Effects of Lake Erie Hypoxia on Fish Habitat Quality and Yellow Perch Behavior and Physiology Leah Zoe Ameida Purdue University

Purdue University Purdue e-Pubs

Open Access Theses Theses and Dissertations

4-2016 Effects of Lake Erie hypoxia on fish habitat quality and yellow perch behavior and physiology Leah Zoe Ameida Purdue University

Follow this and additional works at: https://docs.lib.purdue.edu/open_access_theses Part of the Forest Sciences Commons

Recommended Citation Ameida, Leah Zoe, "Effects of Lake Erie hypoxia on fish habitat quality and yellow perch behavior and physiology" (2016). Open Access Theses. 741. https://docs.lib.purdue.edu/open_access_theses/741

This document has been made available through Purdue e-Pubs, a service of the Purdue University Libraries. Please contact [email protected] for additional information. Graduate School Form

30 Updated ¡ ¢¡£ ¢¡¤ ¥

PURDUE UNIVERSITY GRADUATE SCHOOL Thesis/Dissertation Acceptance

This is to certify that the thesis/dissertation prepared

By Leah Zoe Almeida

Entitled EFFECTS OF LAKE ERIE HYPOXIA ON FISH HABITAT QUALITY AND YELLOW PERCH BEHAVIOR AND PHYSIOLOGY

For the degree of Master of Science

Is approved by the final examining committee:

Tomas Höök Chair Marisol Sepúlveda

Stuart Ludsin

To the best of my knowledge and as understood by the student in the Thesis/Dissertation Agreement, Publication Delay, and Certification Disclaimer (Graduate School Form 32), this thesis/dissertation adheres to the provisions of Purdue University’s “Policy of Integrity in Research” and the use of copyright material.

Approved by Major Professor(s): Dr. Tomas O. Höök

Approved by: Dr. Robert K. Swihart 4/11/2016 Head of the Departmental Graduate Program Date

EFFECTS OF LAKE ERIE HYPOXIA ON FISH HABITAT QUALITY AND

YELLOW PERCH BEHAVIOR AND PHYSIOLOGY

A Thesis

Submitted to the Faculty

Purdue University

Leah Zoe Almeida

In Partial Fulfillment of the

Requirements for the Degree

Master of Science

May 2016

Purdue University

West Lafayette, Indiana

ACKNOWLEDGEMENTS

I would like to thank my advisor, Tomas Höök, for his guidance throughout my

¡¢£¤¥¦¢ §¤¨ ¥¤¤© ¡ ¢ £ ¡ ££¤¤¤¤ ¥¢ ¡¥¢ ¤ ¤§¡¡§£¡ ¥£

Ludsin for their feedback on designing my project, analyzing and interpreting the results, and my thesis drafts. All current and many previous members of the Höök lab including

Dave Coulter, Zach Feiner, Carolyn Foley, Andrew Honsey, Margaret Hutton, Allison

Hyrcik, Nick Kalejs, Tim Malinich, Tim Sesterhenn, Sarah Stein, and Mitch Zischke provided valuable assistance in design, implementation, analysis, and presentation of my research. Thanks to Jay Beugly for his assistance with field sampling and equipment.

Sam Guffey provided a great deal of help in designing and conducting laboratory experiments, molecular techniques, and data analysis. I thank Robert Rode for his advice and help in rearing fish and conducting experiments and Tyler Krieg for helping set up experimental systems. Thanks to all involved in the NOAA EcoFore-Lake Erie Project, particularly Dan Rucinski, Tim Sesterhenn, Daisuke Goto, and Kristin Arend for providing source code and input data. I also thank my sources of funding including the

Purdue University Knox Fellowship and the National Science Foundation Graduate

Research Fellowship Program. This material is based upon work supported by the

National Science Foundation Graduate Research Fellowship under Grant No.1333468.

iii

TABLE OF CONTENTS

Page LIST OF TABLES ...... v LIST OF FIGURES ...... viii CHAPTER 1. INTRODUCTION ...... 1 CHAPTER 2. NUTRIENT LOADING EFFECTS ON FISH HABITAT QUALITY: TRADE-OFFS BETWEEN ENHANCED PRODUCTION AND HYPOXIA ...... 7 2.1 Introduction ...... 7 2.1.1 Model System ...... 10 2.2 Materials and Methods ...... 12 2.2.1 Model Description ...... 12 2.2.2 Model Scenarios ...... 15 2.3 Results ...... 16 2.3.1 Hindcasts ...... 16 2.3.2 Nutrient Loading-Habitat Quality Response Curve ...... 18 2.4 Discussion ...... 20 CHAPTER 3. BEHAVIOR AND PHYSIOLOGICAL RESPONSES OF YELLOW PERCH (Perca flavescens) TO MODERATE HYPOXIA ...... 35 3.1 Introduction ...... 35 3.2 Materials and Methods ...... 41 3.2.1 Fry Collection and Rearing ...... 41 3.2.2 Experiment 1: Willingness to Forage in Hypoxic Water ...... 42 3.2.3 Experiment 2: Physiological Response to Acute Hypoxia ...... 45 3.2.4 Experiment 3: Physiological Response to Chronic Hypoxia ...... 50 3.3 Results ...... 52

Page 3.3.1 Experiment 1: Willingness to Forage in Hypoxic Water ...... 52 3.3.2 Experiment 2: Acute Exposure ...... 53 3.3.3 Experiment 3: Chronic Exposure ...... 54 3.4 Discussion ...... 55

REFERENCES ...... 68

¡ ¡ APPENDICES 68

Appendix A Supporting Model Methods and Results ...... 88 Appendix B Experimental Data ...... 112 Appendix C Supplementary References ...... 123

LIST OF TABLES

Table ...... Page

Table 2.1. Focal species and life-stages and their habitat characteristic preferences used in modelling growth rate potential...... 27

Table 2.2. Different climate year data and nutrient loading used for each GRP model scenario...... 28

Table 3.1. Sequences of specific primer pairs used in quantitative PCR analyses...... 61

Appendix Table ...... Page

ESadult is adult emerald shiner, RGadult is adult round goby. CA, CB: mass dependence function intercept and slope, respectively for a 1 g fish; CQ: water temperature-dependent coefficient of consumption; CTO, CTM, CTL: optimum, upper, and maximum lethal water temperatures for consumption, respectively; CK1, CK4: fraction of the maximum consumption rate for CQ and CTL, respectively; RA, RB: mass dependence function intercept and slope, respectively for a 1 g fish; RQ: rate of respiration increase over low water temperatures; RTO, RTM: optimum and lethal water temperatures for respiration, respectively; ACT: activity, i.e. a constant multiplier of resting metabolism; FA, UA: proportion of consumption to egestion or excretion, respectively...... 101

-1

¡¢£ ¤¡¥¦§¡¥¨© © ¤ ¥¡ ¥ ¤ © § ¤ Appendix Table A.2 ) for all prey types for each species and age class. The species codes are as follows: YPadult is adult yellow perch, YPyoy is young-of-year yellow perch, RSadult is adult rainbow smelt, RSyoy is young-of-year rainbow smelt, ESadult is adult emerald shiner, RGadult is adult round goby...... 103

vii

Appendix Table ...... Page

Appendix Table A.3 Prey selectivity values and equations for all prey types for each species and age class. The species codes are as follows: YPadult is adult yellow perch,

YPyoy is young-of-year yellow perch, RSadult is adult rainbow smelt, RSyoy is young- of-year rainbow smelt, ESadult is adult emerald shiner, RGadult is adult round goby. . 104

Appendix Table B.1 Experiment 1 movement and time in treatment data...... 112

Appendix Table B.2 Experiment 2 (Acute Exposure) fold data...... 115

Appendix Table B.3 Experiment 3 (Chronic Exposure) growth, consumption, and [Hb] data...... 118

Appendix Table B.4 Experiment 3 (Chronic Exposure) fold data...... 119

viii

LIST OF FIGURES

Figure ...... Page

Figure 2.1. A schematic of the growth rate potential (GRP) model including input data

from Rucinski et al. (2014), prey distributions submodel (described in Appendix A1) and

bioenergetics submodel (described in Appendix A2). Arrows indicate which model

output or variables are used in the following model...... 29

Figure 2.2. Growth rate potential (GRP) model output at various time scales. Adult

yellow perch (YPadult) habitat quality (GRP) fluctuates seasonally, daily, and sub-daily

due to changes in physical (e.g., light) and biological (e.g., large zooplankton) factors.

This can be observed in a representative year (e.g., 2002) and, on a shorter time-scale, in

a representative day (e.g., September 15)...... 30

Figure 2.3. Results from the hindcast simulation demonstrating (a) mean annual

temperatures (°C) and annual total phosphorus load (MT); (b) number of hypoxic cells

-1 -1

¢ £ ¤¥¦ §¨© ¡ ¢ £ ¥¦ § § § §¨ ¨ ¡ )

-1

¨ © §¨© ) per year; (d) percent positive GRP cells per year for all species

and life-stages; (e) GRP-99% per year for all species and life-stages. The species codes

are as follows: ES adult is adult emerald shiner, YP YOY is young-of-year yellow perch,

RS YOY is young-of-year rainbow smelt, YP adult is adult yellow perch, RS adult is

adult rainbow smelt, RG adult is adult round goby...... 31

Figure ...... Page

-1 -1

¡¢ £¤¥¦ §¨ © ¨ © Figure 2.4. ) throughout 2002 at three nutrient loads for three

species and life-stages. Column (a) nutrients × 0.1, (b) nutrients × 1.0, (c) nutrients × 1.9

¦ ¤ ¨ § ¤ £ ¤ ¤ ¨¤ ¡ ¤ t yellow perch

(YP Adult), 2) adult rainbow smelt (RS Adult), 3) adult emerald shiner (ES Adult)...... 32

Figure 2.5. The effect of increasing nutrients on percent positive GRP cells for each species and age class: young-of-year yellow perch (YP YOY) (a), young-of-year rainbow smelt (RS YOY) (b), adult emerald shiner (ES Adult) (c), adult yellow perch (YP Adult)

(d), adult rainbow smelt (RS Adult) (e), adult round goby (RG Adult) (f). Different gray lines represent different climate years (1987 to 2005), and the black line represents the mean across all climate years...... 33

Figure 2.6. The effect of increasing nutrients on GRP-99% for each species and age class: young-of-year yellow perch (YP YOY) (a), young-of-year rainbow smelt (RS YOY) (b), adult emerald shiner (ES Adult) (c), adult yellow perch (YP Adult) (d), adult rainbow smelt (RS Adult) (e), adult round goby (RG Adult) (f). Different gray lines represent different climate years (1987 to 2005), and the black line represents the mean across all climate years...... 34

Figure 3.1. Schematic of experimental unit for Experiment 1. Depth of tanks is 24 cm. 62

Figure ...... Page

Figure 3.2. Hemoglobin concentrations in Experiment 2 (acute) for the control and

¢£¤ ¥ ¦ §¤¢£¨© § §¤¢£¨© § §§¨ £ § ¥§ ¢£ ¡ pulations. Center line in

box plots shows median, and upper and lower edges of the box are the 25th and 75th

percentiles, and error bars are the 10th and 90th percentiles...... 63

! " #$% & '( Figure 3.3. Median fold changes of hif-, igf- expressions

in the brain (a), gill (b), liver (c), and muscle (d) for Erie and NC populations in

¤¢§ ¥ §) ¦©§¦£ £* ¡¢£¤ ¥ ¦ §¤¢£¨© § §¤¢£¨© § § §¨+ £ , ¨

indicate the 95 % confidence interval (CI)...... 64

! " #$% & '( Figure 3.4. Median fold changes of hif-, igf- expression

in the brain (a), gill (b), liver (c), and muscle (d) for Erie and NC populations in

¤¢§ ¥ §) ¦©§¦£ £* ¡¢£¤ ¥ ¦ §¤¢£¨© § §¤¢£¨© § § §¨+ £ , ¨

indicate the 80 % confidence interval (CI)...... 65

! " #$% & '( Figure 3.5. Median fold changes of hif-, igf- expression

in the brain (a), gill (b), liver (c), and muscle (d) for Erie and NC populations in

Experiment 3 (chronic) control, moderate, and low treatments. Error bars indicate the

95 % confidence interval (CI)...... 66

! " #$% & '( Figure 3.6. Median fold changes of hif-, igf- expression

in the brain (a), gill (b), liver (c), and muscle (d) for Erie and NC populations in

Experiment 3 (chronic) control, moderate, and low treatments. Error bars indicate the

80 % confidence interval (CI)...... 67

Appendix Figure ...... Page

Appendix Figure A.1 Graphical depictions of the temperature (a), light (b), and DO (c) functions in zooplankton redistribution...... 105

Appendix Figure A.2 Daily changes in physical and biological variables and in GRP for the day of September 15, 2002. Includes a) temperature (°C), b) dissolved oxygen (mg ·

L-1), c) light (ly/day), d) small zooplankton (mg · L-1), e) large zooplankton (mg · L-1), f) chironomids (g · L-1), g) GRP (g · g-1 · d-1) for adult yellow perch, h) GRP (g · g-1 · d-1) for adult rainbow smelt, i) GRP (g · g-1 · d-1) for adult emerald shiner...... 106

Appendix Figure A.3 Physical and biological variables at different time periods throughout the year. Column (a) dawn, (b) midday, (c) dusk, and (d) midnight. Row 1) temperature (°C), 2) dissolved oxygen (mg · L-1), 3) light (ly · day-1), 4) small zooplankton (mg · L-1), 5) large zooplankton (mg · L-1), 6) chironomids (g · L-1)...... 107

-1 -1 ¥¦ Appendix Figure A.4 ¡¢ £¤ ¥ ¤ ) at different time periods throughout the year.

Column (a) dawn, (b) midday, (c) dusk, (d) midnight. Row 1) adult yellow perch GRP,

2) adult rainbow smelt GRP, 3) adult emerald shiner. The black line throughout the graphs indicate the cell where the highest GRP is located in the water column at that time...... 108

xii

Appendix Figure ...... Page

Appendix Figure A.5 Results from the hindcast simulation demonstrating average GRP per year for all species and life-stages. The species codes are as follows: ES adult is adult emerald shiner, YP YOY is young-of-year yellow perch, RS YOY is young-of-year rainbow smelt, YP adult is adult yellow perch, RS adult is adult rainbow smelt, RG adult is adult round goby...... 109

-1

§¨§¢¦ © ¡ ¢ ©© ¥¨©¥ Appendix Figure A.6 ¡ ¢ £ ¤¢ ¥¦ ) at dawn (top row),

-2

§ ¤¤ £¡ © ¤¦ ¡© ¥¤ §¤¥§ ¨ ¡ © ¥¤ © §¡ ©¥©§¤ ¢ ¡ ¢

(bottom row) at various nutrient loads: 0.1 nutrients (column 1), 1.0 nutrients (column 2), and 1.9 nutrients (column 3)...... 110

Appendix Figure A.7 The effect of increasing nutrients on average GRP for each species and age class: young-of-the-year yellow perch (a) young-of-the-year rainbow smelt (b) adult emerald shiner (c) adult yellow perch (d) adult rainbow smelt (e), adult round goby

(f). The different gray lines represent different climate years (1987 to 2005), and the black line represents the mean across all climate years...... 111

Appendix Figure B.1 Differences in initial total length for Erie and NC populations in (a)

! " #

¢¡§ ¢¥¨ ¨¢ ©¥¨¡© ¥¤ £© §¢ ©¦ ¡¢ ¢ ©¦ ¡¢ ¨¡¢ ¨¢¥¨ ¦ ¥¤

Experiment 3 (chronic) control, moderate, and low treatments. Center line in box plots shows median, and upper and lower edges of the box are the 25th and 75th percentiles, and error bars are the 10th and 90th percentiles...... 122

xiii

ABSTRACT

Almeida, Leah Zoe M.S., Purdue University, May 2016. Effects of Lake Erie Hypoxia on Fish Habitat Quality and Yellow Perch Behavior and Physiology. Major Professors: Tomas O. Höök.

Hypoxia is a concern in freshwater, marine, and estuarine systems worldwide including in Lake Erie, North America. Hypoxia may develop more frequently and for a longer duration as a result of excess nutrient loading from human sources, therefore, the incidence of hypoxic areas has increased with human population increases and intensive land-use practices. The effects of hypoxia on fish and other aquatic organisms have been well-studied, but most research examines the negative aspects of hypoxia development on habitats and the effect of severe hypoxia on individuals. In this study we focused on less studied and more subtle effects of hypoxia, including the possible benefits of prey increases that can occur concomitantly with increased hypoxia and the behavioral and physiological responses of yellow perch (Perca flavescens) to moderate hypoxia. First, we examined the balance between the potential beneficial and detrimental aspects of nutrient loading and hypoxia on habitat quality for four fishes and two life-stages using a growth rate potential model (GRP). Although nutrient loading did increase prey biomass in model simulations, it also caused a decline in overall habitat quality due to the spatial and temporal expansion of hypoxia. Nutrient loading also increased habitat quality, but

xiv only in areas where habitat quality was already above average and primarily for species and life-stages that preferred warmer, pelagic habitat conditions. Habitat quality for species that preferred cooler, benthic habitat conditions declined relatively steeply at intermediate nutrient loading levels. Therefore, although nutrient loading may benefit some species by increasing prey resources, hypoxia may alter the ability of many species, particularly benthic species, from accessing peaks in prey or other advantageous resources in the bottom of the water column. The second aspect of our study examined how moderate hypoxic concentrations affected the behavior and physiology of yellow perch. We found modest differences in the expression of some genes that may be altered in response to hypoxia, however, moderate hypoxia did not affect the majority of behavioral and physiological responses examined. This may indicate that yellow perch can utilize areas of moderate hypoxia and the resources therein without many deleterious effects. Overall, our studies demonstrated that the responses of individuals and ecosystems to hypoxia and nutrient loading are complex. Examining fish and ecosystem responses to changes in habitat conditions like hypoxia can improve understanding of current and future large-scale environmental changes.

