

385

456

566

450

(Lep

omis

mac

roch

irus)

Blun

tnos

e m

inno

w0

02

02

00

31

26

87(P

imep

hale

s no

tatu

s)Br

idle

shi

ner

10

00

51

00

04

41

(Not

ropi

s bi

frena

tus)

Broo

k si

lver

side

00

00

00

00

00

410

0(L

abid

esth

es s

iccu

lus)

Broo

k st

ickl

ebac

k0

131

11

11

03

26

6(C

ulae

a in

cons

tans

)Br

ook

trout

00

20

00

10

00

00

(Sal

velin

us fo

ntin

alis

)

App

endi

x 2.

1 (C

ontin

ued)

.

96

Fish

spe

cies

RE1

RE2

RE3

RE4

RE5

RE

6R

E7

RE

8R

E9

RE

10R

E11

RE

12

Fres

hwat

er (C

ontin

ued)

Brow

n bu

llhea

d0

4775

942

572

815

038

310

699

9721

017

3(Ic

talu

rus

nebu

losu

s)B

row

n tro

ut1

34

15

41

01

02

5(S

alm

o tru

tta tr

utta

)C

entra

l mud

min

now

00

12

00

00

00

01

(Um

bra

limi)

Cha

in p

icke

rel

02

412

96

110

01

413

(Eso

x ni

ger )

Cha

nnel

cat

fish

00

01

02

25

23

21

(Icta

luru

s pu

ncta

tus)

Com

ely

shin

er0

00

02

230

01

00

17(N

otro

pis

amoe

nus)

Com

mon

car

p12

376

456

481

420

303

271

205

245

161

124

187

(Cyp

rinus

car

pio)

Com

mon

shi

ner

00

04

21

02

14

1812

0(N

otro

pis

corn

utus

)C

reek

chu

b0

180

24

35

71

02

5(S

emot

ilus

atro

mac

ulat

us)

Cut

lips

min

now

04

02

30

01

00

12

(Exo

glos

sum

max

illing

ua)

Eas

tern

mud

min

now

00

00

02

00

00

00

(Um

bra

pygm

aea)

Eas

tern

silv

ery

min

now

09

08

113

413

2931

4291

7510

(Hyb

ogna

thus

regi

us)

App

endi

x 2.

1 C

ontin

ued)

.

97

Fish

spe

cies

RE1

RE2

RE3

RE4

RE5

RE

6R

E7

RE

8R

E9

RE

10R

E11

RE

12

Fres

hwat

er (C

ontin

ued)

Em

eral

d sh

iner

238

4410

2615

411

987

123

105

387

1180

7355

(Not

ropi

s at

herin

oide

s)Fa

llfis

h0

50

03

25

102

1241

57(S

emot

ilus

corp

oral

is)

Fath

ead

min

now

01

00

62

33

819

227

(Pim

epha

les

prom

elas

)Fr

eshw

ater

dru

m0

01

00

00

01

30

0(A

plod

inot

us g

runn

iens

)G

izza

rd s

had

2294

194

354

1006

323

419

117

8072

7411

8(D

oros

oma

cepe

dian

um)

G

olde

n sh

iner

3313

810

2914

3353

4334

1758

1246

3426

3333

3315

1528

90(N

otem

igon

us c

ryso

leuc

as)

Gol

dfis

h10

8527

243

222

1244

170

660

982

054

146

940

2(C

aras

sius

aur

atus

aur

atus

)G

rass

pic

kere

l0

01

00

00

00

10

24(E

sox

amer

ican

us

verm

icul

atus

)G

reen

sun

fish

01

31

03

00

02

00

(Lep

omis

cya

nellu

s)La

rgem

outh

bas

s6

5421

599

527

965

846

161

031

623

115

616

9(M

icro

pter

us s

alm

oide

s)Lo

gper

ch0

01

00

01

21

42

8(P

erci

na c

apro

des)

App

endi

x 2.

1 (C

ontin

ued)

.

98

Fish

spe

cies

RE1

RE2

RE3

RE4

RE5

RE

6R

E7

RE

8R

E9

RE

10R

E11

RE

12

Fres

hwat

er (C

ontin

ued)

Long

ear s

unfis

h0

02

00

05

20

01

0(L

epom

is m

egal

otis

)Lo

ngno

se d

ace

00

04

11

00

00

24

(Rhi

nich

thys

cat

arac

tae)

Mim

ic s

hine

r0

00

00

00

10

00

1(N

otro

pis

volu

cellu

s)N

orth

ern

hog

suck

er2

00

01

32

13

413

4(H

ypen

teliu

m n

igric

ans)

Nor

ther

n pi

ke0

00

11

10

00

33

47(E

sox

luci

us)

Pug

nose

shi

ner

00

00

00

00

00

20

(Not

ropi

s an

ogen

us)

Pum

pkin

seed

1869

957

2465

7665

0854

9750

2567

0448

1136

8018

2915

78(L

epom

is g

ibbo

sus)

Rai

nbow

trou

t0

00

00

00

10

01

0(S

alve

linus

gai

rdne

ri )

Red

brea

st s

unfis

h3

4420

8173

319

959

115

4313

1510

3313

2374

966

5(L

epom

is a

uritu

s)R

edfin

pic

kere

l0

03

21

01

20

217

112

(Eso

x am

eric

anus

amer

ican

us)

Roc

k ba

ss1

15

73

221

1314

1715

156

(Am

blop

lites

rupe

stris

)

App

endi

x 2.

1 (C

ontin

ued)

.

99

Fish

spe

cies

RE1

RE2

RE3

RE4

RE5

RE

6R

E7

RE

8R

E9

RE

10R

E11

RE

12

Fres

hwat

er (C

ontin

ued)

Ros

yfac

e sh

iner

00

00

00

00

00

01

(Not

ropi

s ru

bellu

s)Sa

tinfin

shi

ner

00

00

83

03

75

2322

5(C

yprin

ella

ana

lost

ana)

Shie

ld d

arte

r0

00

00

00

10

00

0(P

erci

na p

elta

ta)

Smal

lmou

th b

ass

13

513

559

4179

151

159

158

346

(Mic

ropt

erus

dol

omie

ui)

Spo

tfin

shin

er0

37

117

3411

3316

1323

971

(Not

ropi

s sp

ilopt

erus

)Sp

otta

il sh

iner

509

1784

1211

320

052

5008

124

333

3554

641

154

2485

241

041

4202

776

124

(Not

ropi

s hu

dson

ius

)Sw

allo

wta

il sh

iner

00

00

10

03

02

02

(Not

ropi

s pr

ocne

)Te

ssel

late

d da

rter

3558

832

0381

4913

026

8656

5282

4509

6099

8386

5965

1571

7(E

theo

stom

a ol

mst

edi)

Tige

r mus

kellu

nge

00

00

00

00

00

11

(Eso

x lu

cius

cro

ss E

sox

mas

quin

ongy

)Tr

out-p

erch

00

00

00

00

00

040

(Per

cops

is o

mis

com

aycu

s)W

alle

ye0

00

00

00

10

03

0S

tizos

tedi

on v

itreu

m v

itreu

m)

App

endi

x 2.

1 (C

ontin

ued)

.