CHAPTER 1. INTRODUCTION

Hypoxic conditions, low oxygen concentrations, often develop in aquatic systems as a result of cultural eutrophication. Nutrients entering a body of water may increase primary productivity, and high densities of primary producers may ultimately increase biological oxygen demand in the form of respiration by primary producers, consumers, and perhaps most importantly, decomposers. If little mixing exists (e.g., in lakes covered by ice, stagnant wetlands, floodplains, or thermally or salinity stratified bottom waters), high respiration rates can cause water to become hypoxic (Greenbank, 1945; Diaz, 2001;

Townsend & Edwards, 2003). While this process can occur naturally, the addition of excess nutrients from human activities is causing an increase in the worldwide abundance and severity of hypoxic zones (Diaz & Rosenberg, 2008; Zhang et al., 2010).

For aquatic habitats, a potential trade-off exists with eutrophication between increased productivity and the development of hypoxia (Caddy, 1993). Increases in

nutrients can increase prey biomass (Smith et al., 1981; Elmgren, 1989; Blumenshine et

¢£ ¤¥¦§¨ ©¥ §¤¤§ ¨ ¤ ¤¥ ¡¦¨¤ ¢ ©¦ ¡¤¦ § ¡ § ¥ al., 1997) ¡ - igher trophic levels. However, elevated nutrient loads may also increase the occurrence and spatio-temporal extent of hypoxia.

For most fin-fish, attempted avoidance is the first response to hypoxic conditions

(Suthers & Gee, 1986; Prince & Goodyear, 2006; Pollock et al., 2007; Chapman &

Mckenzie, 2009; Pothoven et al., 2012). While species are directly affected by hypoxia through mortality events, most mobile species experience indirect effects of hypoxia when they alter their behavior and distribution to avoid hypoxia (Breitburg, 2002;

Pollock et al., 2007). Distributional changes due to hypoxia may cause fish to experience novel environmental conditions, such as non-preferred temperature or light. Novel habitats may lead to effects such as altered feeding and trophic interactions (Pihl, 1994;

Shoji et al., 2005; Ludsin et al., 2009), which can affect growth rates (Eby et al., 2005), and may potentially alter population vital rates and ultimately influence community composition (Howell & Simpson, 1994; Ludsin et al., 2001).

If fish do not avoid hypoxia, they may engage a variety of physiological mechanisms to prolong survival and minimize the negative consequences of reduced oxygen availability. Increased concentration and binding affinity of hemoglobin is one way in which fish can extract more oxygen from the environment (Powers, 1980; Wells,

2009), and, by decreasing metabolic rate, foraging, and food consumption, an individual can reduce its activity and the amount of oxygen consumed (Poon et al., 2002; Richards,

2009). However, responses such as these to hypoxia are often species-specific.

Many studies have examined the negative effects of hypoxia on habitat quality

(e.g., Pihl, 1994; Arend et al., 2011; Scavia et al., 2014; Hughes et al., 2015), and numerous studies have investigated the responses of organisms to severe hypoxia (e.g.,

Gracey et al., 2001; Rimoldi et al., 2012; Roberts et al., 2012; Froehlich et al., 2015).

Few studies have explored the complex potential interaction of nutrient loading leading to both prey increases and hypoxia; and few have examined the less obvious effects of moderate hypoxia to which fish may be more likely to be exposed as compared to more

3 severe levels. To investigate these more subtle questions, we used two approaches: 1) a species-specific, spatially- and temporally-explicit bioenergetics habitat quality model for four fish species; and 2) laboratory experiments on yellow perch (Perca flavescens) behavior and physiological responses to hypoxia. For both of these approaches, we used the central basin of Lake Erie, North America, as our study system.

Lake Erie has a history of anthropogenic nutrient loading and an annually recurring hypoxic zone that is comparable to many high profile marine and estuarine hypoxic zones (Hawley et al., 2006). The central basin of Lake Erie is mesotrophic, and varying nutrient loading levels have caused the size of the annual hypoxic zone as well as the fish community to fluctuate (Ludsin et al., 2001; Burns et al., 2005; Hawley et al.,

2006; Scavia et al., 2014). Changes in climate over recent years (e.g., warmer temperatures and increased frequency of severe storm events) may also influence the

Lake Erie hypoxic zone and the response of fish and other species to low oxygen

(Michalak et al., 2013; Scavia et al., 2014).

Yellow perch are common and economically important in Lake Erie (Hushak et al., 1988a). They are also relatively tolerant of low oxygen conditions (Petrosky &

Magnuson, 1973), potentially indicating that they encounter hypoxic conditions more frequently as they are less likely to completely avoid the use of such habitat (Hasler,

1945; Roberts et al., 2012). Understanding how this species responds to a range of hypoxic conditions will provide important information for how species with moderate hypoxic tolerances may be responding to environmental conditions in Lake Erie and other similar habitats. Together, these approaches add to current knowledge about hypoxic habitats and the responses of species which inhabit them.

In our habitat quality model of Lake Erie, we observed predicted increases in prey biomass and hypoxia with increased nutrient loading, simultaneously driving beneficial and detrimental changes in habitat quality. Invertebrate prey biomasses (i.e., zooplankton

and chironomids) increased over a range of 0.1 1.9 factor increases in annual nutrient loading over baseline. Concurrently, the vertical spatial extent and duration of the hypoxic zone also increased. Overall, annual habitat quality for most species and life- stages declined slightly with increasing nutrient loads. Habitat quality did increase in parts of the water column and during periods of time where habitat quality was already above average (peak habitat), corresponding to areas and times of peaks in prey biomass.

Species and life-stages that prefer pelagic food sources (zooplankton) and warmer temperatures experienced the greatest increases in peak habitat quality. Alternatively, species and life-stages that prefer benthic food sources (chironomids) and cooler temperatures experienced some increases in peak habitat quality at low and high nutrient loadings; but at moderate nutrient loading, habitat quality declined sharply. Although we observed these general responses to factor load change in annual nutrient loading, a greater amount of variability in the habitat quality response occurred between years as a result of annual meteorological conditions (e.g., timing of loading, temperature, timing and depth of stratification) than due to changes in nutrient levels. Our results indicate that while increased nutrient loading will likely affect species to different degrees based on habitat preferences, annual climate conditions may mask nutrient-related responses of habitat quality.

Individual yellow perch (i.e., foraging behavior and physiological responses to acute and chronic hypoxia) demonstrated little to no response to moderate hypoxic

5 conditions. Although previous studies have demonstrated yellow perch and conspecific

Eurasian perch (Perca fluviatillis) respond to severe hypoxia (< 2.0 mg/L) by decreasing

consumption and growth and increasing expression of certain proteins (e.g., hif- ¡), our results indicate that young-of-year yellow perch may be able to satisfy their oxygen needs in moderate hypoxia through other means or other mechanisms, at least at the durations tested in our experiment. If our results accurately depict the yellow perch response to moderate hypoxia, yellow perch may be able to utilize larger area with moderate oxygen concentration during hypoxic events.

While our study advanced knowledge about the subtle effects of nutrient loading and hypoxia, a great deal remains unknown. Our model only examined vertical habitat quality, however, fish will utilize habitat horizontally, potentially inhabiting an area of moderate hypoxia bordering severe hypoxia to access preferred temperatures and prey.

Understanding how nutrient loading alters habitat horizontally would be valuable for future research and conservation goals. Our experiments attempted to only focus on the effects of moderate hypoxia, yet fish may also be exposed to additional changes in habitat, such as temperature, light, and prey, if they choose to inhabit an area of moderate hypoxia. Additionally, we did not take into account the potential change in activity that exposure to moderate hypoxia may cause in situ (Roberts et al., 2009), which could alter physiological responses.

Nutrient loading and hypoxia affect individuals and ecosystems in myriad ways, and the results of our investigations have further increased understanding of these potential complex responses. Aquatic habitats are mosaics of abiotic and biotic conditions such as prey, light, temperature, and dissolved oxygen. As anthropogenic

6 activities exacerbate nutrient loading into aquatic habitats, the dynamic interaction of habitat conditions may more negatively affect certain species and life-stages than others.

Understanding the capacity of species to tolerate a range of moderate to severe hypoxia will allow for better estimation of habitat choices and resources available to each species.

Further research on the ways in which individuals and their habitats are affected by nutrient loading is needed to better understand population and community responses.

CHAPTER 2. NUTRIENT LOADING EFFECTS ON FISH HABITAT QUALITY: TRADE-OFFS BETWEEN ENHANCED PRODUCTION AND HYPOXIA

2.1 Introduction

Globally, increases in nutrient loading from anthropogenic sources and rising human populations have led to an expansion of hypoxic zones, areas with depleted

-1

£¤¥¦¢§¨© ¢ ¤ ¨¥¤ ¨ ¤ ¡ ¢£¤¥ £ ¡ ¢ (Diaz, 2001). Hypoxia develops when excess nutrients (nitrogen or phosphorus) cause increased production and biomass of aquatic primary producers, such as algae and other phytoplankton, which eventually die and settle to the bottom of the water column as detritus. In turn, high respiration during the decomposition of this detritus in bottom waters depletes available oxygen in the surrounding water. If the water column is stratified, the bottom, oxygen- depleted water cannot fully mix with the upper oxygenated water and hypoxic conditions may establish (Diaz, 2001). While hypoxia develops due to natural processes in many bodies of water, the addition of nutrients from human activities can exacerbate this phenomenon. Increased nutrient loading from rising human populations and global climate change are likely to expand the spatial extent and annual duration of current hypoxic zones and cause new zones to be formed in freshwater, estuarine, and coastal marine systems (Diaz, 2001; Diaz & Rosenberg, 2008; Zhang et al., 2010).

The excess nutrients that cause the development of hypoxia can alter aquatic ecosystems through myriad pathways. Greater levels of nutrients have been shown to

8 increase biomass of phytoplankton, zooplankton, and benthic invertebrates, potentially providing more food for fish as demonstrated in freshwater (Blumenshine et al., 1997), estuarine (Smith et al., 1981), and marine systems (Elmgren, 1989). However, many studies focusing on effects of nutrient loading on fish production do not account for the potential positive influences of additional nutrients (e.g., Pihl et al., 1991; Howell &

Simpson, 1994; Roberts et al., 2009; Arend et al., 2011; but see Breitburg et al., 2009). In contrast, additional nutrient loading may not lead to proportional increases in fish biomasses for various reasons. For example, high nutrient concentrations may favor inedible forms of phytoplankton which are not readily consumed and transferred to higher trophic levels (e.g., cyanobacteria, see Paerl & Fulton, 2006). In addition, hypoxia exacerbated by excessive nutrient loading can cause direct mortality by asphyxiation, or, more frequently, indirectly affect fish as hypoxic conditions cause them to alter behaviors or to disperse to novel habitats to avoid these conditions (Breitburg, 2002; Pollock et al.,

2007). While several past studies have evaluated the potential effects of hypoxia on fish growth (e.g., Eby et al., 2005), trophic interactions (e.g., Pihl, 1994; Shoji et al., 2005;

Ludsin et al., 2009), recruitment (Hughes et al., 2015), and community composition (e.g.,

Howell & Simpson, 1994; Ludsin et al., 2001), most have only focused on deleterious effects of hypoxia and ignored the potentially important benefits of coincident prey increases.

Ultimately, the effect of increased nutrient loading on fish production may be species- and system-specific, depending upon species environmental tolerances and existing productivity levels. For example, planktivorous fish may benefit more immediately from increased nutrient loading than piscivorous fish (see Jeppesen et al.,

2000), and the interaction between physical characteristics of lakes or semi-enclosed seas

(e.g., circulation, flushing rates) and nutrient loading can affect the development of hypoxia (see Silva & Vargas, 2014; Avramidis et al., 2015), where habitat quality may increase with increased nutrient loading in oligotrophic systems but decline with additional loading in meso- or eutrophic systems that become increasingly nutrient- saturated and hypoxic (Caddy, 1993; Paerl & Fulton, 2006). Finally, species-specific effects of nutrient loading and subsequent hypoxia are likely regulated by the vertical structure of aquatic systems and corresponding species distributions. Many studies of aquatic communities within hypoxic zones have documented decreased abundances of fish where oxygen is depleted (e.g., Pihl et al., 1991; Petersen & Pihl, 1995; Eby et al.,

2005); but, other studies have linked increases in nutrients with increases in fish production in areas vertically or horizontally adjacent to hypoxic areas (see reviews by

Caddy, 1993, 2000; Breitburg et al., 2009).

The effects of nutrient loading in aquatic ecosystems can be evaluated across multiple temporal scales. Interannual variation in the size, duration, and magnitude of hypoxic zones has been attributed to temperature and large-scale climate variability (e.g.,

Pacific Decadal Oscillation, Deutsch et al., 2011; El Niño Southern Oscillation, Hughes et al., 2015). Therefore, increased nutrient loading may have a stronger effect on oxygen conditions in one year than another due to the timing of nutrient loading, thermal variation, wind, or other climatic variables influencing characteristics of stratification

(however, see Li et al., 2016). At seasonal timescales, hypoxia may or may not overlap with critical periods for fish growth, reproduction, and survival, thus influencing the effect of nutrients on fish habitat quality. For example, the annually occurring hypoxic

10 zone in Lake Erie, North America, can temporally overlap with both the annual peak of chironomid larva density in bottom substrates and peak summer water temperatures, increasing the negative effects of hypoxia on the habitat quality of species that prefer to consume chironomid larvae and species that prefer cooler temperatures (Arend et al.,

2011). On daily or subdaily scales, many species of fish and zooplankton exhibit diel vertical migration (DVM), moving higher or lower in the water column throughout the day allowing them to follow preferred physical (e.g., light) and biological (e.g., prey, predator avoidance) conditions (Lampert, 1989; Mehner, 2012). Hypoxia has been demonstrated to disrupt these patterns in fish (Baldwin et al., 2002; Ludsin et al., 2009;

Vanderploeg et al., 2009a), potentially providing new refuges for prey (Ludsin et al.,

2009) or increasing overlap between predators and prey (Vanderploeg et al., 2009a).

2.1.1 Model System

In this study, we examined the effects of nutrient loading, primarily phosphorus, on habitat quality using the central basin of Lake Erie, USA, as a model system. Lake

Erie supports large recreational and commercial fisheries (Hushak et al., 1988b) and its hypoxic zone rivals or exceeds the size of most marine or estuarine hypoxic zones

(Hawley et al., 2006). The central basin of Lake Erie is mesotrophic; therefore, the response of this system to additional nutrient loading is unclear. Like other coastal areas

with hypoxia, Lake Erie has experienced decades of anthropogenic nutrient loading

¢£ ¤ ¥¦§¡ ¦ ¡¤¢¡¢ ¨© ¡ © ¤ ¨© ©¢ §¦¦ ¡¢ ¦¢¢ ¦¨¤ ¡ ¤¢© contribu ¡ (Burns et al., 2005; Hawley et al., 2006; Scavia et al., 2014) and concomitant fluctuations in the abundance of hypoxia-intolerant fish species (e.g., burbot, Lota lota; white sucker,

Catostomus commersoni; Ludsin et al., 2001). Recent increases in the size of the summer

hypoxic zone are potentially due in part to warmer temperatures extending the period of

stratification in the summer, greater frequency of storm events increasing nutrient runoff

from the surrounding watershed, and in particular increased run-off of biologically

available, reactive forms of phosphorous (Michalak et al., 2013; Scavia et al. 2014). As

climate change further alters nutrient loading and hypoxic zones in Lake Erie,

investigations on the effects of additional nutrients on fish habitat quality could provide

predictions for the future responses of similarly threatened systems throughout the globe.

We evaluated how nutrient loading may affect habitat quality of four fish species

inhabiting the central basin of Lake Erie, using a one-dimensional (1D) water quality

model (Rucinski et al., 2014) to provide habitat parameters for a bioenergetics growth

rate potential (GRP) model. These types of habitat quality models have been used

extensively (Mason & Brandt, 1996; Roy et al., 2004; Budy et al., 2011), including in

Lake Erie (Arend et al., 2011; Brandt et al., 2011), to consider how physical and

¡¢£¢¤¡¥¦£§¦¨¡¦ £© ¡ ¦ © ¡ © ¦¨©¦¢ ¦ ¡¦ ¦ ¦ ©¥¦ ¢©¡¥¦£ ¡ © ©¨¤

budget, i.e., a species-specific index of habitat quality. We specifically used our GRP

model to create response curves over a range of nutrient factors, which demonstrate how

¦ ©¥¡© ¨ © ¢ © ¦ ¥¦ ¤©¦ ¡ ©¨¡ ¤ £©§©£ ¢ ¦ ¨¡§¡ ¤ §¦¨¡¦ £© ¡ ¡ ¥¦ ©

nutrient levels), and have been used to determine thresholds at which nutrient loading

begins to detrimentally affect aquatic ecosystems (Lyche Solheim et al., 2008; Trolle et

al., 2008; Rucinski et al., 2014; Kuiper et al., 2015). Past applications of response curves

to nutrient loading have generally not considered fish or upper food web responses (but

see Kuiper et al., 2015), and many applications have not thoroughly considered inter-

12 annual effects, despite the potential for timing of nutrient loading and annual physical conditions to strongly mediate system effects of nutrient loading (e.g., Lam et al., 1987;

Greene et al., 2009; Deutsch et al., 2011; Hughes et al., 2015).