100

Fish

spe

cies

RE1

RE2

RE3

RE4

RE5

RE

6R

E7

RE

8R

E9

RE

10R

E11

RE

12

Fres

hwat

er (C

ontin

ued)

Whi

te b

ass

12

12

12

11

00

24

(Mor

one

chry

sops

)W

hite

cra

ppie

01

104

213

121

78

936

(Pom

oxis

ann

ular

is)

Whi

te s

ucke

r6

1618

337

914

311

814

075

3841

8322

1(C

atos

tom

us c

omm

erso

ni )

Yel

low

bul

lhea

d0

00

00

10

00

00

0(Ic

talu

rus

nata

lis )

Yello

w p

erch

116

181

453

274

133

177

196

151

149

148

659

(Per

ca fl

aves

cens

)

Mar

ine

Atla

ntic

cro

aker

3659

20

00

00

00

00

(Mic

ropo

goni

as u

ndul

atus

)A

tlant

ic h

errin

g50

00

00

00

00

00

0(C

lupe

a ha

reng

us h

aren

gus)

Atla

ntic

men

hade

n40

8627

076

2042

065

3533

1699

4422

4120

0130

3410

3767

5191

(Bre

voor

tia ty

rann

us)

Atla

ntic

moo

nfis

h1

30

00

00

00

00

0(S

elen

e se

tapi

nnis

)A

tlant

ic n

eedl

efis

h97

596

381

117

271

2650

2810

070

3155

(Stro

ngyl

ura

mar

ina)

Bay

anch

ovy

1934

2218

3599

2314

542

592

4129

945

8149

6055

652

063

147

38(A

ncho

a m

itchi

lli)

App

endi

x 2.

1 (C

ontin

ued)

.

101

Fish

spe

cies

RE1

RE2

RE3

RE4

RE5

RE

6R

E7

RE

8R

E9

RE

10R

E11

RE

12

Mar

ine

(Con

tinue

d)Bl

uefis

h37

5168

2365

9141

3520

7838

659

32

40

0(P

omat

omus

sal

tatri

x)Bu

tterfi

sh10

91

20

00

00

00

0(P

epril

us tr

iaca

nthu

s)C

reva

lle ja

ck52

136

149

218

223

40

00

00

(Car

anx

hipp

os)

Cun

ner

14

00

00

00

00

00

(Tau

togo

labr

us a

dspe

rsus

)Fo

ursp

ot fl

ound

er2

10

00

00

00

00

0(H

ippo

glos

sina

obl

onga

)G

oose

fish

00

01

00

00

00

00

(Lop

hius

am

eric

anus

)G

ray

snap

per

011

40

00

00

00

00

(Lut

janu

s gr

iseu

s)G

rubb

y0

10

00

00

00

00

0(M

yoxo

ceph

alus

aen

aeus

)In

shor

e liz

ardf

ish

1543

20

00

00

00

00

(Syn

odus

foet

ens)

Long

horn

scu

lpin

01

00

00

00

00

00

(Myo

xoce

phal

us

octo

dece

msp

inos

us)

Look

dow

n12

1913

20

00

00

00

0(S

elen

e vo

mer

)

App

endi

x 2.

1 (C

ontin

ued)

.

102

Fish

spe

cies

RE1

RE2

RE3

RE4

RE5

RE

6R

E7

RE

8R

E9

RE

10R

E11

RE

12

Mar

ine

(Con

tinue

d)N

aked

gob

y13

9017

174

21

00

00

0(G

obio

som

a bo

sc)

Nor

ther

n ki

ngfis

h12

313

712

00

00

00

00

0(M

entic

irrhu

s sa

xatil

is)

Nor

ther

n pu

ffer

824

60

00

00

00

00

(Sph

oero

ides

mac

ulat

us)

Nor

ther

n se

arob

in21

52

00

00

00

00

0(P

riono

tus

caro

linus

)N

orth

ern

star

gaze

r4

130

00

00

00

00

0(A

stro

scop

us g

utta

tus)

Ora

nges

potte

d fil

efis

h1

00

00

00

00

00

0(C

anth

erhi

nes

pullu

s)Pe

rmit

512

00

00

00

00

00

(Tra

chin

otus

blo

chii)

Red

hak

e0

00

00

00

20

00

0(U

roph

ycis

chu

ss)

Rou

gh s

ilver

side

286

879

677

217

341

10

10

21

(Mem

bras

mar

tinic

a)S

cup

00

10

00

00

00

00

(Ste

noto

mus

chr

ysop

s)S

eabo

ard

goby

00

10

00

00

00

00

(Gob

ioso

ma

gins

burg

i)S

ilver

hak

e3

00

069

00

00

00

0(M

erlu

cciu

s bi

linea

ris)

App

endi

x 2.

1 (C

ontin

ued)

.

103

Fish

spe

cies

RE1

RE2

RE3

RE4

RE5

RE

6R

E7

RE

8R

E9

RE

10R

E11

RE

12

Mar

ine

(Con

tinue

d)S

ilver

per

ch4

176

233

00

10

01

00

(Bai

rdie

lla c

hrys

oura

)S

mal

lmou

th fl

ound

er12

60

00

00

00

00

0(E

tropu

s m

icro

stom

us)

Spa

nish

mac

kere

l2

101

40

00

00

00

0(S

com

bero

mor

us m

acul

atus

)S

pot

656

564

351

476

8866

00

00

00

(Lei

osto

mus

xan

thur

us)

Spo

tfin

moj

arra

11

02

00

00

00

00

(Euc

inos

tom

us a

rgen

teus

)S

potte

d go

atfis

h0

00

00

00

00

00

17(P

seud

upen

eus

mac

ulat

us)

Spo

tted

hake

01

00

00

00

00

00

(Uro

phyc

is re

gia)

Stri

ped

anch

ovy

4319

74

20

00

00

00

(Anc

hoa

heps

ctus

)St

riped

killi

fish

10

00

00

00

00

00

(Fun

dulu

s m

ajal

is)

Stri

ped

mul

let

780

284

00

00

00

00

(Mug

il ce

phal

us)

Stri

ped

sear

obin

102

840

00

00

00

00

0(P

riono

tus

evol

ans)

Sum

mer

flou

nder

146

167

9326

106

00

00

00

(Par

alic

hthy

s de

ntat

us)

App

endi

x 2.

1 (C

ontin

ued)

.

104

Fish

spe

cies

RE

1R

E2R

E3R

E4R

E5R

E6R

E7R

E8R

E9

RE1

0R

E11

RE1

2

Mar

ine

(Con

tinue

d)Ta

utog

910

810

00

00

00

00

0(T

auto

ga o

nitis

)W

eakf

ish

247

244

2713

112

00

00

10

(Cyn

osci

on re

galis

)W

indo

wpa

ne7

10

00

00

00

00

0(S

coph

thal

mus

aqu

osus

)W

inte

r flo

unde

r92

984

250

70

20

00

00

0(P

seud

ople

uron

ecte

s am

eric

anus

)

Uni

dent

ified

spe

cies

Eso

x sp

p.0

00

00

00

00

00

1Fu

ndul

us s

pp.

00

10

00

00

01

00

unid

entif

ied

cent

rarc

hid

102

186

947

2252

3782

2942

1458

2775

3290

1096

540

2032

unid

entif

ied

clup

eid

1916

1254

521

5157

5428

828

4605

570

521

5837

190

574

8205

878

494

unid

entif

ied

cypr

inid

22

2010

873

917

136

119

613

410

861

1208

unid

entif

ied

gobi

id0

00

01

00

00

00

0un

iden

tifie

d M

oron

e0

11

186

31

01

40

78un

iden

tifie

d se

arob

in7

20

00

00

00

00

0un

iden

tifie

d st

urge

on0

00

00

02

00

00

0un

iden

tifie

d su

cker

00

00

00

30

00

03

unid

entif

ied

spp.

00

40

221

10

12

01

unid

entif

ied

spp.

00

30

00

00

00

00

unid

entif

ied

spp.

61

43

31

10

22

02

unid

entif

ied

spp.

07

7562

51

00

00

00

App

endi

x 2.

1 (C

ontin

ued)

.

105

Appendix 2.2. Multiple functions (aggregator, mediator, soft barrier, and hard barrier) of the ecotones (from all detected frequencies) on the Hudson fish species in each time period during 1974–2001.

Ecotone No. of selected1

location Aggregator Mediator Soft barrier Hard barrier species/gradient

RE2May-75 3.1 21.9 28.1 46.9 32May-76 6.7 26.7 26.7 39.9 30May-78 15.6 12.5 21.9 50.0 32Jun-74 8.8 11.8 20.6 58.8 34Jun-75 10.9 16.2 24.3 48.6 37Jun-76 12.2 9.8 34.1 43.9 41Jun-77 16.7 13.9 25.0 44.4 37Jul-78 7.0 18.6 14.0 60.4 43Jul-79 26.2 28.6 11.9 33.3 42Jul-84 31.0 21.4 9.5 38.1 42Jul-90 32.3 19.4 12.9 35.4 31Aug-78 10.2 18.4 20.4 51.0 49Aug-84 36.6 14.6 22.0 26.8 41Aug-96 21.4 10.7 17.9 50.0 28Sep-84 31.3 12.5 15.6 40.6 32Nov-76 18.2 13.6 22.7 45.5 22Nov-79 37.5 17.5 12.5 32.5 40Dec-78 0.0 11.1 27.8 61.1 18

RE2-RE3Jun-91 19.0 23.9 21.4 35.7 42Jul-00 20.0 17.1 8.6 54.3 35Sep-79 42.9 34.3 11.4 11.4 35Sep-81 27.9 9.3 23.3 39.5 43Sep-87 46.0 16.2 18.9 18.9 37Sep-99 12.5 6.3 18.8 62.4 32Oct-83 36.8 5.3 21.1 36.8 19Nov-78 16.7 33.3 13.3 36.7 30

RE2-RE4Jul-86 42.1 15.8 18.4 23.7 38Aug-74 22.0 30.0 22.0 26.0 50Aug-79 36.4 25.0 18.1 20.5 44Sep-82 38.5 25.6 18.0 17.9 39Sep-88 51.5 9.1 15.2 24.2 33Sep-96 30.0 20.0 30.0 20.0 20Oct-75 38.9 22.2 22.2 16.7 36Oct-98 22.6 19.4 6.5 51.5 31Nov-80 23.5 29.5 23.5 23.5 17Nov-86 36.0 28.0 12.0 24.0 25

Function (%) for the total selected species

106

Appendix 2.2 (Continued).

Ecotone No. of selected1

location Aggregator Mediator Soft barrier Hard barrier species/gradient

RE2-RE5Jun-94 9.4 12.5 28.1 50.0 32Aug-88 40.6 25.0 28.1 6.3 32Aug-90 56.3 15.6 18.7 9.4 32

RE2-RE6Jul-91 14.3 11.8 42.8 31.1 42Aug-87 30.3 15.2 33.3 21.2 33Sep-00 21.5 21.4 25.0 32.1 28

RE3Aug-96 25.0 17.9 25.0 32.1 28Sep-90 10.0 16.7 26.7 46.6 30

RE3-RE4Apr-74 50.0 11.1 16.7 22.2 18May-78 37.5 37.5 15.6 9.4 32Jun-75 29.8 40.5 10.8 18.9 37Jul-75 25.6 30.2 23.3 20.9 43Aug-81 18.0 35.9 28.2 17.9 39Aug-97 3.6 25.0 42.8 28.6 28Sep-01 3.5 13.8 24.1 58.6 29Oct-84 20.1 23.3 33.3 23.3 30Dec-78 38.9 27.8 11.1 22.2 18

RE3-RE5Aug-98 15.2 18.2 33.3 33.3 33Sep-98 7.0 10.3 37.9 44.8 29Oct-85 20.0 20.0 50.0 10.0 30Oct-92 21.4 25.0 14.3 39.3 28Nov-87 14.3 7.1 35.7 42.9 14

RE3-RE6May-76 60.0 16.6 6.8 16.6 30Aug-83 11.2 25.0 44.4 19.4 36Aug-91 4.7 20.9 37.2 37.2 43

RE4Jun-90 9.5 0.0 14.3 76.2 21

RE4-RE5Jul-90 3.2 9.7 12.9 74.2 31Aug-99 8.2 10.8 32.4 48.6 37Oct-83 15.8 26.3 31.6 26.3 19Oct-93 13.3 20.1 33.3 33.3 30Sep-81 4.7 9.3 44.2 41.8 43

Function (%) for the total selected species

107

Appendix 2.2 (Continued).

Ecotone No. of selected1

location Aggregator Mediator Soft barrier Hard barrier species/gradient

RE4-RE6Aug-86 2.5 20.0 35.0 42.5 40Sep-87 11.1 25.0 36.1 27.8 36Sep-90 10.0 33.3 20.0 36.7 30

RE4-RE7Oct-00 4.5 45.5 22.7 27.3 22

RE4-RE8Nov-93 57.2 28.6 7.1 7.1 14

RE5Jun-91 0.0 4.8 26.2 69.0 42Jun-92 0.0 8.3 25.0 66.7 24Jul-93 0.0 13.9 22.2 63.9 36Sep-93 7.0 3.4 31.0 58.6 29Sep-96 15.0 10.0 30.0 45.0 20Dec-78 50.0 16.7 5.6 27.7 18

RE5-RE6Jun-90 4.6 13.6 27.3 54.5 22Aug-97 8.8 11.8 32.4 47.0 34Oct-98 6.5 16.1 41.9 35.5 31

RE5-RE7Dec-75 50.0 17.0 16.5 16.5 12

RE6Jun-91 0.0 7.1 16.7 76.2 42Jun-92 0.0 16.7 20.8 62.5 24Jul-90 3.2 3.2 13.0 80.6 31Aug-88 3.2 15.6 28.1 53.1 32Dec-78 5.6 16.7 11.0 66.7 18

RE6-7Oct-83 5.3 15.8 31.5 47.4 19Sep-93 6.9 17.2 34.5 41.4 29Sep-95 3.8 18.5 44.4 33.3 27

RE6-RE9Aug-90 7.7 15.4 50.0 26.9 26Oct-94 18.6 29.6 11.1 40.7 27

RE6-10Nov-87 28.6 14.3 7.1 50.0 14Oct-92 11.1 11.1 29.6 48.2 27

Function (%) for the total selected species

108

Appendix 2.2 (Continued).