Our GRP model builds off the coarser Lake Erie hypoxia GRP model presented in

Arend et al. (2011) by incorporating a finer temporal resolution, prey dynamics, and incremental changes in nutrient loading to examine the less-studied trade-off between increased prey availability and hypoxia on fish habitat quality. We modeled habitat quality for adult and young-of-year life-stages of four economically and ecologically important fish species that represent a range of thermal tolerances and trophic guilds at fine vertical (0.5 m) and temporal (10 min) resolution. Our objectives were to: 1) quantify habitat quality throughout the growing season in the central basin of Lake Erie, incorporating not only effects of temperature and oxygen, but also prey and light dynamics at fine temporal and vertical scales, and 2) evaluate the trade-off between forage base production and hypoxia for each species and life-stage by developing nutrient response curves.

2.2 Materials and Methods

2.2.1 Model Description

We modeled GRP, a surrogate for habitat quality, for the adult life-stages of four fish species as well as the young-of-year (YOY) life-stages for two of those species.

These fishes encompass a range of hypoxia sensitivity, warm-water and cool-water

preferences, pelagic and benthic feeders, and native and invasive species (Table 2.1). The

GRP model quantifies how spatio-temporal overlap of environmental conditions (e.g.,

temperature, oxygen, light, and pelagic and benthic prey densities) in a 1D column of

water (48 layers, each 0.5 m deep) in the central basin of Lake Erie potentially affect

habitat quality of the six fish species and life-stages (Fig. 2.1, 2.2).

The input data for the GRP model were obtained from 1D linked thermal and

eutrophication models (Rucinski et al., 2014) which, among other variables, generated

daily depth-specific water temperature (°C); daily depth-specific oxygen concentration

-1 -1

¢£ ¤ £ ¡ ); ten minute intervals of light intensity (ly day ) with depth-specific extinction

-1

¦¥ §¨©§¥ ¦¨ ¡¦¦¨ §¥¦¦¡§ ¡¢£ ¤ factors; daily ¥ ); and daily carbon

-2

¥¥¨¢ ¥¦¥ ¥¦ ¢£ m ) (Fig. 2.1). Temperature was simulated in the thermal

portion of the model, based on the Princeton Ocean Model (Blumberg & Mellor, 1987)

and driven by meteorological observations. Oxygen concentration in the eutrophication

model was determined by temperature, meteorologically controlled mixing regimes,

decomposition of organic matter, sediment interactions, and water column photosynthesis

and respiration. Light throughout the water column was determined with a base light

extinction coefficient estimated from field observations, which was adjusted based on

shading created by phytoplankton production. Zooplankton biomass responded to a

temperature-dependent mortality rate and phytoplankton production, which was

controlled by available phosphorus, light, temperature, advection, and zooplankton

grazing. Settling of carbon to sediment affected production of benthic prey (see below)

and was based on carbon loading, phytoplankton death and grazing, zooplankton death,

14 advection, and sediment interactions. Full documentation of the linked thermal and eutrophication models is presented in Rucinski et al. (2014).

To capture diurnal changes in vertical distributions, total zooplankton densities are re-distributed among habitat cells at each ten minute time step based on temperature, light, and oxygen preferences of zooplankton taxa. Additionally, zooplankton biomass is divided into small (0.2 mm), medium (0.83 mm), and large (1.25 mm) size classes based on proportional abundances of each size class observed in the field (Makarewicz et al.,

1989). Chironomid (benthic prey) biomass in the bottom 1.5 m updates daily based on modeled temperature and settling of carbon to the benthic layer (Appendix A1).

Input data are used to calculate the GRP in each 0.5 m cell during every ten minute time step using existing species-specific bioenergetics submodels (Kitchell et al.,

1977; Fig. 2.1; Table A1; Appendix A2). For each application (year) of the model, simulations begin on April 15 and end on December 29, thereby encompassing the growing season for most fish species in Lake Erie and the time periods before, during, and after hypoxia occur. Each day, the model calculates habitat conditions in every cell at

ten minute time steps (Fig. 2.1, 2.2). The model is reset at the beginning of each year, and

¡¢£ ¤ ¡¥¦ § ¥¨ © ¢ ¨¡§ ¤ ¡ ¥¦ §§ ¨ ¥¡§¥ ¦ ¥§ ¢¡ ¨ ¦§ ¨ nutrient loading, and lighting, which, in addition to daily nutrient loading, are incorporated into the thermal and water quality model to determine the environmental input variables used in the GRP model. The daily nutrient loading of each year is either applied with the respective climate year (Hindcast Scenario) or scaled to create a nutrient response curve (Nutrient Response Curve Scenario; Table 2.2).

2.2.2 Model Scenarios

Hindcasts

We modeled GRP for each species and life-stage using the nutrient loading of each respective climate year 1987-2005 (Rucinski et al., 2014). We examined relationships between physical and biological variables and GRP at sub-daily, daily, and

seasonal scales, and we assessed annual patterns of environmental conditions, percent

¡¢£¤£¥¦ §¨© ¦ ¢ ¦¦ ¦ ¢¦ £ ¤¦£¤¦¥ ¢ ¡ ¦ ¦ ¤ ¦¥¦ ¦

GRP, and GRP-99% values for all fish species and life-stages. The percent of positive

GRP cells provides an overall summary of GRP throughout a year, assuming that positive

GRP indicates habitat quality that is beneficial (Brandt et al., 2011). Average GRP is calculated as the average of the GRP from every time step and depth cell throughout a year. GRP-99%, or the 99th percentile of GRP values calculated as the average of the highest 1% (out of 17,902 cells) of all GRP cells in a year, demonstrates if nutrient loading and resulting hypoxia increased or decreased the GRP in the areas with the best

GRP in that year. Such spatial and temporal peaks in habitat quality may contribute disproportionately to annual performance of individuals (Forsman & Lindell, 1991).

Nutrient Response Curves

In order to create a nutrient response curve that accounted for annual variability in temperature, vertical mixing, timing of nutrient loads, and light, we modeled GRP for each climate year with 19 nutrient loads: 0.1 increments from 0.1 to 1.9 times the annual nutrient load during 1997, an approximately average loading year (Table 2.2) (Rucinski

16 et al., 2014). Note that while the total annual nutrient load was scaled to 1997, all other climate variables, including timing of nutrient loading, were specific to the climate year modeled. With the resulting 361 simulations, we examined how incremental changes in phosphorus loading affect hypoxia, prey production, and GRP for each species and life- stage. We used annual GRP summary indices of percent positive GRP cells, average

GRP, and GRP-99% for all species and life-stages to examine how habitat quality changed at each incremental increase of nutrient loading for each simulation year.

Percent positive GRP, average GRP, and GRP-99% were calculated as in the hindcast simulation. We determined which nutrient factor produced the maximum and minimum value of each index for each species and life-stage.

2.3 Results

2.3.1 Hindcasts

Our hindcast model integrated physical and biological factors on a ten minute time step to produce retrospective depictions of habitat quality. As an example, the year

2002 was a moderate climate year (Fig. 2.2). On a representative day after stratification and hypoxia had developed (e.g., September 15, 2002), the daily temperature and DO were constant as daily means, but light gradually increased at the surface of the water throughout the day. Light reached its maximum intensity and deepest depth at midday.

The zooplankton biomass was re-distributed in the water column each time step

17 throughout the day based on light, temperature, and DO. On 15 September 2002, adult yellow perch had higher GRP values when light was strong enough to allow for efficient visual foraging and where zooplankton was most dense. Throughout the year, physical and biological factors affect GRP in this way at ten minute time-steps, and total annual trends in habitat quality (e.g., percent positive GRP, average GRP, and GRP-99%) were determined based on these finely resolved interactions. During 2002, 37 % of cells (6552 of 17,902) displayed positive adult yellow perch GRP, the average GRP was 9.22 × 10-5

-1 -1 -1 -1

¡¢ £¤¥ ¤¦§¨© ¦§ § ¡ ¡¢ ¡ , and the GRP-99% .

Multi-year patterns of input variables demonstrated that total phosphorus (TP) loading to the central basin of Lake Erie varied among years, but showed no obvious temporal trend (Fig. 2.3a). Average annual temperature also fluctuated, but generally increased across years (Fig. 2.3a). These two factors appeared to contribute to 1) the modeled spatial and temporal extent of hypoxic conditions (Fig. 2.3b) and 2) the mean modeled densities of zooplankton and chironomid prey (Fig. 2.3c). Despite annual changes in prey availability and temperature, habitat quality remained fairly stable for most years (Fig. 2.3d, e, Fig. A5). However, for all metrics of habitat quality, there was a large decrease in 1998 and 1999 for almost all species and age-classes (Fig. 2.3d, e, Fig.

A5). In 1998, this appeared to be due to high temperatures and high TP loading causing an increase in hypoxic conditions and negatively affecting zooplankton biomass. In 1999, there were still high temperatures, but low TP loading resulting in decreased hypoxic conditions and low zooplankton and chironomid biomass.

2.3.2 Nutrient Loading-Habitat Quality Response Curve

Our response curve analyses demonstrated the importance of climatic conditions

(e.g., thermal conditions, vertical mixing, and timing of nutrient loading) on habitat quality. In fact, differences in climate year had a stronger influence on annual GRP than a

19 × range in annual nutrient loading (see Fig. 2.5; Fig. 2.6), thereby demonstrating the importance of developing response curves using multiple climate years. Despite this variation, habitat quality averaged across climate years responded to nutrient loading.

Throughout the 2002 climate year, a moderate year in terms of temperature, color contour maps of GRP for adult yellow perch (moderate response to nutrient loading), adult rainbow smelt (severe response to nutrient loading), and adult emerald shiner (mild response to nutrient loading) demonstrated the seasonal effects of altered nutrient loading

(Fig. 2.4). While changes in increasing nutrient loading caused an increase in the duration and vertical extent of the hypoxic zone, higher nutrient loading slightly positively affected GRP for all species and life-stages in early and mid-summer in the middle of the water column. The annual peaks in GRP for adult yellow perch and rainbow smelt occurred in the bottom of the water column with the lowest nutrient factor (× 0.1), but, at higher nutrient loadings, the earlier development of the hypoxic zone appears to overlap this spatiotemporal zone of high GRP and peak GRP values were instead observed earlier in the year (early June) and higher in the water column.

At an annual scale, increasing nutrient loading generally caused a gradual decrease in percentage of positive GRP cells (Fig. 2.5) as well as average GRP (Fig. A7).

Highest percent positive GRP was achieved at low nutrient loading factors for all species

19 and life-stages (except adult yellow perch) and vice versa for percent positive GRP minima. Adult yellow perch prefer chironomid prey more than other species and are

relatively tolerant of low oxygen. Therefore, their percent positive GRP may not have

¡¢ £¤¥¦¡¦§¢¨©¡¤¨©¡ ¢£¡ ¡¢ ¦ ¦ ¨¤ ¥ ¡§¢ ¥ ¥¨ ¡¥¨¢ ¡ §¢¡ © ¤¥¤ ¦ biomass increased with nutrients, and the benthic habitat did not become entirely hypoxic until nutrient loading was relatively high. Average GRP demonstrated similar trends with

most species and life-stages having maximum and minimum average GRP at × 0.1 0.2 and × 1.9 respectively. The exception was average GRP for young-of-year yellow perch, which slightly increased with nutrient loading (Fig. A7). This was most likely due to the increase in maximum habitat quality for young-of-year yellow perch. However, in both of these habitat quality metrics, climate year appeared to be more influential than nutrient load (spread in gray lines compared to change in black line in Fig. 2.5; Fig. A7).

Changes in GRP-99% were sensitive to climate year and were species specific.

Across years, pelagic and more tolerant species and size classes (young-of-year yellow perch, young-of-year rainbow smelt, and emerald shiners) exhibited increased GRP-99% with increasing nutrient loading. In contrast, more sensitive and benthic oriented species

(adult yellow perch, adult rainbow smelt, and round goby) experienced sharp declines in

GRP-99% at nutrient levels which caused hypoxia to overlap with peak chironomid abundance and warm epilimnetic temperatures (Fig. A3; Fig. 2.4). Generally, highest

GRP-99% was achieved at low nutrient loadings for adult yellow perch, adult rainbow smelt, and round goby, but high loading for young-of-year yellow perch, young-of-year rainbow smelt, and emerald shiner, and vice versa for GRP-99% minima. However,

20 differences across loading scenarios were relatively small compared to differences in

GRP-99% across different climate years (spread of gray lines in Fig. 2.6).

2.4 Discussion

Simulation results demonstrated that there are species-specific habitat quality trade-offs between increased prey and hypoxia due to nutrient loading (Caddy, 1993,

2000; Arend et al., 2011). Additionally, due to the large effect of variability in annual meteorological conditions on hypoxic zones (Lam et al., 1987; Paerl, 2006; Greene et al.,

2009; Deutsch et al., 2011; Hughes et al., 2015), habitat quality may fluctuate greatly depending on climate year conditions. Due to the complex influences of climate variables, the analysis presented here suggests that on an annual basis external nutrient loading has less of an influence than thermal regime on fish habitat quality, and thereby demonstrates how such climate variability can obscure habitat responses to changes in nutrient loading.

Response curve analyses demonstrated not only the mediating influence of annual climatic conditions, but also the non-linear and species-specific responses of fish habitat quality to changes in nutrient loading. As demonstrated by the percent of positive GRP cells, increases in nutrient loading caused areas of poor habitat quality to expand in the water column and to persist for longer periods throughout the year, causing habitat quality to decline overall. In general, benefits provided by increased nutrient loading through increases in prey were overwhelmingly observed during times and in the water

21 column where habitat quality was already advantageous for specific species and life- stages. Average GRP demonstrated how these contradictory effects slightly reduced average habitat quality for most species and life-stages.

The complex, interactive effects of increased nutrient loading on habitat quality were most obvious in the GRP-99% curves for adult yellow perch, rainbow smelt, and round goby. At very low nutrient loadings, GRP-99% of these species showed a slight positive relationship between nutrient loading and habitat quality. However, this positive relationship was interrupted by a sharp decline in GRP-99% at relatively low nutrient loading levels (i.e., loading factors of approximately × 0.2), as advantageous habitat became disadvantageous with hypoxia expansion. This response occurred at loading factors well below current nutrient loading levels. Across a range of greater nutrient loading levels (loading factors > 0.8), there was also a positive trend in GRP-99% response curves for adult yellow perch and rainbow smelt. However, at these greater nutrient loading levels, GRP-99% never reached levels observed at very low nutrient loadings. Nonetheless, if the curves were only observed at nutrient loadings above × 0.2, it would appear that increased nutrient loadings only benefit the best areas of habitat quality. Thus, depending on the history of nutrient loading to a system and the nutrient loading range across which fish population responses are considered, one could arrive at very different conclusions regarding nutrient loading and fish population relationships.

The observed trend across high nutrient loading levels may partially explain why benthic and bentho-pelagic fisheries landings can show positive relationships to nutrient inputs and are not linked to the total extent of related hypoxic zones in some systems (see review by Breitburg et al., 2009). However, fisheries landings could potentially be even

22 greater if nutrient levels were reduced to levels outside of those observed during recent times in strongly anthropogenically affected systems; a hypothesis supported by studies that have found that the size and abundance of demersal species has declined in hypoxic areas due to reduced growth or emigration (Howell & Simpson, 1994; Petersen & Pihl,

1995; Hughes et al., 2015).

GRP-99% response curves appear to be strongly influenced by timing and location of hypoxia relative to prey resources and temperatures. With increasing nutrient loading, the overlap of the hypoxic zone with the annual peak in chironomid biomass increased, encompassing the entire chironomid peak in many years and extending the amount of time cool waters in the hypolimnion were disadvantageous (Fig. 2.4; Fig. A6).

The overlap of hypoxia with benthic macroinvertebrate prey may not only limit access by mobile predators, but also decreases the health and abundance of benthic prey populations, causing habitat degradation beyond the occurrence of hypoxia (Powers et al., 2005). Evidence of the importance of prey and timing was also apparent in the hindcast analysis. The precipitous decline in zooplankton during two years strongly influenced percent positive GRP, average GRP, and GRP-99% for almost all species and life-stages, including species that generally prefer chironomids (adult yellow perch and round goby).

For fish and other organisms, peak habitat quality often indicates times of peak prey abundance and ideal environmental conditions, as was the case in our model. These areas of peak habitat quality may allow for increased growth (Forsman & Lindell, 1991), and contribute disproportionately to total annual growth. Habitat which enhances growth may also be critical for survival (Post & Evans, 1989), and reproductive success (van den

Berghe & Gross, 1989) and be a strong mediator of ecological interactions, e.g., competition (Mittelbach, 1981). Therefore, in systems similar to our modeled system,

hypoxia may lead to a spatial mismatch of seasonally important resources, and thereby

¢£¤¥¦¡§¨ £©©¡ ¤©¥ ¡ ¥¡ £¥¤£¤ £§¥¤¨ £ ©¥¤ ¡ ¡

Continued climate change will likely also affect overlap with resources by altering the timing of resource availability (Li & Smayda, 1998; see review by Durant et al., 2007). Changes in climate will likely increase the duration of thermal stratification in aquatic systems like Lake Erie, and may also increase levels of nutrient run-off from terrestrial environments by increasing the frequency of storms (Michalak et al., 2013).

Although nutrient-enriched systems may mediate some negative impacts of a warmer climate, such as increased metabolic demand (McElroy et al., 2015), it is likely that climate changes will affect the timing of peaks in prey biomasses, in addition to altering the duration, extent, and severity of hypoxic zones (Michalak et al., 2013). This may induce further spatio-temporal mismatch with resources for benthic- and pelagic-reliant species.