Ecotone No. of selected1

location Aggregator Mediator Soft barrier Hard barrier species/gradient

RE7May-76 3.3 3.3 30.1 63.3 30

RE7-RE8Jun-90 33.3 14.3 9.5 42.9 21

RE7-RE9Dec-78 16.7 16.7 22.2 44.4 18

RE7-RE11Aug-87 21.2 27.3 18.2 33.3 33Aug-97 29.4 14.7 20.6 35.3 34Sep-87 19.4 22.3 19.4 38.9 36Sep-90 16.6 26.7 26.7 30.0 30Oct-88 27.6 6.9 31.0 34.5 29

RE8Aug-89 23.3 16.7 20.0 40.0 30Oct-00 4.5 9.1 18.2 68.2 22

RE8-9Apr-76 0.0 0.0 45.2 54.8 31

RE8-RE10Nov-85 13.1 30.4 17.4 39.1 23Dec-75 0.0 0.0 25.0 75.0 12

RE8-11Apr-78 4.0 8.0 40.0 48.0 25Sep-95 48.1 11.2 7.4 33.3 27

RE9Nov-77 2.9 2.9 26.6 67.6 34

RE9-RE10Jun-90 13.6 9.1 9.1 68.2 22Oct-00 22.7 22.7 22.7 31.9 22

RE9-RE11Nov-93 0.0 7.2 57.1 35.7 14

RE10-RE11Apr-76 0.0 6.5 32.3 61.2 31Nov-77 0.0 8.8 32.4 58.8 34

Function (%) for the total selected species

109

Appendix 2.2 (Continued).

Ecotone No. of selected1

location Aggregator Mediator Soft barrier Hard barrier species/gradient

RE11Jun-90 19.0 19.0 4.8 57.2 21Oct-92 11.1 7.4 22.2 59.3 27Oct-00 31.9 22.7 4.5 40.9 22Nov-87 21.4 14.3 0.0 64.3 14

Function (%) for the total selected species

1A fish species with less than 5 individuals collected along the Hudson River estuary gradient in each of the 116 time periods was not included in the analysis.

Appendix 2.3. Multiple functions (neutral, partly inhibit, and completely inhibit) of the ecotone peripheries (from all detected frequencies) on the Hudson fish species in each time period during 1974–2001.

Ecotone periphery No. of selected1

location between Neutral Partly inhibit Completely inhibit species/gradient

RE1-RE2Apr-78 8.0 20.0 72.0 25Apr-79 12.5 29.2 58.3 24May-74 26.1 4.3 69.6 23May-77 12.2 24.2 63.6 33Jun-78 12.5 20.0 67.5 40Jul-76 13.3 15.6 71.1 45Aug-80 12.2 9.8 78.0 41Aug-82 8.9 23.5 67.6 34Aug-89 10.0 10.0 80.0 30Aug-93 14.7 2.9 82.4 34Sep-75 8.3 22.9 68.8 48Sep-83 6.3 18.7 75.0 32Sep-85 11.8 23.5 64.7 34Sep-86 3.0 9.1 87.9 33Oct-74 11.8 17.6 70.6 34Oct-77 15.2 27.3 57.5 33Oct-81 16.7 13.9 69.4 36Nov-77 14.7 23.5 61.8 34

RE2-RE3Jun-79 13.9 44.4 41.7 36Jun-95 28.6 4.8 66.6 21Aug-75 11.1 31.1 57.8 45Nov-75 10.0 23.3 66.7 30

Function (%) for the total selected species

110

Appendix 2.3 (Continued).

Ecotone periphery No. of selected1

location between Neutral Partly inhibit Completely inhibit species/gradient

RE3-RE4Jun-88 17.1 17.1 65.8 35Jul-96 18.0 25.6 56.4 39Sep-76 18.0 41.0 41.0 36Sep-80 11.1 33.3 55.6 36Sep-94 8.6 17.1 74.3 35Oct-78 25.7 34.3 40.0 35Oct-80 15.4 26.9 57.7 26Oct-86 25.8 29.0 45.2 31Oct-95 13.3 30.0 56.7 30

RE4-RE5Sep-92 17.3 31.0 51.7 29

RE5-RE6Jul-97 8.6 40.0 51.4 35Jul-99 21.2 12.1 66.7 33Aug-94 16.2 29.0 54.8 31Aug-95 12.9 38.7 48.4 31

RE7-RE8Aug-00 14.3 40.0 45.7 35Oct-06 12.5 54.2 33.3 24

RE8-RE9Apr-79 16.7 25.0 58.3 24

RE10-RE11Jun-74 35.3 35.3 29.4 34Jun-78 22.5 35.0 42.5 40Jul-75 30.2 30.2 39.6 43Aug-79 29.5 29.5 41.0 44Sep-75 22.9 14.6 62.5 48Sep-80 30.6 19.4 50.0 36Oct-74 14.7 38.2 47.1 34Nov-80 11.8 5.9 82.3 17Nov-86 24.0 24.0 52.0 25

RE11-RE12May-74 13.1 21.7 65.2 23May-77 9.1 30.3 60.6 33May-78 21.9 18.7 59.4 32May-80 16.7 13.9 69.4 36Jun-79 13.9 16.7 69.4 36Jun-80 20.0 32.5 47.5 40Jun-95 9.6 33.3 57.1 21Jul-76 19.5 31.7 48.8 41Jul-77 14.0 25.5 60.5 43Jul-78 28.6 26.2 45.2 42

Function (%) for the total selected species

111

Appendix 2.3 (Continued).

Ecotone periphery No. of selected1

location between Neutral Partly inhibit Completely inhibit species/gradient

RE11-RE12 (Continued)Jul-97 8.6 20.0 71.4 35Jul-00 20.6 32.4 47.0 34Aug-75 20.0 28.9 51.1 45Aug-76 29.2 20.8 50.0 48Aug-78 28.6 22.4 49.0 49Aug-80 26.8 19.5 53.7 41Sep-76 25.6 18.0 56.4 39Sep-78 28.9 15.8 55.3 38Sep-79 22.9 22.9 54.2 35Sep-82 20.5 18.0 61.5 39Sep-94 20.0 20.0 60.0 35Sep-99 18.8 43.8 37.4 32Sep-01 16.7 13.3 70.0 30Oct-76 10.3 25.6 64.1 39Oct-78 20.0 11.4 68.6 35Oct-79 22.3 19.4 58.3 36Oct-80 15.4 26.9 57.7 26Oct-93 10.3 13.8 75.9 29Nov-78 16.7 10.0 73.3 30Nov-79 13.4 23.3 63.3 30

Function (%) for the total selected species

1A fish species with less than 5 individuals collected along the Hudson River estuary gradient in each of the 116 time periods was not included in the analysis.

112

REFERENCES

Able, K. W. 2005. A re-examination of fish estuarine dependence: evidence for

connectivity between estuarine and ocean habitats. Estuarine Coastal and Shelf Science 64:5–17.

Able, K. W., and M. P. Fahay. 1998. The first year in the life of estuarine fishes in the

middle Atlantic Bight. Rutgers University Press, New Brunswic, New Jersey.

Abood, K. A., and S. G. Metzger. 1996. Comparing impacts to shallow-water habitats

through time and space. Estuaries 19:220–228.

Armand, A. D. 1992. Sharp and gradual mountain timberlines as a results of species

interaction. Pages 360–378 in Hansen and di Castri, editors. Landscape boundaries: Consequences for biotic diversity and ecological flows. Springer-Verlag, New York.

ASA Analysis & Communication, I. 2004. 2001 year class report for the Hudson

River estuary monitoring program. Dynegy Roseton L.L.C., Entergy Nuclear Indian Point 2 L.L.C., Entergy Nuclear Indian Point 3 L.L.C., and Mirant Bowine L.L.C., New York.

Attrill, M. J., and S. D. Rundle. 2002. Ecotone or ecocline: ecological boundaries in

estuaries. Estuarine, Coastal and Shelf Science 55:929–936.

Baker, J., K. French, and R. J. Whelan. 2002. The edge effect and ecotonal species:

Bird communities across a natural edge in southeastern Australia. Ecology 83:3048–3059.

Barnthouse, L. W., R. J. Klauda, D. S. Vaughan, and R. L. Kendall, editors. 1988.