Given the large influence of annual climatic conditions on fish habitat quality in the response curve simulations, our results add to the discussion of the response of hypoxic systems to reduced nutrients. Hypoxic systems have demonstrated regime shifts, recovery after reaching a threshold, delayed responses, and linear recovery to nutrient reductions (Kemp et al., 2009). Nutrient loads are often the strongest factor determining water column respiration, which has been shown to be the primary mechanism controlling interannual variability in the size of hypoxic zones (Li et al., 2016); and the increase in hypoxic zones in the past century has been linked to anthropogenic activities

24 rather than climatic factors (Jenny et al., 2016). However, once a system reaches nutrient loading levels which consistently result in hypoxia, climate factors may have a greater influence than nutrients on annual variation in habitat quality. In lakes that experience anthropogenically-induced hypoxia, climate variability has been found to drive the extent of persistent hypoxic conditions after nutrient loads have been reduced, potentially indicating hysteresis, an altered ecosystem response due to the history of phosphorus loading (Jenny et al., 2014). Delayed responses may also develop due to legacy organic matter that may remain in sediments for decades (Sharpley et al., 2013). Our model resets annually and therefore cannot account for legacy sediment carbon or across year regime shifts, and hence model predictions would likely differ if multiyear nutrient loading is considered. Nonetheless, model applications demonstrate that variations in climatic conditions can greatly mediate the response of habitat quality to reduced nutrients.

We expand upon previous work (Arend et al., 2011) by including not only temperature and oxygen effects, but also the influence of light and prey on habitat at sub- daily scales. As our GRP model is highly vertically and temporally resolved, we account for the scale-dependence of GRP models, demonstrated through past analyses (Mason &

Brandt, 1996). Still, our model is a simplification of the environment. The eutrophication model we utilized uses simplified taxa to represent phytoplankton and zooplankton, which may not account for potential changes in lower trophic level composition that may occur with eutrophication (Paerl & Fulton, 2006). Additionally, by using a one- dimensional column of water, we do not consider horizontal dynamics which could affect total habitat quality of the system (e.g., horizontal migrations in response to hypoxia). In fact, we do not consider potential movement of fish at all. Integrating movement into

25 models can demonstrate how individuals respond to their habitat, but the rules that dictate movement are extremely important and incorrect assumptions and simplifications can drastically alter the outcomes (Zollner & Lima, 1999; Humston et al., 2004; Kristiansen et al., 2009). An advantage of GRP models is that they allow for fine-scale spatio- temporal quantification of habitat quality based on overlap of physical and biological characteristics (Brandt et al., 1992). This requires limited assumptions about behavior across trophic levels. Moreover, GRP measures are related to performance of simulated fish and associated with vital rates of fish in natural systems (Brandt et al., 1992). Habitat quality should indicate population performance after accounting for situations in which habitat quality and density are not linked (e.g., seasonal variations in habitat, population sinks, etc.) (Van Horne, 1983; Pulliam, 1988). Thus, model results can be used to comparatively demonstrate how populations may respond to changes in nutrient loading through changes in habitat. Nutrient loading appears to increase the quantity of poor quality habitat, but potentially increases the quality of good habitat. This may benefit some populations, but others may experience a greater mismatch with beneficial habitat conditions. Further, results suggest that life-stages respond to nutrient loading differently, which may alter demographic distributions and communities.

As human populations grow, the addition of nutrients from anthropogenic sources into bodies of water is expected to continue to increase (Diaz, 2001; Diaz & Rosenberg,

2008; Zhang et al., 2010). With climate change, there may also be various ecosystem changes coupled with increased demand for aquatic species as food sources and ecosystem changes due to climate change, thereby increasing pressure on fishery species in eutrophic water bodies (Caddy, 2000; Breitburg, 2002). Species that experience severe

26 habitat limitation due to hypoxia, such as adult rainbow smelt, as well as species that are thought to be more resilient to the effects of hypoxia, such as adult yellow perch, may fare worse as hypoxia eliminates high quality benthic habitat for a longer part of the year.

Alternatively, thermally tolerant species such as emerald shiner may flourish as increased available prey densities improve their habitat quality. Understanding how environmental changes such as eutrophication dynamically affect habitat quality will improve the prediction of species responses and lead to better management strategies.

Table 2.1. Focal species and life-stages and their habitat characteristic preferences used in modelling growth rate potential.

Species/Life-stage Hypoxia Thermal Feeding Native sensitivity tolerance preference status YPadult Tolerant Cool Benthivore Native Yellow perch (Perca flavescens) adult YPyoy Tolerant Warm Benthivore Native Yellow perch (Perca flavescens) young-of-year RSadult Sensitive Cold Planktivore Non- Rainbow smelt (Osmerus mordax) native adult RSyoy Sensitive Warm Planktivore Non- Rainbow smelt (Osmerus mordax) native young-of-year RGadult Tolerant Warm Benthivore Non- Round goby (Neogobius native melanostomus) adult ESadult Sensitive Warm Planktivore Native Emerald shiner (Notropis atherinoides) adult

Table 2.2. Different climate year data and nutrient loading used for each GRP model scenario.

Scenario Climate Year Nutrient Loading Total GRP years

Year simulated

Hindcast 1987 2005 1987 2005 19

Nutrient Response 1987 2005 0.1 1.9 × nutrient 361 Curves loads of 1997

CHAPTER 3. BEHAVIOR AND PHSIOLOGICAL RESPONSES OF YELLOW PERCH (PERCA FLAVESCENS) TO MODERATE HYPOXIA

3.1 Introduction

Hypoxia, or the depletion of oxygen to relatively low concentrations (Vaquer-

Sunyer & Duarte, 2008) can occur naturally in aquatic systems. However, it may also develop or be exacerbated through anthropogenically-mediated nutrient loading

(Delorme, 1982; Diaz & Rosenberg, 2008; Zhang et al., 2010). Human-induced eutrophication can enhance primary production, ultimately leading to high rates of decomposition and hypoxic conditions in a variety of systems such as lakes covered by ice (Greenbank, 1945), wetlands and floodplains (Townsend & Edwards, 2003), and bottom waters of vertically stratified systems (Diaz, 2001). Due to changing climate and continued increasing anthropogenic nutrient loading, the number of hypoxic zones has increased worldwide and will likely continue to increase in the future (Diaz, 2001; Diaz

& Rosenberg, 2008; Zhang et al., 2010).

Hypoxic zones are most commonly defined by oxygen concentrations that cause lethality (often dissolved oxygen, DO, concentrations < 2.0 mg/L). However, fish and other aquatic species may be affected sub-lethally by low oxygen concentrations that exceed 2.0 mg/L (see Hrycik et al. in Review; Vaquer-Sunyer & Duarte, 2008;

Vanderplancke et al., 2014; Hughes et al., 2015). Because many organisms sense and actively avoid very low, potentially lethal oxygen concentrations (Wannamaker & Rice,

2000; Pollock et al., 2007) in many systems aquatic, organisms are likely exposed to moderate hypoxia much more often and for a longer duration than severe low oxygen concentrations. However, experiments that demonstrate an effect of hypoxia are often performed at extreme hypoxia (~ 2.0 mg/L or below) (e.g., Gracey et al. 2001, Arend et al. 2011, Rimoldi et al. 2012, Roberts et al. 2012, Froehlich et al. 2015). Analyzing the responses of fish to moderate hypoxia may provide a more holistic understanding of the behaviors and physiological responses of fish to potentially common, but less severe oxygen limitations.

In general, the first response of fishes and most mobile organisms to hypoxia is avoidance (Suthers & Gee, 1986; Wannamaker & Rice, 2000; Prince & Goodyear, 2006;

Pollock et al., 2007; Chapman & Mckenzie, 2009; Pothoven et al., 2012). Many factors, however, may influence how long a fish will remain in a hypoxic zone and whether or not they re-enter oxygen depleted water. Field observations show that the abundance of fish found within hypoxic waters decreases with oxygen concentration, but there are not always clear thresholds (Roberts et al., 2012; Hughes et al., 2015). Additionally, in some instances, fish may reside outside of the hypoxic zone, but forage into areas of low oxygen for brief periods to consume preferred prey items. Central mudminnows (Umbra limi) and juvenile anchovies (Anchoa spp.), for example, have been shown to forage in hypoxic waters at DO concentrations below 0.4 mg/L and below 1.0 mg/L, respectively

(Rahel & Nutzman, 1994; Taylor et al., 2007), however, more juvenile anchovies foraged in the hypolimnion when hypoxia was less than 2.0 mg/L than when hypoxia was more severe, below 1.0 mg/L (Taylor et al., 2007). Yellow perch (Perca flavescens) will also dive into a hypoxic hypolimnion to forage on chironomids (Hasler, 1945; Roberts et al.,

¡¢£ ¤ ¥¦£§¨ 2009). Hydroacoustic surveys traced yellow perch diving into hypolimnion in DO concentrations from 4.0 mg/L to 1.0 mg/L, but more fish were found diving into the hypolimnion at sites with higher DO concentrations (Roberts et al., 2012). These results could indicate a graded response of fish in periodic use of hypoxic areas; i.e., fish may be more willing to dive into areas with moderate hypoxia, or may spend less time in more severe hypoxia when they do dive.

To survive and minimize negative effects when exposed to hypoxia for brief or prolonged periods, a variety of physiological responses within the fish are activated,

including adjustments to blood hemoglobin concentrations, metabolism, and potentially

¦¨ £¨§ decreased food consumption. Field observations have documented an increase in f © hemoglobin concentration in hypoxic environments (Saint-Paul, 1984), but in such natural conditions the precise duration and concentration of DO exposure are unknown.

Experimentally, where DO can be controlled, hemoglobin responses are more evident under severe hypoxia (< 2.0 mg/L) (up to six hours or until equilibrium loss, Robb &

Abrahams, 2003; after 15 days, Sun et al., 2012), and do not always occur at moderate

hypoxia (2.5 3.5 mg/L) (after six hours, Robb & Abrahams, 2003; after 30 days, Tran-

Duy et al., 2008). Food consumption has also been shown to decrease in response to severe hypoxia (Roberts et al., 2011; Sun et al., 2012), and studies that examine a range of DO concentrations show that the food consumption response becomes more severe as

DO decreases (Chabot & Dutil, 1999; Tran-Duy et al., 2012). These physiological and behavioral responses can prevent immediate death when exposed to hypoxia briefly by increasing the amount of oxygen extracted from the environment or by decreasing the use

38 of oxygen within the organism (i.e., by decreasing foraging activity and metabolism), and may allow for eventual acclimation to hypoxia (Powers, 1980; Wells, 2009).

Physiological responses to hypoxia such as these are mediated by changes in the

¥¦ § ¨©¤ expression of a multitude of proteins. Hypoxia- ¡¢£¤ - is one of the main proteins associated with physiological responses to hypoxia as the dimer it forms, hypoxia-inducible factor 1, controls many known hypoxia-regulated genes and is linked to energy-saving responses during hypoxia (Semenza, 1999; Poon et al., 2002). Many studies have examined its potential for use as a biomarker for exposure to acute or chronic hypoxia (e.g., sea bass (Dichentrarchus labrax), Terova et al., 2008; Eurasian perch (Perca fluviatilis), Rimoldi et al., 2012; Pacific herring (Clupea pallasii), Froehlich et al., 2015), but it may only be detectible at severe hypoxic levels (see Froehlich et al.

2015). There is some evidence that other proteins involved in growth and stress

§© ¤ § ¡ responses, such as insulin-like growth factor- -

(HSP70), and suppressor of cytokine signaling 3 (SOCS3), may be affiliated with the

hypoxic response and may be activated by HIF- (Gracey et al., 2001; Delaney &

Klesius, 2004; Zhang et al., 2012). Other proteins involved with reproduction and growth, such as aromatase (CYP19a1a), may also be adjusted during the down-regulation of processes not necessary for survival. Again, expression of many of these proteins has only been linked to a hypoxic response when hypoxic exposure is severe (e.g., 0.8 mg/L for six days, Gracey et al., 2001; drop from 4.9 to 0.3 mg/L over three days, Delaney &

Klesius, 2004; and 0.8 mg/L for five days, Zhang et al., 2012). Given that fish do respond to moderate hypoxia, a greater examination of the change in protein and gene

39 expression at less extreme oxygen concentrations could elucidate a link between molecular changes and physiological and behavioral responses in fish.

Fish responses to hypoxia are species-specific, and potentially also specific to different populations within the same species. Different temperature tolerances have been observed in populations of the same species that are adapted to different regions or environmental conditions (Feminella & Matthews, 1984; Fields et al., 1987; Conover &

Present, 1990), and physiologically, fish have the potential to adapt hypoxic tolerance through genetically-linked changes in hemoglobin (Powers, 1980; Wells, 2009) and proteins (Rytkönen et al., 2008). Therefore, to consider the potential effect of varying tolerances between populations of a species that may encounter moderate hypoxia regularly, we conducted a series of experiments using two separate hatchery populations of yellow perch.

Yellow perch are regularly found in systems that become hypoxic and are not considered particularly sensitive to low oxygen (Petrosky & Magnuson, 1973); therefore, they may encounter moderate hypoxia more frequently. They have been shown to respond to severe hypoxia, via avoidance (Suthers & Gee, 1986; Roberts et al., 2009,

2012) and decreased consumption at DO concentrations < 2.0 mg/L (Roberts et al., 2011,

2012). Our two study populations originated in different locations (Lake Erie and North

Carolina), but had been raised in hatchery conditions for approximately 20 generations.

We assumed that the founding Lake Erie population would have occasionally encountered hypoxic conditions, potentially adapting the observed hypolimnetic diving foraging behavior (Roberts et al., 2009, 2012). The founding North Carolina population was less known; however, it was assumed that the population was likely adapted to

40 warmer temperatures that may also have become hypoxic regularly. Both populations would have encountered periods of hypoxia in hatchery conditions as well. Although the exact underlying differences between the populations were unknown, it was predicted that they would respond slightly different to hypoxic exposure with more mild responses from the population most adapted to hypoxia.

With the two populations of yellow perch, the three experiments in the present study assessed yellow perch behavioral responses to moderate hypoxia and evaluated how moderate hypoxia may alter gene expression and physiological responses. With the goal of evaluating if less severe DO concentrations induced behavioral and physiological adjustments, all experiments were conducted at more moderate DO concentrations than most previous studies. It was assumed that responses would be stronger in lower DO, and moderate in treatments with higher DO, however all of our responses would be less severe than in previous studies with severely low DO. Experiment 1 examined yellow perch foraging in moderate and low hypoxia. It was expected that movement between hypoxia and normoxia would be more common in treatments with more severe hypoxia, but overall time foraging in hypoxic waters would decrease in the lower DO (Experiment

1: Willingness to forage in hypoxic water). In experiment 2 yellow perch were exposed to moderate to severe hypoxia over a short time period, mimicking brief exposure during a foraging foray or rapid changes in DO in a shallow system. Given the moderate DO concentration, it was hypothesized that hemoglobin and gene expression would adjust slightly, but not as severely as in previous experiments with conspecifics at very low DO

(Experiment 2: Physiological response to acute hypoxia). Experiment 3 tested physiological responses of yellow perch after prolonged exposure to moderate hypoxia.

Although responses to chronic hypoxic exposure have been shown in yellow perch and conspecific Eurasian perch (Carlson et al., 1980; Roberts et al., 2011; Rimoldi et al.,

2012), a link between protein expression and physiological responses has not been established under moderate to low hypoxia. It was anticipated that consumption and growth would decrease slightly in each decreasing DO treatment, and hemoglobin concentration and gene expression would also adjust slightly in lower DO (Experiment 3:

Physiological response to chronic hypoxia).

3.2 Materials and Methods

3.2.1 Fry Collection and Rearing

During the first week of June 2014, two populations of yellow perch fry were

¡¢£ ¤ ¥¦§¨ ©¥¡ ¤ ¨¥¨ ¦ ¥¨ ¡ obtained in Ohio, which originated from Lake

Erie (Erie), and from the Purdue Aquaculture Research Lab (ARL), which originated from North Carolina (NC). Both populations were reared at the ARL in separate flow- through 3200 L tanks with temperatures initially around 26 °C to spur growth and gradually adjusted to 15 °C throughout the four months of rearing to reach target experimental temperatures. Experimental temperatures were planned to be near 15 °C to mimic the temperature of the hypolimnion of a stratified lake in late summer (Roberts et al., 2009, 2012). Temperature and DO were recorded weekly until the last month of rearing when both were recorded daily. pH was recorded weekly for the extent of rearing

42 and experimentation. It was not necessary to record ammonia and nitrates as the flow- through tanks were cleaned daily and the flow relatively quickly refreshed the tanks with well water containing almost no ammonia or nitrates. The light period followed natural light and dark periods, adjusting throughout rearing with sunset and sunrise times.

Yellow perch were initially fed to satiation twice daily with a combination of

Otohime 200 micron larval diet for marine fish and Zeigler 0.6 0.85 mm crumble, transitioning throughout the first two months to Zeigler 1.5 mm crumble. At the end of

July, fish were adjusted to once daily satiation feedings, and they were gradually transitioned to Zeigler 2.5 mm slow sink crumble in the last month of rearing. The 2.5 mm slow sink feed was used for feeding in all three experiments.

3.2.2 Experiment 1: Willingness to Forage in Hypoxic Water

A balanced 2 × 3 experimental design was used to assess the willingness of both populations of yellow perch to forage under three DO treatments. For each treatment and population combination, 20 replicates were included without reusing fish over the course of four trials. Each trial was 7 days (6 acclimation days, 1 filming day), consisted of five randomly distributed replicates of each treatment × population combination, and all four trials were run over the course of 4.5 consecutive weeks. Four fish were placed in each experimental unit which consisted of two 10 L tanks, connected with a PVC pipe 5 cm in

diameter (Fig. 3.1). One of the randomly selected tanks in the experimental unit was the

¡¢£¤¥¦£§¨© ££ ¨£ © ©£© £ ©¨¥© ¦ ! "# $%

&£ ¨£¨© ¨£ ¡&£©¨£ ¨§ ¨© ©©'¥¨£ ¨ ¨ £ ¨£©¨£ ¨ £(£)

43 moderate, control) 20 to 40 minutes prior to feeding and maintained for approximately 20 minutes while food was available. Ambient air was bubbled into both the Refuge and

Treatment tanks via diffusers until it was necessary to adjust the DO in the Treatment tank, at which point the diffuser in the Treatment tank was replaced with a diffuser that bubbled nitrogen gas (>99.99% N2) into the tank. In the low hypoxia treatment, the

Treatment tank DO was adjusted to around 2.5 mg/L (actual 2.7 mg/L ± 0.5 SD). In the moderate hypoxia treatment, the Treatment tank DO was adjusted to around 4.5 mg/L

(actual 4.5 mg/L ± 0.6 SD). The Treatment tank for the control treatment was not adjusted, but the diffuser was removed and replaced to replicate the replacement of the air diffuser with the N2 diffuser in the hypoxic treatments. The DO for the Treatment tank of the control experimental unit was maintained at 9.0 mg/L (± 0.5 SD). At the end of the 20 minutes of feeding, the N2 diffuser was replaced with the air diffuser in the

Treatment tanks.

Water flowed into the tanks by gravity flow at the rate of 100 mL · min-1 from plastic airline tubing attached to three head tanks, and temperature was 16.8 ± 1.2 (mean

± SD of all experimental units in trial), 18.2 ± 0.6, 15.2 ± 0.5, and 14.9 ± 0.6 °C for trials one through four respectively. Light was maintained at a 12h:12h light:dark regime for all trials. At least once during acclimation and hypoxic exposure, both Treatment and

Refuge tank pH was documented. Trial pH values were 8.2 ± 0.1 (mean ± SD), 8.3 ± 0.1,

8.7 ± 0.1, and 8.7 ± 0.1, respectively.

At the beginning of each trial, the Treatment and Refuge tanks of each experimental unit were assigned using a random number table. Over the 7 day trials, fish were acclimated to the experimental units for two days, starved for one day, and exposed

44 to the hypoxic treatments for the last four days. Throughout acclimation and treatment, the fish were provided with food (20 pellets of Zeigler 2.5 mm slow sink feed) once daily in the Treatment tank only. At the time of the feeding and after the DO had reached the target level in the Treatment tanks, the temperatures of the Treatment and Refuge tanks were recorded with a YSI meter. The DO in the Treatment and Refuge tanks and the number of fish in the Treatment tank 0, 10, and 20 minutes after the food was added were also recorded. On the starvation day, this protocol was still followed but no food was added. The Treatment tank and the 20 minute feeding protocol were videotaped on day 7 with a Canon Vixia HF G20 portable camcorder. Camcorders were set up to shoot for 20 minutes without interruption and people were only visible to fish during the three time periods at 0, 10, and 20 minutes when temperature and DO were recorded. All fish were removed from tanks and euthanized with MS-222 at the end of filming on day seven.

Analysis

When the videos were reviewed, two metrics were recorded: a binary value of whether or not any fish moved into or out of the Treatment tank during the 20 minutes at feeding (binomial movement) and the average time at least one fish was in the Treatment tank (total time in Treatment). Total time in Treatment was averaged for the four fish in the experimental unit to obtain a single value for each replicate. The data from replicates where it was not possible to tell if fish were moving in and out of the Treatment tank due to poor film quality were not considered in the analyses. In total, twelve replicates were removed due to this: three from Erie low DO, two from Erie moderate DO, two from Erie

45 control, one from North Carolina low DO, two from North Carolina moderate DO, and two from North Carolina control.

These data were analyzed with R (R Development Core Team, 2013) packages

£¤¥ ¦¡§ ¨© ¥ ¡¢ (Pinheiro et al., 2016) (Venables & Ripley, 2002). A generalized linear mixed model was used with trial as a random factor and treatment and population as the fixed, additive factors. Total time in Treatment was analyzed with non-parametric statistics because the data deviated significantly from normal distributions after attempted

transformation according to the Kolmogorov-Smirnov test. Specifically, total time spent

¢ ¤ £ ¦¦ ¤ ¨© ¥ in Treatment was analyzed using a Kruskal- ¦ ¢

(Venables & Ripley, 2002). All analyses for this and all following experiments were performed in R (R Development Core Team, 2013).

3.2.3 Experiment 2: Physiological Response to Acute Hypoxia

Yellow perch from both populations were briefly exposed to severe hypoxia by gradually decreasing the DO concentration in a tank from normoxic (actual 9.6 mg/L ±

0.5 SD) to approximately 2.0 mg/L (actual 2.1 mg/L ± 0.4 SD) over 30 minutes and maintaining the target concentration for 30 minutes, similar to the experiment by Rimoldi et al. (2012). For each population, there were six hypoxic exposure and six control replicate tanks (40 x 18 x 15.25 cm) with three fish in each. Water flowed into the tanks by gravity flow at the rate of 100 mL · min-1 from plastic air tubes attached to three head tanks, and temperature was maintained at 15.3 ± 0.1 °C. Air was bubbled into the head

46 tanks with a diffuser, maintaining normoxic levels in the replicate tanks while fish acclimated.

Fish were acclimated to their tanks for one day before the trial began. Two trials were conducted over two consecutive days, with three replicates of each treatment × population combinations on each day. Tanks were not reused in each trial (i.e., 2 trials, 3 replicates per trial, 4 treatments, 24 total tanks, 72 total yellow perch). At the start of the trial, a diffuser bubbling N2 gas was placed into the hypoxic treatment tanks and a diffuser bubbling ambient air was placed into the control tanks. The DO was monitored with a YSI DO probe every ten minutes, except in the first 10 minutes when it was monitored every five minutes. Immediately after the hour-long trial, fish were euthanized with MS-222. Blood samples were collected with a heparinized capillary tube after

cutting into the anterior spinal artery and immediately transferred to tubes containing

¡¢£¤¥ ¦§¨ ¡ ©¢ ©¦ ¥ ¢ -Aldrich part # D5941). Tissue samples of gill, brain, liver, and muscle were dissected from each fish. For each organ, samples of similar mass from all fish in a tank were pooled together and frozen immediately in liquid nitrogen (n = 24 final combined samples per tissue type). Blood samples, identified by individual fish, were used to determine hemoglobin concentration using the cyanmethemoglobin method

(n = 72, nested design). The other tissue samples were stored at -80 °C for up to six months until RNA extraction.

RNA Extraction and Quantitative Real-Time PCR

Liver, muscle, gill, and brain tissues were analyzed with qPCR after RNA was extracted and transformed into cDNA using the following procedures. Liver and muscle

47 samples were ground to a powder in liquid nitrogen using a mortar and pestle. Brain and gill samples, as well as powdered liver and muscle samples, were immersed in 1 ml

QIAzol® Lysis Reagent (QIAGEN Sciences product #79306) and placed on ice, sonicated for one minute and placed on ice again, with a repetition of the sonication if necessary to achieve homogeneity. Samples were then frozen in QIAzol reagent at -20 °C for 12-48 hours. RNA was extracted from the homogenized samples using a protocol recommended by the manufacturer. RNA concentration and purity was determined by

A260/A280 with a NanoDrop 2000c spectrophotometer (Thermo Scientific). Only samples with a A260/A280 ratio of 1.8 to 2.0 were carried into the next step; otherwise, they were cleaned via ethanol precipitation. Except in the cases where insufficient

amounts of RNA was extracted, 2.0 µg of RNA was treated with DNase (Thermo

¡¢£¤¥ ¢¦¢ ¡ §¨©¥ ¨ ¡¡©¢¤ ¥ ¥£ ¨¤ ¦¨¡¥©£© ¢¤¥©¡¥¢¤ ¨¤ ¥£¤

was reverse-transcribed into cDNA (Applied Biosystems part #4368814) according to the

¨¤ ¦¨¡¥©£© ¢¤¥©¡¥¢¤ ¢¤¨¦ inal volume of 20 µl. When the samples had lower total amounts of RNA, 0.05 µg to 0.55 µg of RNA were used to create cDNA with a final volume of 20 µl. The thermocycler protocol for creating cDNA was as follows: 25 °C for

10 minutes, 37 °C for 120 minutes, and then 85 °C for 5 minutes. Samples of cDNA were stored in at -20 °C.

If available, published sequences of forward and reverse primers for the focal

proteins were used, otherwise, those of conspecifics were used and verified. Primers

targeting -actin (used as reference), igf- cyp19a1a, hsp70, and socs3 were obtained

from the literature for yellow perch, and primers targeting hif- was obtained for

Eurasian perch (Table 3.1). To verify the hif- primers, the product of a PCR reaction

48 was found to be a single band using agarose gel electrophoresis, so it was then cloned

into a pMD-20 T-vector (Clontech Laboratories Inc. product # 3270) according to the

¡¢£¤¡¥¦£§¨§© ¢ ¦§£¥¦ ¢ ¡¢ ¨£¨¢¥ ¨ ¡¦¦¨£§ £ ¨ ¨¢ ¥ §¨ ¡¥ ¦ ¢¡¢

ABI PRISM 3700 genetic analyzer using dye terminator chemistry. The DNA sequence that was amplified was submitted to GenBank with accession # KX050166, and was

found to be identical to the Eurasian perch hif- sequence (GenBank accession

#EF100706).

To determine if there was an effect of treatment on the expression of the reference

gene ( -actin), a two-way ANOVA with the factors treatment and population was performed on the Cq values from each tissue after normality was verified with a

Kolmogorov-Smirnov test. There was a significant effect of population in the expression

of -actin in brain for the acute experiment (F1,21 = 4.3, p = 0.05); therefore, we interpreted any population results from the acute experiment brain tissue with caution and

determined that -actin was a reasonable reference gene in all other cases. Attempts to use the gene rplp0 as a second reference gene were hampered by the appearance of multiple PCR products in gill, brain, and muscle tissues, but not liver tissues; therefore, it was abandoned.

qPCR was performed with Bio-Rad iQTM SYBR® Green Supermix using 1.6 ng of cDNA (or 0.32 ng of cDNA if the sample had a lower concentration) following manufacturer instructions for a total reaction mix volume of 20 µl. Thermocycling and fluorescence measurement was performed on a CFX ConnectTM Real-Time System with the following protocol: polymerase activation and DNA denaturing at 95 °C occurred for

3 minutes and amplification consisted of 40 cycles of 10 seconds of denaturation at 95 °C

followed by 60 seconds of annealing/extension and fluorescence measurement at 53 °C.

The melt curve analysis occurred between 55 and 95 °C in 0.5 °C increments with 5

seconds per step. Only a single melting peak was observed for each qPCR target. Primer

efficiency was determined using the method of Pfaffl (2001). Efficiencies were between

1.73 and 2.10, and the linear range of the assay coincided with cDNA dilutions of 1:10 to

1:6250.

Analysis

The hemoglobin concentration data were evaluated for deviation from normal

distributions with a Kolmogorov-Smirnov test and for homogeneity of variance with

¡¢¡£¡¤¥ ¦ ¡¥¦ §¥¨£© ©¡ ¥¦ ¦¤ (Hui et al., 2008). Hemoglobin concentrations

were analyzed with a nested design in a general linear mixed model with R packages

£¡¤ £ ¤ ¨ £ ¦ £ ¡¡ £ ¦¥ £

(Pinheiro et al., 2016)

¡ ¦¡£¦ £ § ¦¨£ ¡¡ ¨ ¡ ¦¥ ¦

The Cq values obtained from the RT-qPCR were converted to fold change

! "

¦ ¡ ¡¦ accor¨ £ © ¦ Pfaffl (2001). Values were then log transformed before

statistical analysis. Due to skewness of data, unique distributions of medians were

created for the log fold change data of each treatment × population group using a

bootstrapping resampling technique with replacement. From these distributions, the 95%

and 80% confidence intervals were determined. Difference between groups was

evaluated by examining the overlap of these confidence intervals. Strong differences

between treatments were identified by no overlap of the 95% confidence intervals, and

50 more moderate differences were identified by the overlap of the 80% confidence

intervals.

¡¢£¡ ¤¥ £ ¡ ¦¦¡§¨¥© ¡¥ ¡ ¡ ¥§ § ¨¡¢ ¦ ¦ ¢ ¡ ¥ ¦ £¡¡¥ ¢

¡ ¡¢ £¡¤¥ £ ©¢©§ ¨©£ ¥¨ £¡ tissues to evaluate the strength of effects, with a lar £ effect. The effect size is a comparison of the mean difference between the treatment and control responses, scaled by standard deviation; in this case, the standard deviation of

fold values for each gene and tissue in each treatment × population combination.

¡¢£¡¤¥ £ ¨ ¡¥ ¨¡ ¡ ¤¥ ¢ ¡¦¦¡§¨¥© ¡ ¢ ¢¥¨¥¦ ¥ ¥ ¡ ¥© ¡ (Cohen,

1977; Hedges & Olkin, 1985). This was used to rank the effect sizes observed among various gene expressions and tissue types.

3.2.4 Experiment 3: Physiological Response to Chronic Hypoxia

For ten days, yellow perch from each population were exposed to one of three DO treatments: low (actual 2.5 mg/L ± 0.8 SD), moderate (actual 3.7 mg/L ± 0.5 SD), and control (actual 9.0 mg/L ± 0.5 SD). There were four replicate tanks per treatment × population combination (i.e., 24 tanks total) with three fish in each tank. The aquarium system described above under Experiment 2 was reused, temperature was maintained at

14.7 ± 0.4 °C, and pH was maintained between 8.2 and 8.7. Each head tank flowed to replicate tanks of different treatments, and ambient air and/or N2 gas was bubbled into the head tanks with a diffuser. The fish were fed in excess during a five day acclimation period prior to beginning the experiment. Dissolved oxygen was maintained at normoxic

levels during acclimation (actual 9.0 mg/L ± 0.2 SD). On day six, the fish were starved

and tank DO levels were gradually dropped throughout the day to treatment levels.

Feeding resumed and treatment DO levels were maintained for the subsequent ten days.

Regular measurements were taken throughout the experiment. All fish total

lengths and weights were measured at three times: before the experiment began, at the

end of the acclimation period, and at the end of the experiment to obtain tank averages of

size throughout the experiment. Dissolved oxygen, temperature, and the number of food

pellets consumed while food was available for six to eight hours were documented daily.

Immediately after the experiment, fish were euthanized with MS-222 and blood and

tissue samples from gill, brain, liver, and muscle were collected and processed as

described above for the acute exposure experiment. Gill, brain, liver, and muscle

samples of similar mass from all fish in a tank were again pooled together.

Analysis

Tank average consumption (g · g-1 · d-1), tank average growth in length (mm), and

individual fish hemoglobin concentration were evaluated for deviation from normal

distributions with a Kolmogorov-Smirnov test and for homogeneity of variance with

¡¢¡£¡¤¥ ¦ ¡¥¦§ ¨©£¥ ¦ ©£¦¡¢ ¦¡¥ £ £¦ © £© ¥¦ ¦ ©£¡¢¡£

after attempted transformations, therefore, consumption was analyzed with a Kruskal-

¥¦¡¥¦¥© ¡ ¤§ © ¦ ¥£ !¡ th a two-way ANOVA,

and hemoglobin concentration was analyzed with a nested design in a general linear

" ¡ ©¡ ¦ ¡¥£ ¡¤£ ¤§ #©¦ ¡ ¡ ©© £ £¥ ¥$ ¦£

was a random factor and treatment and population were fixed factors.

After Cq values were converted to log(fold) values, unique distributions of medians were created for the log fold change data of each treatment × population group using a bootstrapping resampling technique (see above). The same method as in Experiment 2

was used to evaluate differences between groups using the 95% and 80% confidence

¡¢ £¤¥¦¥§¨© £ £ ¨ ¥¡ ¥¨¥§¨ ¥§§¥¢£ ¥¨ ¡¢£¥¢£ £ ¨ ¤£ £ £¤ £¡¢ .

3.3 Results

3.3.1 Experiment 1: Willingness to Forage in Hypoxic Water

The number of tanks in which at least one fish moved between the Treatment and

Refuge tanks was significantly higher in the North Carolina population (low = 9, moderate = 5, control = 8) than the Lake Erie population (low = 4, moderate = 3, control

= 2) (PopulationNC: t1,101 = 2.76, p = 0.007); however, movement was unrelated to hypoxic treatment. No effect of DO treatment or population on the time spent in the

2 2

! "

£¥¢£¡¢ ¢ ¥¡ ¥¨ £¢£¢£ ¤£¥¢£¡¢ © © § ¥¢ ¡ ¤ 2,104 1,104 =

1.88, p = 0.17). Summarized treatment-specific data are presented in Table B1.

3.3.2 Experiment 2: Acute Exposure

Dissolved oxygen levels were maintained within the expected bounds over the

course of the 60 minute experiment. Initial DO levels in the hypoxic exposure and

control tanks averaged 7.9 mg/L (± 0.1 SE) and 8.1 mg/L (± 0.1 SE) respectively. After

30 minutes, DO averaged 2.3 mg/L (± 0.2 SE) and 9.7 mg/L (± 0.2 SE) in the exposure

and control tanks, respectively. After 60 minutes, DO was stable around 2.0 mg/L (± 0.1

SE) and 9.7 mg/L (± 0.1 SE).