Science, law, and Hudson River power plants: a case study in environmental impact assessment. American Fisheries Society, Monograph 4, Bethesda, Maryland.

Becker, G. C. 1983. Fishes of Wisconsin. University of Wisconsin Press, Madison.

113

Bell, R. E., R. D. Flood, S. Carbotte, W. B. F. Ryan, F. O. Nitsche, S. Chilrud, R. Arko, V. Ferrini, A. Slagle, C. Bertinato, and M. Turrin. 2006. Hudson River estuary program benthic mapping project. Hudson River Estuary Program, New York State Department of Environmental Conservation, New York.

Bovee, K. D. 1982. A guide to stream habitat analysis using the instream flow

incremental methodology: Instream Flow Information Paper 5. FWS/OBS–82/26, U.S. Fish and Wildlife Service.

Bovee, K. D. 1986. Development and evaluation of habitat suitability criteria for use

in the Instream Flow Incremental Methodology. Instream Flow Information Paper 21: 86(7), U.S. Fish and Wildlife Service, Washington DC.

Bowersox, M. A., and D. G. Brown. 2001. Measuring the abruptness of patchy

ecotones. Plant Ecology 156:89–103.

Boyle, R. H. 1979. The Hudson River, Expanded edition. W.W. Norton and Company,

New York.

Cadenasso, M. L., S. T. A. Pickett, K. C. Weathers, and C. G. Jones. 2003. A

framework for a theory of ecological boundaries. BioScience 53:750–758.

Cadenasso, M. L., M. M. Traynor, and S. T. A. Pickett. 1997. Functional location of

forest edges: gradients of multiple physical factors. Canadian Journal of Forest Research 27:774–782.

Clements, F. E. 1897. Peculiar zonal formations of the Great Plains. American

Naturalist 31:968–970.

Clements, F. E. 1905. Research methods in ecology. University Pubishing Company,

Nebraska, USA.

Coch, N. K., and H. J. Bokuniewicz. 1986. Oceanographic and geologic framework of

the Hudson system. Northeastern Geology 8:96–108.

Csillag, F., B. Boots, M.-J. Fortin, K. Lowell, and F. Potvin. 2001. Multiscale

characterization of boundaries and landscape ecological patterns. Geomatica 55:291–307.

114

Daniels, R. A., K. E. Limburg, R. E. Schmidt, D. L. Strayer, and R. C. Chambers. 2005. Changes in fish assemblages in the tidal Hudson River, New York. American Fisheries Society Symposium 45:471–503.

Degraaf, D. A., and L. H. Bain. 1986. Habitat use by and preferences of Juvenile

Atlantic salmon in two Newfoundland Rivers. Transactions of the American Fisheries Society 115:671–681.

Delcourt, P. A., and H. R. Delcourt. 1992. Ecotone dynamics in space and time. Pages

19–54 in Hansen and di Castri, editors. Landscape Boundaries: Consequences for biotic diversity and ecological flows. Springer-Verlag, New York.

Dew, I. M. 2001. Theoretical model of a new fishery under a simple quota

management system. Ecological Modelling 143:59–70.

di Castri, F., and A. J. Hansen. 1992. The environment and development crises as

determinants of landscape dynamics. Pages 3–18 in Hansen and di Castri, editors. Landscape Boundaries: Consequences for biotic diversity and ecological flows. Springer-Verlag, New York.

di Castri, F., A. J. Hansen, and M. M. Holland, editors. 1988. A new look at ecotones:

emerging international projects on landscape boundaries (special issue 17). Biology International. IUBS, Paris, France.

Fagan, W. F., R. S. Cantrell, and C. Cosner. 1999. How habitat edges change species

interactions. American Naturalist 153:165–182.

Fagan, W. F., M. J. Fortin, and C. Soykan. 2003. Integrating edge detection and

dynamic modeling in quantitative analyses of ecological boundaries. BioScience 53:730–738.

Fielding, A. H. 2002. What are appropriate characteristics of an accuracy measure?

Pages 271–280 in Scott, Heglund, Morrison, and al., editors. Predicting species occurrences: issues of accuracy and scale. Island Press, Washington.

Fischer, H. B., E. J. List, R. C. Y. Koh, J. Imberger, and N. H. Brooks. 1979. Mixing

in inland and coastal waters. Academic Press, San Diego, California.

115

Ford, A. 1999. Modeling the Environment: an introduction to system dynamics modeling of environmental systems. Island Press, Washington, D.C.

Forman, R. T. T., and P. N. Moore. 1992. Theoretical foundations for understanding

boundaries in landscape mosaics. Pages 236–258 in Hansen and di Castri, editors. Landscape Boundaries: Consequences for biotic diversity and ecological flows. Springer-Verlag, New York.

Fortin, M.-J. 1994. Edge detection algorithms for two dimensional ecological data.

Ecology 75:956–965.

Fortin, M.-J., R. J. Olson, S. Ferson, L. Iverson, C. Hunsaker, G. Edwards, D. Levine,

K. Butera, and V. Klemas. 2000. Issues related to the detection of boundaries. Landscape Ecology 15:453–466.

Fruin, J. J. 1971. Pedestrian planning and design. Metropolitan Association of Urban

Designers and Environmental Planners, New York.

Gaff, H., D. L. DeAngelis, L. J. Gross, and R. Salinas. 2000. A dynamic landscape

model for fish in the Everglades and its application to restoration. Ecological Modelling 127:33–52.

Gauch, H. G., Jr. 1982. Multivariate analysis in community ecology, first edition.

Cambridge University Press, New York.

Gertseva, V. V., V. I. Gertsev, and N. Y. Ponomarev. 2003. Tropho-ethological

polymorphism of fish as a strategy of habitat development: a simulation model. Ecological Modelling 167:159–164.

Gertseva, V. V., J. E. Schindler, V. I. Gertsev, N. Y. Ponomarev, and W. R. English.

2004. A simulation model of the dynamics of aquatic macroinvertebrate communities. Ecological Modelling 176:173–186.

Gido, K. B., C. W. Hargrave, W. J. Matthews, and G. D. Schnell. 2002. Structure of

littoral-zone fish communities in relation to habitat, physical, and chemical gradients in a southern reservoir. Environmental Biology of Fishes 63:253–263.

116

Gilmore, R. G., Jr. 1995. Environmental and bigeographic factors influencing ichthyofaunal diverity in the Indian River Lagoon. Bulletin of Marine Science 57:153–170.

Gottlieb, S. J. 1998. Nutrient removal by age-0 Atlantic menhaden (Brevoortia

tyrranus) in Chesapeake Bay and implications for seasonal management of the fishery. Ecological Modelling 112:111–130.

Gray, E. V. S., and J. R. Stauffer, Jr. 1999. Comparative microhabitat use of

ecologically similar benthic fishes. Environ. Biol. Fishes. 56:443–453.

Guay, J. C., D. Boisclair, D. Rioux, M. Leclerc, M. Lapointe, and P. Legendre. 2000.

Development and validation of numerical habitat models for juveniles of Atlantic salmon (Salmo salar). Canadian Journal of Fisheries and Aquatic Science 57:2065–2075.

Guisan, A., and N. E. Zimmermann. 2000. Predictive habitat distribution models in

ecology. Ecological Modelling 135:147–186.

Haddad, N. M. 1999. Corridor use predicted from behaviors at habitat boundaries.

American Naturalist 153:215–227.

Haefner, J. W. 1996. Modeling biological systems: principles and applications.