A two-sample t-test demonstrated that a significant difference existed in total

body length between the two populations in this experiment (Fig. B1a; NC: 89.0 mm ±

4.5 SD, Erie: 101.0 mm ± 6.0 SD; t2,22 = 5.57, p < 0.001). Therefore, it was not possible

to separate effects of size from effects of population. Hemoglobin concentration was

significantly higher in the North Carolina population (NC: 49.8 mg/ml ± 12.2 SD, Erie:

44.4 mg/ml ± 9.5 SD; t1,20 = 2.98, p = 0.007), however, treatment was not significantly

different (Fig. 3.2; t1,20 = -1.84, p = 0.080). No strong differences (95% CI) were

identified in gene expression data (Fig. 3.3). Moderate differences (80% CIS) were

observed in the Erie population gill expression of hif- ¡ and brain expression of socs3

(Fig. 3.4). Both genes were higher in exposure compared to control treatments.

Additional trends were apparent in the expression of igf- ¢ in gill tissue. An examination

¦§ ¨§© § §¤¤§ ¥ § ¤£

£¤ ¥ og transformed fold changes indicated that the

§ ¥ §¤ ¤§ ¥ § ¨§© § ¤ £ § !§ £" £¤ ¢ igf- in gills of

the Erie population. Expression of igf- ¢ in gills also had the largest effect size for the

North Carolina population at 0.87 ± 0.56 (Treatment: F1,21 = 2.82, p = 1.0; Population:

¢ © §¢£ ¢ F1,21 ¡¢£¤¥ ¦ ¡¢ §¨

3.3.3 Experiment 3: Chronic Exposure

Dissolved oxygen levels were maintained within the expected bounds over the course of the 10 day experiment, except for days nine and eleven. On those two days,

DO in tanks with the most severe hypoxic treatments (~2.5 mg/L) rose to above 5.0 mg/L and the tanks with the moderate hypoxic treatments (~4.0 mg/L) rose to above 6.0 mg/L.

Dissolved oxygen was lowered again to treatment levels as soon as possible, and the longest deviation from target DO levels was less than 24 hours.

A two-sample t-test demonstrated that total body length differed between the two populations in this experiment as well (Fig. B1b; NC: 90.0 mm ± 8.0 SD, Erie: 99.0 mm

± 8.0 SD; t2,22 = 2.78, p =0.01), however, consumption, growth, and hemoglobin concentration did not differ between populations or treatments (Table B3). Gene

expression of igf- was higher in NC Control compared to Moderate treatments in brain tissue, and hsp-70 in brain tissue was higher in Erie Control than Moderate (based on lack of overlap of 95% CI; Fig. 3.5a). When considering an 80% CI, gene expression of igf-

was higher in Control compared to Moderate oxygen treatments in brain tissue of Erie

perch, igf- was lower in Control compared to Low oxygen treatments in gill tissue of

NC perch, and hif- was lower in Control compared to Moderate oxygen treatments in

liver tissue of Erie perch (Fig. 3.6). There were various differences in gene expression

! " # $%&'( ) * "

¦ ¦ § ¢ ¤¢ ¨¢ ©

g effect sizes for all log transformed fold changes indicated that the largest effect size

¨© ¨ ¤ ¨ ¨¨ was ¡¢£ ¤ ¡¥¢ ¦§ igf- in gills between the control and low treatments for the North Carolina population (Treatment: F2,20 = 0.84, p = 1.0;

Population: F1,20 = 0.66, p = 1.0). The highest effect size for the Erie population was 0.25

± 0.62 in the hepatic expression of hif- between control and moderate treatments

2 2

¨¨ ! ¢¡""# ! $ ¡ %&'( ¦ 2,20 1,21 = 1.73, p = 1.0).

3.4 Discussion

Individual behavioral and physiological responses to low DO concentrations are the bases for potential population-level and system effects of hypoxia. However, most responses have only been examined at extremely low DO concentrations, which may be only partially informative as fish can detect and avoid severe hypoxia (Wannamaker &

Rice, 2000; Pollock et al., 2007). Yellow perch foraging behavior and physiological responses to acute and chronic hypoxia in the present study indicate that moderate hypoxia may minimally affect this species. In all three experiments, yellow perch responded mildly or not at all to the hypoxic concentrations tested. These results indicate that yellow perch may be able to utilize a wide range of habitat during the development of hypoxia or greater habitat in close proximity to severe hypoxic zones.

Yellow perch will forage in hypoxic habitats, and, in conjunction with previous research, the results of this study indicate that they may do so without strong negative consequences. Fluctuating DO, as fish would experience in hypoxic foraging, has been

56 found to not affect yellow perch consumption and growth, particularly with moderate changes in DO. Fluctuations of 1-1.5 mg/L over the course of a day did not affect yellow perch growth in the study by Carlson et al. (1980), which may be similar to fluctuations yellow perch may experience if diving into a hypoxic zone around 4.0 mg/L (Roberts et al., 2012). Large changes in DO from 8 mg/L to 2 mg/L also did not affect yellow perch consumption compared to constant high DO when fluctuations occurred over the course of a day (12 hours) or rapidly (30 minutes; Roberts et al. 2012), indicating that even ventures into severe hypoxia may not greatly affect yellow perch.

In our study, behavioral results demonstrated that DO did not influence the willingness of yellow perch to move in or out of hypoxia, nor did it influence the amount of time perch spent foraging. Physiological results demonstrated no change in consumption or growth over prolonged exposure to hypoxia, and no changes in hemoglobin concentration were induced by hypoxia over acute or chronic exposures.

However, moderate differences and trends were observed in gene expression of proteins, potentially indicating that the yellow perch response is tempered at higher than lethal DO concentrations. Collectively, these results demonstrate that the dynamic way in which yellow perch respond to moderate hypoxia may require little adjustment of physiological mechanisms of protection, or that yellow perch can satisfy their oxygen needs at these concentrations and for these durations through other responses, such as hyperventilation or use of glycogen stores (Tran-Duy et al., 2008; Huang et al., 2015).

Results from our study conflict with those from previous research, but this may be due to differences in the temperatures at which studies were conducted. In previous studies yellow perch consumption does negatively respond to chronic hypoxia at constant

57 moderate (3.5 mg/L) and lower (2.0 mg/L) concentrations (Carlson et al., 1980; Roberts et al., 2011, 2012). The experimental temperatures of these studies were much warmer

(20 and 26 °C) compared with the present study (~15 °C). Experiments performed at cooler temperatures (11 °C), similar to temperatures yellow perch would experience in the hypolimnion of a lake such as Lake Erie, demonstrate results consistent with the current study (Roberts et al., 2011). Higher temperatures result in greater metabolic rates, higher respiration rates, and reduce the solubility of oxygen in water; therefore, consumption rates are higher at warmer temperatures and fishes have lower tolerances for hypoxia (Guderley, 2004; Sørensen et al., 2014; He et al., 2015). The lack of strong behavioral or physiological responses in the present three experiments may be more indicative of potential responses in vertically stratified hypoxic systems, as the temperatures used align with observed temperatures in hypolimnetic hypoxic zones.

In addition to temperature, past studies demonstrate that the duration of hypoxia exposure and severity of hypoxia can greatly affect responses. Gene expression for hif-

¡ is species-specific and quite sensitive to duration and severity of exposure. In Pacific

herring (Clupea pallasii) liver tissue, hif- ¡ was most expressed after two to eight hours

of exposure to DO of 2.3 mg/L (Froehlich et al., 2015); after which, hif- ¡ expression appeared to decline. Similarly, sea bass (Dicentrarchus labrax) exposed to DO of 1.9

mg/L for four hours significantly increased liver hif- ¡ expression (Terova et al., 2008).

However, sea bass liver hif- ¡ expression also continuously increased after exposure to moderate hypoxia (4.3 mg/L) after 48 hours, 5 days, and 15 days (Terova et al., 2008).

Studies on Eurasian perch have demonstrated a significant increase in hif- ¡ in brain and liver after DO exposure of 0.4 mg/L for approximately 40 minutes, but a decrease in

¡ muscle hif- ¡ expression and no change in brain or liver hif- expression after DO exposure of 2.8 mg/L for 15 days (Rimoldi et al., 2012). The durations and DO levels tested in the current study were estimations of environmental conditions yellow perch are most likely to encounter. Therefore, it may be reasonable to conclude that yellow perch are fairly tolerant in their response to moderate hypoxia.

It was not possible to unequivocally evaluate the effect of hypoxia on the different populations of yellow perch given the small (~12 mm) difference in average length between populations in the acute and chronic exposure experiments. The effect of hypoxia on different sized fish has been debated, but there is evidence that smaller fish are more tolerant of hypoxia (Robb & Abrahams, 2003; Tran-Duy et al., 2012).

However, in the three experiments presented herein, the differences between populations were observed at control levels as well as in the hypoxic treatments indicating that the populations may have had different baseline physiology and behavioral mechanisms. In

Experiment 1, in which there was not a difference in the sizes of individuals between populations, yellow perch from the North Carolina population did demonstrate a greater willingness to move regardless of hypoxic treatment. Experiment 2 control hemoglobin levels were also different between populations, and Experiment 3 results demonstrated potential moderate differences in control gene expression levels. Additionally, the differences in size of the two populations in Experiment 2 and 3 were relatively minor compared to the size differences used in experiments examining the effects of hypoxia on larger and smaller fish (Robb & Abrahams, 2003; Tran-Duy et al., 2012). Therefore, these experiments may demonstrate that inherent variation between populations with relatively similar histories and rearing experiences may influence tolerance, but likely at

59 lower hypoxic concentrations than tested in this study. Future work on this subject should examine responses to moderate and severe hypoxia on populations with greater differences in the habitat to which they are adapted in order to evaluate if population adaptation can alter response to hypoxia.

Due to individual-level tolerances and response mechanisms, hypoxia is known to alter species distributions and affect interactions on an ecosystem-level (Howell &

Simpson, 1994; Breitburg, 2002; Pollock et al., 2007; Ludsin et al., 2009; Pothoven et al., 2012). Moderate hypoxia is more pervasive than severe hypoxia in aquatic systems, therefore, yellow perch may be able to utilize a greater amount of habitat given their mild response to moderate hypoxia. Juvenile yellow perch may be able to use moderate hypoxia as refuges from more sensitive predators, and have access to prey attempting to utilize moderate hypoxia as a refuge from predators (Ludsin et al., 2009; Vanderploeg et al., 2009b). However, fishing pressure may be greater on yellow perch populations if they selectively utilize areas of moderate or fluctuating hypoxia proximate to severe hypoxia (Kraus et al., 2015). And, other aspects of eutrophication, such as an increase in inedible prey (i.e., cyanobacteria; see Paerl and Fulton 2006), are likely to negatively impact yellow perch.

Hypoxia can affect fish behavior (Wannamaker & Rice, 2000; Chapman &

Mckenzie, 2009; Roberts et al., 2012) and can initiate a physiological response (Chabot

& Dutil, 1999; Rimoldi et al., 2012; Tran-Duy et al., 2012), but the DO concentrations at which these responses occur depend on a variety of factors such as species-specific physiology (He et al., 2015), acclimation (Carlson et al., 1980), and temperature (Roberts et al., 2011; He et al., 2015). Yellow perch appear to be largely unaffected by moderate

60 levels of hypoxia which they are more likely to encounter regularly in hypoxic systems.

However, indirect effects of hypoxia on yellow perch, such as redistributed competitors and altered predator-prey relationships, may greatly impact yellow perch in natural environments. Results from this study should assist in future management of hypoxic zones and increase knowledge of the variety of fish responses to hypoxia. Understanding how individual species, such as yellow perch, respond to environmental changes is valuable to better protect and manage aquatic habitats as anthropogenic impacts, such as eutrophication and climate change, continue to alter ecosystems.

Table 3.1. Sequences of specific primer pairs used in quantitative PCR analyses.

Gene Accession # Primer Sequence 1

¢ -actin AY332493 F GCCTCTCTGTCCACCTTCCA R GGGCCGGACTCATCGTACT

hif- ¡ KX050166 F ATGGACACAGGAATTGTACC R TGCATCCCTTGACTTCTCCTT 1 igf- ¢ AY332492 F CGCAGGGCACAAAGTGGAC R CCCAGTGTTGCCTCGACTTG cyp19a1a2 DQ984126 F TCTGGGTTTGGGGCCACTTC R ACCGCTGATGCTCTGCTGAG hsp70 KX050165 F TGTTGGTCGGTGGCTCAA R TTGAAGAAGTCCTGAAGCAGCTT socs33 HQ696607 F ACGAGCGAGAAGGTGTTTCAG R CTCATTGCGGAGTCAAACTTGC 1Pierron et al. (2009) 2Lynn et al. (2009) 3Shepherd et al. (2012)

Figure 3.2. Hemoglobin concentrations in Experiment 2 (acute) for the control and

¢£¤ ¥ ¦ §¤¢£¨© § §¤¢£¨© § ¡ treatments of the Erie and NC populations. Center line in box plots shows median, and upper and lower edges of the box are the 25th and 75th percentiles, and error bars are the 10th and 90th percentiles.

REFERENCES

Andersen T, Hessen DO (1991) Carbon, nitrogen, and phosphorus content of freshwater

zooplankton. Limnology and Oceanography, 36, 807 814.

Arend KK, Beletsky D, DePinto J V. et al. (2011) Seasonal and interannual effects of

hypoxia on fish habitat quality in central Lake Erie. Freshwater Biology, 56, 366

383.

Avramidis P, Bekiari V, Christodoulou D, Papatheodorou G (2015) Sedimentology and

water column stratification in a permanent anoxic Mediterranean lagoon

environment, Aetoliko Lagoon, western Greece. Environmental Earth Sciences, 73,

5687 5701.

Baek MJ, Yoon TJ, Bae YJ (2012) Development of Glyptotendipes tokunagai (Diptera:

Chironomidae) under different temperature conditions. Environmental Entomology,

41, 950 958.

Baldwin CM, Beauchamp DA, Gubala CP (2002) Seasonal and diel distribution and

movement of cutthroat trout from ultrasonic telemetry. Transactions of the

American Fisheries Society, 131, 143 158. van den Berghe EP, Gross MR (1989) Natural selection resulting from female breeding

competition in a Pacific salmon (Coho: Oncorhynchus kisutch). Evolution, 43, 125

140.

Blaxter JHS (1986) Development of sense organs and behavior of teleost larvae with

special reference to feeding and predator avoidance. Transactions of the American

Fisheries Society, 115, 98 114.

Blumberg AF, Mellor GL (1987) A description of a three-dimensional coastal ocean

circulation model. In: Three-dimensional coastal ocean models, Vol. 4 (ed Heaps

NS), pp. 1 16. American Geophysical Union.

Blumenshine SC, Vadeboncoeur Y, Lodge DM, Cottingham KL, Knight SE (1997)

Benthic-pelagic links: responses of benthos to water-column nutrient enrichment.

Journal of the North American Benthological Society, 16, 466 479.

Brandt SB, Mason DM, Patrick EV (1992) Spatially-explicit models of fish growth rate.

Fisheries, 17, 23 35.

Brandt SB, Costantini M, Kolesar S, Ludsin SA, Mason DM, Rae CM, Zhang H (2011)

Does hypoxia reduce habitat quality for Lake Erie walleye (Sander vitreus)? A

bioenergetics perspective. Canadian Journal of Fisheries and Aquatic Sciences, 68,

857 879.

Breck JE (1993) Hurry up and wait: Growth of young bluegills in ponds and in

simulations with an individual-based model. Transactions of the American Fisheries

Society, 122, 467 480.

Breck JE, Gitter MJ (1983) Effect of fish size on the reactive distance of bluegill

(Lepomis macrochirus) sunfish. Canadian Journal of Fisheries and Aquatic

Sciences, 40, 162 167.

Breitburg D (2002) Effects of hypoxia, and the balance between hypoxia and enrichment,

on coastal fishes and fisheries. Estuaries, 25, 767 781.

Breitburg DL, Hondorp DW, Davias LA, Diaz RJ (2009) Hypoxia, nitrogen, and

fisheries: Integrating effects across local and global landscapes. Annual Review of

Marine Science, 1, 329 349.

Budy P, Baker M, Dahle SK (2011) Predicting fish growth potential and identifying

water quality constraints: A spatially-explicit bioenergetics approach. Environmental

Management, 48, 691 709.

Burns NM, Rockwell DC, Bertram PE, Dolan DM, Ciborowski JJH (2005) Trends in

temperature, secchi depth, and dissolved oxygen depletion rates in the central basin

of Lake Erie, 1983 2002. Journal of Great Lakes Research, 31, 35 49.

Caddy JF (1993) Toward a comparative evaluation of human impacts on fishery

ecosystems of enclosed and semi-enclosed seas. Reviews in Fisheries Science, 1,

57 95.

Caddy JF (2000) Marine catchment basin effects versus impacts of fisheries on semi-

enclosed seas. ICES Journal of Marine Science, 57, 628 640.

Carlson AR, Blocher J, Herman LJ (1980) Growth and survival of channel catfish and

yellow perch exposed to lowered constant and diurnally fluctuating dissolved

oxygen concentrations. The Progressive Fish-Culturist, 42, 73 78.

Chabot D, Dutil J-D (1999) Reduced growth of Atlantic cod in non-lethal hypoxic

conditions. Journal of Fish Biology, 55, 472 491.

Chapman LJ, Mckenzie DJ (2009) Behavioral responses and ecological consequences,

Vol. 27. Elsevier, 25-77 pp.

Cohen J (1977) Statistical power analysis for the behavioral sciences (rev. ed.).

Conover DO, Present TMC (1990) Countergradient variation in growth rate:

Compensation for length of the growing season among Atlantic silversides from

different latitudes. Oecologia, 83, 316 324.

Delaney MA, Klesius PH (2004) Hypoxic conditions induce Hsp70 production in blood,

brain and head kidney of juvenile Nile tilapia Oreochromis niloticus (L.).

Aquaculture, 236, 633 644.

Delorme LD (1982) Lake Erie oxygen; The prehistoric record. Canadian Journal of

Fisheries and Aquatic Sciences, 39, 1021 1029.

Deutsch C, Brix H, Ito T, Frenzel H, Thompson L (2011) Climate-forced variability of

ocean hypoxia. Science (New York, N.Y.), 333, 336 339.

Diaz RJ (2001) Overview of hypoxia around the world. Journal of environmental quality,

30, 275 281.

Diaz RJ, Rosenberg R (2008) Spreading dead zones and consequences for marine

ecosystems. Science (New York, N.Y.), 321, 926 929.

Duffy WG (1998) Population dynamics, production, and prey consumption of fathead

minnows (Pimephales promelas) in prairie wetlands: a bioenergetics approach.

Canadian Journal of Fisheries and Aquatic Sciences, 55, 15 27.

Durant JM, Hjermann D, Ottersen G, Stenseth NC (2007) Climate and the match or

mismatch between predator requirements and resource availability. Climate

Research, 33, 271 283.

Eby LA, Crowder LB, McClellan CM, Peterson CH, Powers MJ (2005) Habitat

degradation from intermittent hypoxia: Impacts on demersal fishes. Marine Ecology

Progress Series, 291, 249 261.

¡¢£¤¥¦ § ¨© ¦ ¢¦ ¥ ¥¥¢¥ ¡ ¥ ¦¥¤£¡

and at the turn of the century. Ambio, 18, 326 332.

Feminella JW, Matthews WJ (1984) Intraspecific differences in thermal tolerance of

Etheostoma spectabile (Agassiz) in constant versus fluctuating environments.

Journal of Fish Biology, 25, 455 461.

Fields R, Lowe SS, Kaminski C, Whitt GS, Philipp DP (1987) Critical and chronic

thermal maxima of northern and Florida largemouth bass and their reciprocal F1 and

F2 hybrids. Transactions of the American Fisheries Society, 116, 856 863.

Forsman A, Lindell LE (1991) Trade-off between growth and energy storage in male

Vipera berus (L.) under different prey densities. Functional Ecology, 5, 717 723. © Froehlich HE, Roberts SB, Essington TE (2015) Evaluating hypoxia-inducible factor-

mRNA expression in a pelagic fish, Pacific herring Clupea pallasii, as a biomarker

for hypoxia exposure. Comparative Biochemistry and Physiology Part A: Molecular

& Integrative Physiology, 189, 58 66.

Fulford RS, Rice JA, Miller TJ, Binkowski FP, Dettmers JM, Belonger B (2006)

Foraging selectivity by larval yellow perch (Perca flavescens): Implications for

understanding recruitment in small and large lakes. Canadian Journal of Fisheries

and Aquatic Sciences, 63, 28 42.

Goedkoop W, Demandt M, Ahlgren G (2007) Interactions between food quantity and

quality (long-chain polyunsaturated fatty acid concentrations) effects on growth and

development of Chironomus riparius. Canadian Journal of Fisheries and Aquatic

Sciences, 64, 425 436.

Gracey AY, Troll JV, Somero GN (2001) Hypoxia-induced gene expression profiling in

the euryoxic fish Gillichthys mirabilis. Proceedings of the National Academy of

Sciences of the United States of America, 98, 1993 1998.

Graeb BDS, Dettmers JM, Wahl DH, Cáceres CE (2004) Fish size and prey availability

affect growth, survival, prey selection, and foraging behavior of larval yellow perch.

Transactions of the American Fisheries Society, 133, 504 514.

Greenbank J (1945) Limnological conditions in ice-covered lakes, especially as related to

winter-kill of fish. Ecological Monographs, 15, 343 392.

Greene RM, Lehrter JC, Hagy JD (2009) Multiple regression models for hindcasting and

forecasting midsummer hypoxia in the Gulf of Mexico. Ecological Applications, 19,

1161 1175.

Guderley H (2004) Metabolic responses to low temperature in fish muscle. Biological

reviews of the Cambridge Philosophical Society, 79, 409 427.

Hanson PC, Johnson TB, Schindler DE, Kitchell JF (1997) Fish bioenergetics 3.0 for

Windows, Vol. WISCU-T-97.

Hartman KJ, Margraf FJ (1992) Effects of prey and predator abundances on prey

consumption and growth of walleyes in western Lake Erie. Transactions of the

American Fisheries Society, 121, 245 260.

Hasler AD (1945) Observations on the Winter Perch Population of Lake Mendota.

Ecology, 26, 90 94.

Hawley N, Johengen TH, Rao YR et al. (2006) Lake Erie hypoxia prompts Canada-U.S.

study. Eos, Transactions, American Geophysical Union, 87, 313 324.

He W, Cao Z-D, Fu S-J (2015) Effect of temperature on hypoxia tolerance and its

underlying biochemical mechanism in two juvenile cyprinids exhibiting distinct £¤¥ hypoxia sensitivities. Comparative Biochemistry and Physiology - ¡¢

Molecular and Integrative Physiology, 187, 232¦ 241.

Hedges LV, Olkin I (1985) Statistical methods for meta-analysis.

Hergenrader GL, Hasler AD (1967) Seasonal changes in swimming rates of yellow perch

in Lake Mendota as measured by sonar. Transactions of the American Fisheries

Society, 96, 373¦ 382.

Van Horne B (1983) Density as a misleading indicator of habitat quality. The Journal of

Wildlife Management, 47, 893¦ 901.

Howell P, Simpson D (1994) Abundance of marine resources in relation to dissolved

oxygen in Long Island Sound. Estuaries, 17, 394.

Hrycik AR, Almeida LZ, Höök TO (in Review) Sub-lethal effects on fish provide insight

into a biologically-relevant threshold of hypoxia.

Huang C-Y, Lin H-C, Lin C-H (2015) Effects of hypoxia on ionic regulation, glycogen

utilization and antioxidative ability in the gills and liver of the aquatic air-breathing

fish Trichogaster microlepis. Comparative Biochemistry and Physiology Part A:

Molecular & Integrative Physiology, 179, 25¦ 34.

Hughes BB, Levey MD, Fountain MC, Carlisle AB, Chavez FP, Gleason MG (2015)

Climate mediates hypoxic stress on fish diversity and nursery function at the land ¦

sea interface. Proceedings of the National Academy of Sciences, 112, 8025¦ 8030.

Hui W, Gel YR, Gastwirth JL (2008) lawstat: An R package for law, public policy and

biostatistics. Journal of Statistical Software, 28.

Humston R, Olson DB, Ault JS (2004) Behavioral assumptions in models of fish

movement and their influence on population dynamics. Transactions of the

American Fisheries Society, 133, 1304 1328.

Hushak LJ, Winslow JM, Dutta N (1988a) Economic value of Great Lakes sportfishing:

¤¥¦§¨¦© ¦© ¨¦¢ § £ ¦ The case of private-¡¢£ Transactions of the American

Fisheries Society, 117, 363 373.

§¨£ ¦©§¢ ¤¤£ ¡! "¢©¢ # ¦"$£ ¢ ¥ % £¤ £ § & '¢ ¤¥¦§¨¦© (

¢£¤ *¦§¨¦© ¦© ¨¦¢ § £ ¦ The Case of Private- ) Transactions of the

American Fisheries Society, 117, 363 373.

Jenny J-P, Arnaud F, Alric B, Dorioz J-M, Sabatier P, Meybeck M, Perga M-E (2014)

Inherited hypoxia: A new challenge for reoligotrophicated lakes under global

warming. Global Biogeochemical Cycles, 28, 1413 1423.

Jenny J-P, Francus P, Normandeau A et al. (2016) Global spread of hypoxia in freshwater

ecosystems during the last three centuries is caused by rising local human pressure.

Global Change Biology, 22, 1481 1489.

Jeppesen E, Jensen JP, Søndergaard M, Lauridsen T, Landkildehus F (2000) Trophic

structure, species richness and diversity in Danish lakes: Changes along a

phosphorus gradient. Freshwater Biology, 45, 201 218.

Kemp WM, Testa JM, Conley DJ, Gilbert D, Hagy JD (2009) Temporal responses of

coastal hypoxia to nutrient loading and physical controls. Biogeosciences, 6, 2985

3008.

Kitchell JF, Stewart DJ, Weininger D (1977) Applications of a bioenergetics model to

yellow perch (Perca flavescens) and walleye (Stizostedion vitreum vitreum). Journal

of the Fisheries Research Board of Canada, 34, 1922 1935.

Kraus RT, Knight CT, Farmer TM et al. (2015) Dynamic hypoxic zones in Lake Erie

compress fish habitat, altering vulnerability to fishing gears. Canadian Journal of

Fisheries and Aquatic Sciences, 72, 797 806.

Kristiansen T, Jørgensen C, Lough RG, Vikebø F, Fiksen Ø (2009) Modeling rule-based

behavior: Habitat selection and the growth-survival trade-off in larval cod.

Behavioral Ecology, 20, 490 500.

Kuiper JJ, van Altena C, de Ruiter PC, van Gerven LPA, Janse JH, Mooij WM (2015)

Food-web stability signals critical transitions in temperate shallow lakes. Nature

Communications, 6, 7727.

Lam DCL, Schertzer WM, Fraser AS (1987) A post-audit anaylsis of the NWRI nine-box

water quality model for Lake Erie. Journal of Great Lakes Research, 13, 782 800.

Lampert W (1989) The adaptive significance of diel vertical migration of zooplankton.

Functional Ecology, 3, 21 27.

Lantry BF, Stewart DJ (1993) Ecological energetics of rainbow smelt in the Laurentian

Great Lakes: An interlake comparison. Transactions of the American Fisheries

Society, 122, 951 976.

Lee VA, Johnson TB (2005) Development of a bioenergetics model for the round goby

(Neogobius melanostomus). Journal of Great Lakes Research, 31, 125 134.

Li Y, Smayda TJ (1998) Temporal variability of chlorophyll in Narragansett Bay, 1973-

1990. ICES Journal of Marine Science, 55, 661 667.

Li M, Lee YJ, Testa JM, Li Y, Ni W, Kemp WM, Di Toro DM (2016) What drives

interannual variability of hypoxia in Chesapeake Bay: Climate forcing versus

nutrient loading? Geophysical Research Letters, n/a n/a.

Ludsin SA, Kershner MW, Blocksom KA, Knight R, Stein RA (2001) Life after death in

Lake Erie: Nutrient controls drive fish species richness, rehabilitation. Ecological

Applications, 11, 731 746.

Ludsin SA, Zhang X, Brandt SB, Roman MR, Boicourt WC, Mason DM, Costantini M

(2009) Hypoxia-avoidance by planktivorous fish in Chesapeake Bay: Implications

for food web interactions and fish recruitment. Journal of Experimental Marine

Biology and Ecology, 381, 121 131.

Lyche Solheim A, Rekolainen S, Moe SJ et al. (2008) Ecological threshold responses in

European lakes and their applicability for the Water Framework Directive (WFD)

implementation: Synthesis of lakes results from the REBECCA project. Aquatic

Ecology, 42, 317 334.

Lynn SG, Powell K a., Westneat DF, Shepherd BS (2009) Seasonal and sex-specific

mRNA levels of key endocrine genes in adult yellow perch (perca flavescens) from

lake erie. Marine Biotechnology, 11, 210 222.

Makarewicz JC, Lewis TW, Bertram P (1989) Phytoplankton and zooplankton

composition, abundance and distribution and trophic interactions: Offshore region

of Lakes Erie, Lake Huron and Lake Michigan, 1985, Vol. GLNPO Repo.

Marsden JE, Charlebois P, Wolfe K, Jude DJ, Rudnicka S (1996) The round goby

(Neogobius melanostomus): A review of European and North American literature.

Mason DM, Brandt SB (1996) Effects of spatial scale and foraging efficiency on the

predictions made by spatially-explicit models of fish growth rate potential.

Environmental Biology of Fishes, 45, 283 298.

May CJ, Aday DD, Hale RS, Denlinger JCS, Marschall EA (2012) Modeling habitat

selection of a top predator: Considering growth and physical environments in a

spatial context. Transactions of the American Fisheries Society, 141, 215 223.

¢£¤¥¦ § ¨ © ¦¥ ¡ EJ, Schneider FD, Hetjens H, Le Merrer P, Coleman RA,

Emmerson M (2015) Size-balanced community reorganization in response to

nutrients and warming. Global Change Biology, 21, 3971 3981.

Mehner T (2012) Diel vertical migration of freshwater fishes - proximate triggers,

ultimate causes and research perspectives. Freshwater Biology, 57, 1342 1359.

Michalak AM, Anderson EJ, Beletsky D et al. (2013) Record-setting algal bloom in Lake

Erie caused by agricultural and meteorological trends consistent with expected

future conditions. Proceedings of the National Academy of Sciences of the United

States of America, 110, 6448 6452.

Mittelbach GG (1981) Foraging efficiency and body size: A study of optimal diet and

habitat use by bluegills. Ecology, 62, 1370 1386.

Oetken M, Jagodzinski LS, Vogt C, Jochum A, Oehlmann J (2009) Combined effects of

chemical and temperature stress on Chironomus riparius populations with differing

genetic variability. Journal of environmental science and health. Part A,

Toxic/hazardous substances & environmental engineering, 44, 955 962.

Paerl HW (2006) Assessing and managing nutrient-enhanced eutrophication in estuarine

and coastal waters: Interactive effects of human and climatic perturbations.

Ecological Engineering, 26, 40 54.

Paerl HW, Fulton RS (2006) Ecology of harmful cyanobacteria. In: Ecology of Harmful

Algae, Vol. 189, pp. 95 109.

Petersen JK, Pihl L (1995) Responses to hypoxia of plaice, Pleuronectes platessa, and

dab, Limanda limanda, in the south-east Kattegat: distribution and growth.

Environmental Biology of Fishes, 43, 311 321.

Petrosky BR, Magnuson JJ (1973) Behavioral responses of northern pike, yellow perch

and bluegill to oxygen concentrations under simulated winterkill conditions. Copeia,

1973, 124 133.

Pfaffl MW (2001) A new mathematical model for relative quantification in real-time RT-

PCR. Nucleic Acids Research, 29, 2002 2007.

Pierron F, Bourret V, St-Cyr J, Campbell PGC, Bernatchez L, Couture P (2009)

Transcriptional responses to environmental metal exposure in wild yellow perch

(Perca flavescens) collected in lakes with differing environmental metal

concentrations (Cd, Cu, Ni). Ecotoxicology, 18, 620 631.

Pihl L (1994) Changes in the diet of demersal fish due to eutrophication-induced hypoxia

in the Kattegat, Sweden. Canadian Journal of Fisheries and Aquatic Sciences, 51,

321 336.

Pihl L, Baden SP, Diaz RJ (1991) Effects of periodic hypoxia on distribution of demersal

fish and crustaceans. Marine Biology, 108, 349 360.

Pinheiro J, Bates D, DebRoy S, Sarkar D, R Core Team (2016) nlme: Linear and

Nonlinear Mixed Effects Models.

Pollock MS, Clarke LMJ, Dubé MG (2007) The effects of hypoxia on fishes: From

ecological relevance to physiological effects. Environmental Reviews, 15, 1 14.

Poon WL, Hung CY, Randall DJ (2002) The effects of aquatic hypoxia on fish. In: Fish

Physiology, Toxicology, and Water Quality: Proceedings of the Sixth International

Symposium, La Paz, BCS Mexico, January 22-26, 2001, pp. 31 50. Ecosystems

Research Division.

Post JR, Evans DO (1989) Size-dependent overwinter mortality of young-of-the-year

yellow perch (Perca flavescens): Laboratory, in situ enclosure, and field

experiments. Canadian Journal of Fisheries and Aquatic Sciences, 46, 1958 1968.

Pothoven SA, Vanderploeg HA, Ludsin SA, Höök TO, Brandt SB (2009) Feeding

ecology of emerald shiners and rainbow smelt in central Lake Erie. Journal of Great

Lakes Research, 35, 190 198.

Pothoven SA, Vanderploeg HA, Höök TO, Ludsin SA, Sprules G (2012) Hypoxia

modifies planktivore zooplankton interactions in Lake Erie. Canadian Journal of

Fisheries and Aquatic Sciences, 69, 2018 2028.

Powers DA (1980) Molecular ecology of teleost fish hemoglobins: strategies for adapting

to changing environments. American Zoologist, 20, 139 162.

Powers SP, Peterson CH, Christian RR, Sullivan E, Powers MJ, Bishop MJ, Buzzelli CP

(2005) Effects of eutrophication on bottom habitat and prey resources of demersal

fishes. Marine Ecology Progress Series, 302, 233 243.

Prince ED, Goodyear CP (2006) Hypoxia-based habitat compression of tropical pelagic

fishes. Fisheries Oceanography, 15, 451 464.

Pulliam HR (1988) Sources, sinks, and population regulation. The American Naturalist,

132, 652 661.

R Development Core Team R (2013) R: A Language and Environment for Statistical

Computing (ed Team RDC). R Foundation for Statistical Computing, 1, 409.

Rahel FJ, Nutzman JW (1994) Foraging in a lethal environment: Fish predation in

hypoxic waters of a stratified lake. Ecology, 75, 1246 1253.

Reynolds SK, Benke AC (2005) Temperature-dependent growth rates of larval midges

(Diptera: Chironomidae) from a southeastern U.S. stream. Hydrobiologia, 544, 69

75.