Chapman & Hall, New York.

Hansen, A. J., and F. di Castri, editors. 1992. Landscape Boundaries: Consequences

for biotic diversity and ecological flows. Springer-Verlag, New York.

Hansen, A. J., F. di Castri, and R. J. Naiman. 1988. Ecotones: what and why? Pages 9-

45 in di Castri, Hansen, and Holland, editors. A new look at ecotones: emerging international project on landscape boundaries (special issue 17). Biological International, IUBS, Paris.

Harding, E. K. 2002. Modelling the influence of seasonal inter-habitat movements by

an ecotone rodent. Biological Conservation 104:227–237.

Harrington, R. W., Jr. 1947. Observations on the breeding habits of the yellow perch,

Perca flavescens (Mitchill). Copeia 1947:199–200.

117

Hatzenbeler, G. R., M. A. Bozek, M. J. Jennings, and E. E. Emmons. 2000. Seasonal variation in fish assemblage structure and habitat structure in the nearshore littoral zone of Wisconsin Lakes. North American Journal of Fisheries Management 20:360–368.

Helfman, G. S. 1979. Twilight activities of yellow perch, Perca flavescens. J. Fish.

Res. Board Can. 36:173–179.

High Performance Systems, Inc. 2002. STELLA(R) Research, version 7.0.3 for

Windows. Hanover, NH.

Hill, M. O. 1979. DECORANA: a FORTRAN program for detrended correspondence

analysis and reciprocal averaging. Cornell University, Ithaca, New York.

Hoeinghaus, D. J., K. O. Winemiller, and D. C. Taphorn. 2004. Compositional change

in fish assemblages along the Andean piedmont-Llanos floodplain gradient of the Rio Portuguesa, Venezuela. Neotropical Ichthyology 2:85–92.

Holland, M. M. 1988. SCOPE-MAB technical consultations of landscape boundaries:

report of a SCOPE/MAB workshop on ecotone, January 5–7, 1987. Pages 47–106 in di Castri, Hansen, and Holland, editors. A new look at ecotones: emerging international project on landscape boundaries (special issue 17). Biological International. IUBS, Paris.

Houston, J. 2001. Status of the bluntnose minnow, Pimephales notatus, in Canada.

The Canadian Field-Naturalist 115:145–151.

Hurst, T. P., K. A. McKown, and D. O. Conover. 2004. Interannual and long-term

variation in the Nearshore fish community of the mesohaline Hudson River estuary. Estuaries 27:659–669.

Jaureguizara, A. J., R. Mennic, C. Bremeca, H. Mianzan, and C. Lastaa. 2003. Fish

assemblage and environmental patterns in the Rio de la Plata estuary. Estuarine, Coastal and Shelf Science 56:921–933.

Johnson, A. R., J. A. Wiens, B. T. Milne, and T. O. Crist. 1992. Animal movements

and population dynamics in heterogeneous landscapes. Landscape Ecology 7:63–75.

118

Johnston, C. A., J. Pastor, and G. Pinay. 1992. Quantitative methods for studying landscape boundaries. Pages 107–125 in Hansen and di Castri, editors. Landscape Boundaries: Consequences for biotic diversity and ecological flows. Springer-Verlag, New York.

Johnston, V. R., and E. P. Odum. 1956. Breeding bird populations in relation to plant

succession on the piedmont of Georgia. Ecology 37:50–62.

Kilpatrick, H. J., A. M. LaBonte, J. S. Barclay, and G. Warner. 2004. Assessing

strategies to improve bowhunting as an urban deer management tool. Wildlife Society Bulletin 32:1177–1184.

Kirchhofer, A. 1995. Morphologicla variability in the ecotone–an important factor for

the conservation of fish species richness in Swiss rivers. Hydrobiologia 303:103–110.

Kolasa, J., and M. Zalewski. 1995. Notes on ecotone attributes and functions.

Hydrobiologia 303:1–7.

Kot, M. 2001. Elements of mathematical ecology. Cambridge University Press,

Cambridge, UK.

Krivtsov, V., Corliss, J., Bellinger, E., Sigee, D., 2000. Indirect regulation rule for

consecutive stages of ecological succession. Ecol. Model. 133, 73–81.

Kupschus, S. 2003. Development and evaluation of statistical habitat suitability

models: and example based on juvenile spotted seatrout Cynoscion nebulosus. Marine Ecology Progress Series 265:197–212.

Kupschus, S., and D. Tremain. 2001. Association between fish assemblages and

environmental factors in nearshore habitats of a subtropical estuary. Journal of Fish Biology 58:1383–1403.

Kupschus, S., and D. Tremain. 2001. Associations between fish assemblages and

environmental fiactors in nearshore habitats of a subtropical estuary. Journal of Fish Biology 58:1383–1403.

119

Lachavanne, J.-B. 1997. Why study biodiversity in land-inland water ecotones? Pages 1–33 in Lachvanne and Juge, editors. Biodiversity in land-inland water ecotones. UNESCO, Paris.

Lay, D. 1938. How valuable are woodland clearings to birdlife? Wilson Bulletin

50:254–256.

Leopold, A. 1933. Game management. Charles Scribner's Sons, New York.

Lewis, M. A. 1997. Variability, patchiness, and jump dispersal in the spread of an

invading population. Pages 46–69 in Tilman and Kareiva, editors. Spatial ecology: the role of space in population dynamics and interspecific interactions. Princeton University Press, Princeton, New Jersey.

Lidicker, W. Z. J. 1999. Responses of mammals to habitat edges: an overview.

Landscape Ecology 14:333–343.

Limburg, K. E., M. A. Moran, and W. H. McDowell. 1986. The Hudson River

ecosystem. Springer-Verlag, New York.

Lucas, M. C., and E. Baras. 2001. Migration of freshwater fishes. Blackwell Science,

London.

Lumia, R. 1998. Flood of January 19–20, 1996 in New York State. Water-Resources

Investigations Report, U.S. Geological Survey, Albany, New York.

Mancroni, W. J., M. W. Daley, and W. Dey. 1992. Recent Trends and Patterns in

selected water quality parameters in the mid-Hudson River estuary. Pages 59–81 in Smith, editor. Estuarine Research in the 1980s. State University of New York Press, Albany.

Marín, V. H. 1997. A simple-biology, stage-structured population model of the spring

dynamics of Calanus chilensis at Mejillones del Sur Bay, Chile. Ecological Modelling 105:65–82.

Marshall, S., and M. Elliott. 1998. Environmental Influences on the Fish Assemblage

of the Humber Estuary. Estuarine, Coastal and Shelf Science 46:175–184.

120

Martino, E. J., and K. W. Able. 2003. Fish assemblages across the marine to low salinity transition zone of a temperate estuary. Estuarine Coastal and Shelf Science 56:969–987.

Matthysen, E. 2002. Boundary effects on dispersal between habitat patches by forest

birds (Parus major, P. caeruleus). Landscape Ecology 17:509–515.

McCausland, W. D., E. Mente, G. J. Pierce, and I. Theodossiou. 2006. A simulation

model of sustainability of coastal communities: aquaculture, fishing, environment and labour markets. Ecological Modelling 193:271–294.

Miller, B. A., and S. Sadro. 2003. Residence time and seasonal movements of juvenile

coho salmon in the ecotone and lower estuary of Winchester Creek, South Slough, Oregon. Transactions of the American Fisheries Society 132:546–559.

Morrison, K. A., N. Therien, and B. Coupal. 1985. Simulating fish redistribution in the

LG-2 reservoir after flooding. Ecological Modelling 28:97–111.

Morrison, M. L., and L. S. Hall. 2002. Standard terminology: toward a common

languge to advance ecological understanding and application. Pages 43–52 in Scott, Heglund, Morrison, Haufler, Raphael, Wall, and Samson, editors. Predicting species occurrences. Island Press, Washington.

Mundahl, N. D., and C. G. Ingersoll. 1983. Early autumn movements and densities of

johny (Etheostoma nigrum)and fantail (E. Flabellare) darters in a southwestern Ohio stream. Ohio J. Sci. 83:103–108.

Naiman, R. J., and H. Décamps, editors. 1990. The ecology and management of

aquatic-terrestrial ecotones. The Parthenon Publishing Group, Paris.

National Oceanic and Atmospheric Administration. 1989. Estuaries of the United

States: vital statistics of a natural resource base. Strategic Environmental Assessments Division, Rockville, MD.

Neilson, R. P., G. A. King, R. L. DeVelice, and J. M. Lenihan. 1992. Regional and

local vegetation patterns: The responses of vegetation diversity to subcontinental air masses. Pages 129–149 in Hansen and Castri, editors. Landscape boundaries: Consequences for biotic diversity and ecological flows. Springer-Verlag, New York, New York.

121

Nieder, W. C., E. Barnaba, S. E. G. Findlay, S. Hoskins, N. Holochuck, and E. A. Blair. 2004. Distribution and abundance of submerged aquatic vegetation and Trapa natans in the Hudson River estuary. Journal of Coastal Research 45:150-161.

Nisbet, R. M., and W. S. C. Gurney. 1982. Modelling fluctuating populations. Jonh

Wiley & Son, New York.

Nitsche, F. O., R. Bell, S. M. Carbotte, W. B. F. Ryan, and R. Flood. 2004. Process-

related classification of acoustic data from the Hudson River Estuary. Marine Geology 209:131-145.

Noordhuis, R., D. T. van der Molen, and M. S. van den Berg. 2002. Response of

herbivorous watr-birds to the return of Chara in Lake Veluwemeer, the Netherlands. Aquatic Botany 72:349–367.

O'Connor, R. J. 2002. The conceptual basis of species distribution modeling: time for

a paradigm shift? Pages 25–33 in Scott, Heglund, Morrison, Haufler, Raphael, Wall, and Samson, editors. Predicting species occurrences. Island Press, Washington.

Odum, E. P. 1971. Fundamentals of ecology, third edition. W.B. Saunders Company,

Philadelphia.

Okubo, A., and S. A. Levin. 2001. Diffusion and ecological problems: Modern

perspectives, second edition. Springer, New York.

Olden, J. D., D. A. Jackson, and P. R. Peres-Neto. 2002. Predictive models of fish

species distributions: A note on proper validation and chance predictions. Transactions of the American Fisheries Society 131:329–336.

Ottaviani, D., G. J. Lasinio, and L. Boitani. 2004. Two statistical methods to validate

habitat suitability models using presence-only data. Ecological Modelling 179:417–443.

Pace, M. L., S. E. G. Findlay, and D. Fishcher. 1998. Effects of an invasive bivalve on

the zooplanktion community of the Hudson River. Freshwater Biology 39:103–116.

122

Pajak, P., and R. J. Neves. 1987. Habitat suitability and fish production: a model evaluation for rock bass in two Virginia streams. Transactions of the American Fisheries Society 116:839–850.

Pan, Y., and D. J. Raynal. 1995. Decomposing tree annual volume increments and

constructing a system dynamic model of tree growth. Ecological Modelling 82:299-312.

Paukert, C. P., and D. W. Willis. 2002. Seasonal and diel habitat selection by bluegills

in a shallow natural lake. Transactions of the American Fisheries Society 131:1131–1139.

Petts, G. E. 1990. The role of ecotones in aquatic landscape managment. Pages 227–

261 in Naiman and Decamps, editors. The ecology and managment of aquatic-terrestrial ecotones. UNESCO and The Parthenon Publishing Group, Paris, France.

Phillips, P. J., D. A. eckhardt, D. A. Freehafer, G. R. Wall, and H. H. Ingleston. 2002.

Regional patterns of pesticide concentration in surface waters of New York in 1997. Journal of the American Water Resources Association 38:731–745.

Pinay, G., H. Décamps, E. Chauvet, and E. Fustec. 1990. Functions of ecotone in

fluvial systems. Pages 141–169 in Naiman and Decamps, editors. The ecology and managment of aquatic-terrestrial ecotones. UNESCO and The Parthenon Publishing Group, Paris, France.

Przybylski, M., P. Biro, M. Zalewski, I. Tatrai, and P. Frankiewicz. 1991. The

structure of fish communities in streams of the northern part of the catchment area of Lake Balaton (Hungary). ACTA Hydrobiologica 33:135–148.

Pusey, B., A. H. Arthington, and M. G. Read. 1993. Spatial and temporal variation in

fish assemblage structure in the Mary River, south-eastern Queensland: the influence of habitat structure. Environmental Biology of Fishes 37:355–380.

Rakocinski, C. F., D. M. Baltz, and J. W. Fleeger. 1992. Correspondence between

environmental gradients and the community structure of marsh-edge fishes in a Louisiana estuary. Marine Ecology-Progress Series 80:135–148.

Reyes, E., F. H. Sklar, and J. W. Day. 1994. A regional organism exchange model for

simulating fish migration. Ecological Modelling 74:255–276.

123

Rhoads, D. C., and S. Cande. 1971. Sediment profile camera for in situ study of organism-sediment relations. Limnology and Oceanography 16:110–114.

Rhoads, D. C., and J. D. Germano. 1982. Characterization of organism-sediment

relations using sediment profile imaging: an efficient method of remote ecological monitoring of the seafloor (RemotsTM System). Marine EcologyProgress Series 8:115–128.

Ries, L., and D. M. Debinski. 2001. Butterfly responses to habitat edges in the highly

fragmented prairies of central Iowa. Journal of Animal Ecology 70:840–852.

Ries, L., R. J. Fletcher, Jr., J. Battin, and T. D. Sisk. 2004. Ecological responses to

habitat edges:mechanisms, models, and variability explained. Annual Reviews of Ecology, Evolution, and Systematics 35:491–522.

Risser, P. G. 1990. The ecological importance of land-water ecotones. Pages 7–21 in

Naiman and Decamps, editors. The ecology and managment of aquatic-terrestrial ecotones. Parthenon Publishing Group, Paris, France.

Ristich, S. S., M. Crandall, and J. Fortier. 1977. Benthic and epibenthic

macroinvertebrates of the Hudson River: Distribution, natural history and community structure. Estuarine, Coastal and Shelf Science 5:255–266.

Ross, J. A., S. F. Matter, and J. Roland. 2005. Edge avoidance and movement of the

butterfly Parnassius smintheus in matrix and non-matrix habitat. Landscape Ecology 20:127–135.

Rundle, S. D., M. J. Attrill, and A. Arshad. 1998. Seasonality in macroinvertebrate

community composition across a neglected ecological boundary, the freshwater-estuarine transition zone. Aquatic Ecology 32:211–216.

Schiemer, F., M. Zalewski, and J. E. Thorpe. 1995. The importance of aquatic-

terrestrial ecotones for freshwater fish. Hydrobiologia 303.

Schmidt, R. E., and S. Cooper. 1996. A catalog of barriers to upstream movement of

migratory fishes in Hudson River tributaries. Final report to Hudson River Foundation, New York.

124

Schmidt, R. E., R. J. Klauda, and J. M. Bartels. 1988. Distributions and movements of the early life stages of three species of alosa in the Hudson River, with comments on mechanisms to reduce interspecific completition. Pages 193–215 in Smith, editor. Fisheries research in the Hudson River. State University of New York Press, Albany, New York.

Sisk, T. D., and N. M. Haddad. 2002. Incorporating the effects of habitat edges into

landscape models: Effective area models for cross-boundary management. Pages 208–240 in Liu and Taylor, editors. Integrating landscape ecology into natural resource management. Cambride University Press, Cambridge, UK.

Sisk, T. D., and C. R. Margules. 1993. Habitat edges and restoration: methods for

quantifying edge effects and predicting the results of restoration efforts. Pages 45–55 in Saunders, Hobbes, and Ehrlich, editors. Nature conservertion 3: Reconstruction of fragmented ecosystems. Surrey Beatty and Sons Pty Limited Chipping Norton, New South Wales, Australia, Chipping Norton, New South Wales.

Skalski, G. T., and J. E. Gilliam. 2000. Modeling diffusive spread in a heterogeneous

population: a movement study with stream fish. Ecology 81:1685–1700.

Smith, C. L. 1985. The inland fishes of New York State. New York State Department

of Environmental Conservation, Albany, New York.

Sparrevohn, C. R., A. Nielsen, and J. G. Stottrup. 2002. Diffusion of fish from a single

release point. Canadian Journal of Fisheries and Aquatic Science 59:844–853.

Stamps, J. A., M. Buechner, and V. V. Krishnan. 1987. The effects of edge

permeability and habitat geometry on emigration from patches of habitat. American Naturalist 129:533–552.

Stanne, S. P., R. G. Panetta, and B. E. Forist. 1996. The Hudson. Rutgers University

Press, New Brunswick, New Jersey.

Sterman, J. D. 2000. Business dynamics: Systems thinking and modeling for a

complex world. McGraw-Hill, Singapore.

Strayer, D. L., N. F. Caraco, J. J. Cole, S. Findlay, and M. L. Pace. 1999.

Transformation of freshwater ecosystems by bivalves. BioScience 49:19–27.

125

Strayer, D. L., K. A. Hattala, and A. W. Kahnle. 2004. Effects of an invasive bivalve (Dreissena polymorpha) on fish in the Hudson River estuary. Canada Journal of Fisheries and Aquatic Science 61:924–941.

Strayer, D. L., J. Powell, P. Ambrose, L. C. Smith, M. L. Pace, and D. T. Fischer.

1996. Arrival, spread, and early dynamics of a zebra mussel (Dreissena polymorpha) population in the Hudson River estuary. Canada Journal of Fisheries and Aquatic Science 53:1143–1149.

Strayer, D. L., M. E. Power, W. F. Fagan, S. T. A. Pickett, and J. Belnap. 2003. A

classification of ecological boundaries. BioScience 53:723–729.

ter Braak, C. J. F., and P. Smilauer. 1998. Canoco reference manual and user's guide

to Canoco for windows: Software for canonical community ordination (version 4). Microcomputer Power, Ithaca, New York.

Texas Instruments, I. 1976. 1975 Water qualtity standard operating procedures.

Consolidated Edison Company of New York, New York.

Texas Instruments, Inc. 1977. 1974 year-class report for the multiplant impact study of

the Hudson River estuary. Consolidated Edison Company of New York, New York.

Texas Instruments Inc. 1981. 1979 year class report for the muliplant inmpact study of

the Hudson River estuary. Consolidated Edison Company of New York, New York.

Tilman, D., and P. Kareiva, editors. 1997. Spatial ecology: the role of space in

population dynamics and interspecific interaction. Princeton University Press, Princeton, New Jersey.

Trautman, M. B. 1957. The fishes of Ohio. Ohio State University Press, Columbus,

Ohio.

Turchin, P. 1998. Quantitative analysis of movement: measuring and modeling

populaiton redistribution in animals and plants. Sinauer Associates, Inc., Sunderland, Massachusetts.

126

Vernberg, W. B., and F. J. Vernberg. 1976. Physicological adaptations of estuarine animals. Oceanus 19:48–54.

Wagner, C. M. 1999. Expression of the estuarine species minimum in littoral fish

assemblages of the lower Chesapeake Bay tributaries. Estuaries 22:304–312.

Wall, G. R., and P. J. Phillips. 1998. Pesticides in the Hudson River basin, 1994–96.

Northeastern Geology and Environmental Sciences 20: 299–307.

Ward, J. V., and J. A. Stanford. 1983. The serial discontinuity concept of lotic

ecosystems. Pages 347–356 in Fontaine and Bartell, editors. Dynamics of lotic ecosystems. Ann Arbor Science Publishers, Ann Arbor, MI.

Watson, J. E. M., R. J. Whittaker, and T. P. Dawson. 2004. Habitat structure and

proximity to forest edge affect the abundance and distribution of forest-dependent birds in tropical coastal forests of southeastern Madagascar. Biological Conservation 120:311–327.

Weinstein, M. P., S. L. Weiss, and M. F. Walters. 1980. Multiple determinants of

community structure in shallow marsh habitats, Cape Fear River estuary, North Carolina. Marine Biology 58:227–243.

Wells, A. W., and J. R. Young. 1992. Long-term variability and predictability of

Hudson River physical and chemical characteristics. Pages 29–58 in Smith, editor. Estuarine research in the 1980s. State University of New York Press, Albany, New York.

Wells, L. 1968. Seasonal depth distribution of fish in southeastern Lake Michigan.

Fisheries Bulletin 67, U.S Fish and Wildlife Service.

Whitfield, A. K. 1999. Ichthyofaunal assemblages in estuaries: A South African case

study. Reviews in Fish Biology and Fisheries 9:151–186.

Whittaker, R. H. 1975. Communities and Ecosystems, second edition. Macmillan

Publishing, New York.

Wiens, J. A. 1992. Ecological flows across landscape boundaries: a conceptual

overview. Pages 217–235 in Hansen and di Carstri, editors. Landscape boundaries:

127

Consequences for biotic diversity and ecological flows. Springer-Verlag, New York, New York, USA.

Wiens, J. A. 2002. Riverine landscapes: taking landscape ecology into the water.

Freshwater Biology 47:501–515.

Wiens, J. A., C.S.Crawford, and J.R.Gosz. 1985. Boundary dynamics: a conceptual

framework for studying landscape ecosystems. Oikos 45:421–427.

Wiens, J. A., N. C. Stenseth, B. Van Horne, and R. A. Ims. 1993. Ecological

mechanisms and landscape ecology. Oikos 66:369–380.

Willis, T. V., and J. J. Magnuson. 2000. Patterns in fish species composition across the

interface between streams and lakes. Canadian Journal of Fisheries and Aquatic Science 57:1042–1052.

Winemiller, K. O., and M. A. Leslie. 1992. Fish assemblages across a complex,

tropical freshwater/marine ecotone. Environmental Biology of Fishes 34:29–50.

Zalewski, M. 1997. Fish diversity and ecotonal habitat. Pages 183-203 in Lachavanne

and Juge, editors. Biodiversity in land-inland water ecotones. UNESCO and The Parthenon Publishing Group, Paris.

Zalewski, M., J. E. Thorpe, and R. J. Naiman. 2001. Fish and riparian ecotones-a

hypothesis. Ecohydrology and Hydrobiology 1:11–14.