Richards JG (2009) Metabolic and molecular responses of fish to hypoxia. In: Fish

Physiology: Hypoxia, Vol. 27, pp. 443 485. Elsevier.

Rimoldi S, Terova G, Ceccuzzi P, Marelli S, Antonini M, Saroglia M (2012) HIF-¡¢

mRNA levels in Eurasian perch (Perca fluviatilis) exposed to acute and chronic

hypoxia. Molecular Biology Reports, 39, 4009 4015.

Robb T, Abrahams M V. (2003) Variation in tolerance to hypoxia in a predator and prey

species: An ecological advantage of being small? Journal of Fish Biology, 62, 1067

1081.

Roberts JJ, Höök TO, Ludsin SA, Pothoven SA, Vanderploeg HA, Brandt SB (2009)

Effects of hypolimnetic hypoxia on foraging and distributions of Lake Erie yellow

perch. Journal of Experimental Marine Biology and Ecology, 381.

Roberts JJ, Brandt SB, Fanslow D, Ludsin SA, Pothoven SA, Scavia D, Höök TO (2011)

Effects of Hypoxia on Consumption, Growth, and RNA:DNA Ratios of Young

Yellow Perch. Transactions of the American Fisheries Society, 140, 1574 1586.

Roberts JJ, Grecay PA, Ludsin SA, Pothoven SA, Vanderploeg HA, Höök TO (2012)

Evidence of hypoxic foraging forays by yellow perch (Perca flavescens) and

potential consequences for prey consumption. Freshwater Biology, 57, 922 937.

Rose KA, Tyler JA, Chambers RC, Klein-MacPhee G, Danila DJ (1996) Simulating

winter flounder population dynamics using coupled individual-based young-of-the-

year and age-structured adult models. Canadian Journal of Fisheries and Aquatic

Sciences, 53, 1071 1091.

Rose KA, Rutherford ES, Mcdermot DS, Forney JL, Mills EL (1999) Individual-based

model of yellow perch and walleye populations in Oneida Lake. Ecological

Monographs, 69, 127 154.

Roy D, Haffner GD, Brandt SB (2004) Estimating fish production potentials using a

temporally explicit model. Ecological Modelling, 173, 241 257.

¤¥¦ §¤ ¨© ¨ ¤¥ £¤¨ ¦ § ¨ ¤¥ !§ "#¤ $¦ % ¡¢£ ypoxia

response to nutrient loads and physical variability. Journal of Great Lakes Research.

Rytkönen KT, Ryynänen HJ, Nikinmaa M, Primmer CR (2008) Variable patterns in the

& ¥ ¤ ¥ molecular evolution of the hypoxia-inducible factor-1 alpha (HIF- st

fishes and mammals. Gene, 420, 1 10.

Saint-Paul U (1984) Physiological adaptation to hypoxia of a neotropical characoid fish

Colossoma macropomum, Serrasalmidae. Environmental Biology of Fishes, 11, 53

62.

Scavia D, Allan JD, Arend KK et al. (2014) Assessing and addressing the re-

eutrophication of Lake Erie: Central basin hypoxia. Journal of Great Lakes

Research, 40, 226 246.

Semenza GL (1999) Regulation of mammalian oxygen homeostasis by hypoxia-inducible

factor 1. Annu. Rev. Cell Dev. Biol., 15, 551 578.

Sharpley A, Jarvie HP, Buda A, May L, Spears B, Kleinman P (2013) Phosphorus

legacy: Overcoming the effects of past management practices to mitigate future

water quality impairment. Journal of Environment Quality, 42, 1308.

Shepherd BS, Rees CB, Binkowski FP, Goetz FW (2012) Characterization and evaluation

of sex-specific expression of suppressors of cytokine signaling (SOCS)-1 and -3 in

juvenile yellow perch (Perca flavescens) treated with lipopolysaccharide. Fish and

Shellfish Immunology, 33, 468 481.

Shoji J, Masuda R, Yamashita Y, Tanaka M (2005) Effect of low dissolved oxygen

concentrations on behavior and predation rates on red sea bream Pagrus major larvae

by the jellyfish Aurelia aurita and by juvenile Spanish mackerel Scomberomorus

niphonius. Marine Biology, 147, 863 868.

Silva N, Vargas CA (2014) Hypoxia in Chilean Patagonian fjords. Progress in

Oceanography, 129, 62 74.

Smith S V, Kimmerer WJ, Laws EA, Brock RE, Walsh TW (1981) Kaneohe Bay sewage

diversion experiment: Perspectives on ecosystem responses to nutritional

perturbation. Pacific Science, 35, 279 395.

Sørensen C, Munday PL, Nilsson GE (2014) Aerobic vs. anaerobic scope: Sibling species

of fish indicate that temperature dependence of hypoxia tolerance can predict future

survival. Global Change Biology, 20, 724 729.

Stevens MM (1998) Development and survival of Chironomus tepperi Skuse (Diptera:

¢£¤¥¦ § ¨© §¥ ¨©¥ ©¤¥¦ ©¤ Chironomidae) at a range of constant temperatures. ¡

Journal of Freshwater Entomology, 20, 37 41.

Sun H, Li J, Tang L, Yang Z (2012) Responses of crucian carp Carassius auratus to long-

term exposure to nitrite and low dissolved oxygen levels. Biochemical Systematics

and Ecology, 44, 224 232.

Suthers IM, Gee JH (1986) Role of hypoxia in limiting diel spring and summer

distribution of juvenile yellow perch (Perca flavescens) in a prairie marsh. Canadian

Journal of Fisheries and Aquatic Sciences, 43, 1562 1570.

Tassou KT, Schulz R (2012) Combined effects of temperature and pyriproxyfen stress in

a full life-cycle test with Chironomus riparius (Insecta). Environmental Toxicology

and Chemistry, 31, 2384 2390.

Taylor JC, Rand PS, Jenkins J (2007) Swimming behavior of juvenile anchovies (Anchoa

spp.) in an episodically hypoxic estuary: Implications for individual energetics and

trophic dynamics. Marine Biology, 152, 939 957.

Terova G, Rimoldi S, Corà S, Bernardini G, Gornati R, Saroglia M (2008) Acute and

!"" #$ "%&' chronic hypoxia affects HIF-

Aquaculture, 279, 150 159.

Townsend SA, Edwards CA (2003) A fish kill event, hypoxia and other limnological

impacts associated with early wet season flow into a lake on the Mary River

floodplain, tropical northern Australia. Lakes and Reservoirs: Research and

Management, 8, 169 176.

Tran-Duy A, Schrama JW, van Dam AA, Verreth JAJ (2008) Effects of oxygen

concentration and body weight on maximum feed intake, growth and hematological

parameters of Nile tilapia, Oreochromis niloticus. Aquaculture, 275, 152 162.

Tran-Duy A, van Dam AA, Schrama JW (2012) Feed intake, growth and metabolism of

Nile tilapia (Oreochromis niloticus) in relation to dissolved oxygen concentration.

Aquaculture Research, 43, 730 744.

Trolle D, Skovgaard H, Jeppesen E (2008) The Water Framework Directive: Setting the

phosphorus loading target for a deep lake in Denmark using the 1D lake ecosystem

model DYRESM-CAEDYM. Ecological Modelling, 219, 138 152.

Vanderplancke G, Claireaux G, Quazuguel P, Huelvan C, Corporeau C, Mazurais D,

Zambonino-Infante JL (2014) Exposure to chronic moderate hypoxia impacts

physiological and developmental traits of European sea bass (Dicentrarchus labrax)

larvae. Fish Physiology and Biochemistry, 41, 233 242.

Vanderploeg HA, Ludsin SA, Ruberg SA et al. (2009a) Hypoxia affects spatial

distributions and overlap of pelagic fish, zooplankton, and phytoplankton in Lake

Erie. Journal of Experimental Marine Biology and Ecology, 381, S92 S107.

Vanderploeg HA, Ludsin SA, Cavaletto JF et al. (2009b) Hypoxic zones as habitat for

zooplankton in Lake Erie: Refuges from predation or exclusion zones? Journal of

Experimental Marine Biology and Ecology, 381, S108 S120.

Vaquer-Sunyer R, Duarte CM (2008) Thresholds of hypoxia for marine biodiversity.

Proceedings of the National Academy of Sciences of the United States of America,

105, 15452 15457.

Venables WN, Ripley BD (2002) Modern Applied Statistics with S, Fourth edn. Springer,

New York.

Vogt C, Pupp A, Nowak C et al. (2007) Interaction between genetic diversity and

temperature stress on life-cycle parameters and genetic variability in midge

Chironomus riparius populations. Climate Research, 33, 207 214.

Wannamaker CM, Rice JA (2000) Effects of hypoxia on movements and behavior of

selected estuarine organisms from the southeastern United States. Journal of

Experimental Marine Biology and Ecology, 249, 145 163.

Wells RMG (2009) Blood-gas transport and hemoglobin function: Adaptations for

functional and environmental hypoxia. In: Fish Physiology: Hypoxia, Vol. 27, pp.

255 299. Elsevier.

Willis DW, Guy CS, Murphy BR (1991) Development and evaluation of a standard

weight (WS) equation for yellow perch. North American Journal of Fisheries

Management, 11, 374 380.

Zhang J, Gilbert D, Gooday AJ et al. (2010) Natural and human-induced hypoxia and

consequences for coastal areas: Synthesis and future development. Biogeosciences,

7, 1443 1467.

Zhang Z, Wells MC, Boswell MG et al. (2012) Identification of robust hypoxia

biomarker candidates from fin of medaka (Oryzias latipes). Comparative

Biochemistry and Physiology - C Toxicology and Pharmacology, 155, 11 17.

Zollner PA, Lima SL (1999) Search strategies for landscape-level interpatch movements.

Ecology, 80, 1019 1030.

133

APPENDICES

Appendix A Supporting Model Methods and Results

Appendix A1: Prey abundance and distribution methods

Zooplankton

-1

££ ¤§¤ ¤ ¡¥ § ¨¥§ ¡ §¨¡¤¡ ¡¡¢£¤¥¦§ ¡¥ ) and

redistributed into small (0.2 mm, e.g., Rotifera, Cyclopoida, Copepoda nauplii), medium

(0.83 mm, e.g., Copepoda), or large (1.25 mm, e.g., Cladocera) sizes (s) at a ten minute

time step (i) for each depth (d) with the following equation:

#$ % $ & '

()*)+

,-,.

" 34567

/012

Eq. 1

-1 -1

8 89 : 8 ;<

¨ ¨¥ § ¤¥ ¨¨¥ ¤ ¡¥

Wh Daphnia (Andersen

> =

& Hessen, 1991) Ztot

= >

¨ £

Prop ¨ -based proportions of zooplankton that are small (0.423), medium (0.199),

> =

size at each depth. Except for the small zooplankton, these proportions are based on

equations that account for the light level, temperature, and dissolved oxygen at each

depth at each 10 minute time step. The equations were calibrated to redistribute

zooplankton in model year 2005 to match field observations from 2005 (Vanderploeg et

al., 2009b). For simplification, the small zooplankton proportion is evenly divided across

all depths (1/48). The medium zooplankton proportion is the average of the small and

large proportions. The large zooplankton proportion for each depth (d) and each ten

minute time step (i) is:

£¤ ¥ ©£ ¥ ©£ ¥

¦¨ ¦¨ ¦¨

¡¢

£¤¥ ©£ ©£ ¥

¦¨ ¨ ¦¨ ¦¨

Eq. 2

where

!"#$% & ' ( ))* )

¦¨

,-.-/

¡¢

!"#$% & ' ( ))* )

¦¨

012 ,-.-/

Eq. 3

(see Fig. A1a),

' "

; 8¨9 :

¦¨

= >

" '

8¨9: ;

¦¨ +

7 4 < ?

56© © @

-.-/ ,-./

¡¢ A

Eq. 4

-1

CDEFGHCI GJ KDFLB MNNN F O P B day or

8¨9: ;

¦¨

= >

8¨9: ;

¦¨

< © ? R 4 56©7

,-./ -. -/

¡¢ A

Eq. 5

-1

& +

!" ' !"()%* , -.

#$%

2 3

/01

& +

' !"()%* , - . !"

#$ %

Eq. 6

-1

5 6 7 6 89 : ; 4 DOcrit (see Fig. A1c), where tempd,i, lightd,i, and DOd,i,

-1 -1

6 : ; : ), and dissolved oxyg ),

respectively, at depth d and time step i.

Chironomidae

-2 >¦ The density of chironomid larvae : ), is determined daily as a product of

chironomid production and survival. We developed an equation from Goedkoop et al.

(2007) to calculate chironomid production based on carbon (i.e. food, mg C) and

temperature (°C):

CDEFGHIJEK L MNOMPOQRS HTDUEK VMNWXPP QYZQ [

Eq. 7

\ ]^ _`a_ _ \ _` a_ _

[ MN V MNMWMQI DT I GD MNMMOXQI DTIGD

Eq. 8

¢£¤¥ The temperature function, T(f), was determined from Reynolds & Benke (2005) ¡S scaling factor adapted from Goedkoop et al. (2007) to match the modeled output to the

-2

¤ £ £ £ ¥¤¤

Erie and was set equal to 2.6. Chironomid survival is determined as a function of daily temperature (i) using an equation developed with data from studies by Stevens (1998),

Vogt et al. (2007), Oetken et al. (2009), Baek et al. (2012), and Tassou & Schulz (2012):

!"#$#%&'() ' & ' +

, -. ,, -.

#! # %&'()'&' + ! $

Eq. 9

Additionally, the average monthly lengths observed in 2005 are used in the foraging model as the prey length. Lengths gradually change from day to day through linear interpolation between sampling dates allowing for a smooth transition in between each monthly average.

Appendix A2: Growth rate potential calculation

Our bioenergetics sub-model is an adaptation of a model presented by Arend et al.

-1 -1 -1

day-1) minus metabolism and wastes (Kitchell et al., 1977) using published parameters

for yellow perch (Hanson et al., 1997), rainbow smelt (Lantry & Stewart, 1993), emerald

shiner (Hanson et al., 1997; Duffy, 1998), and round goby (Lee & Johnson, 2005).

-1

¢£ ¢£ ¤ § ¦ ¢ ¤§ §¡ Metabolism is respiration + specific

-1 -1 -1

¢ ¢§ ¦§ ¢ ¦ ¦ ¦§ ¦£ ¦ ¤§ ¢§ ¦§ §¡ ¢ day ) (Table A1).

For adult round gobies, the respiration equation in the calculation of metabolism includes

an activity term. Since round gobies do not have swim bladders, respiration above the

benthic layers (e.g., bottom 1.5 m) in the water column is multiplied by 6.0 to account for

additional energetic costs (Arend et al., 2011).

We calculate consumption in a separate foraging sub-model integrating ten

minute time step input data throughout the calculations (see Appendix A3). Bioenergetics

models were developed and are typically applied at daily time steps. Thus, the output of

-1 -1

daily rate. Mean daily GRP values average all ten minute time steps in each day

Appendix A3: Foraging sub-model calculation

Consumption potential (i.e., the amount of energy an individual fish could

potentially consume each time step) in each depth cell during a ten minute time step is

modeled as a function of oxygen, temperature, light, and prey densities (Fig. A2, A3).

The foraging sub-model is based on a multi-species type-2 functional response similar to

Rose et al. (1999), but it also incorporates encounter rate and proportional capture

efficiency through mechanistic equations (e.g., volume searched based on swimming

speed, reactive distance, and the effect of light). The sub-model determines the potential

consumption (Cons) of each prey type (i) by a hypothetical individual fish at each depth

(d) using the equation:

¥¦¨© ¨

§ ¡¢ "#$%

£¤

Eq. 1

(')&' )* + ,-.+/+01)2/+(*.1) 3,*'4565 where Cmax is species-specific, temperature-&'

1 -1

6d , Kitchell et al., 1977), fish mass (g), k is the half saturation constant (i.e., the fastest

-1

78'3,*',*9:.0: , ; .2: 0 1 /8&01)2/+ ' (3' <=)/+7'3 (3' < 6 2 (12 2 . ) calibrated for each

prey type for each species and life-stage to account for handling time and prey preference

-1

'33,*' 4)/+7'3 (3' < 6 2 (Table A2), ER is the encoun * , Eq. 4-10), Q is the proportional

capture efficiency (Eq. 11), and fDO is a dissolved oxygen function (Eq. 2-3). Variable k

was calibrated to limit the maximum value of Cons for the year 2002 to 70% or less of

Cmax when nutrient loading was 1.0x as previous models have found that Cmax is more realistic at 0.6 or 0.8 times its value (May et al., 2012).

As in Arend et al. (2011), the dissolved oxygen function (fDO) is only applied to consumption when ambient oxygen concentration falls below a critical threshold

(DOcrit), thereby reducing individual consumption when dissolved oxygen concentration drops below tolerable levels.

fDO d = (1 / DOcrit) * DO d

Eq. 2

DOcrit d = 0.1683 * Temperature d + X

Eq. 3

¡¢£¢¤ ¥¦§¨¡¢ ©¦§ § ¢© ¢ ¢¨£¨¦ ¨¡¢ ¢ © ¤ ¥¦§ £ ¨¡¢ DO X more tolerant species (i.e., yellow perch and round goby) and 1.635 for the more sensitive species (i.e., rainbow smelt and emerald shiner) (Arend et al., 2011).

Encounter rates (ER) are calculated at each depth d and for each prey type i as the product of volume of water searched (VS, Eq. 4-5) times prey density for each depth and

-1 2 -1 § prey type. Volume searched, VS § for zooplankton or m for chironomids) represents the hypothetical volume of water an individual fish could search for prey each second and is calculated for zooplankton as the volume of half a cylinder with a length equal to the distance a fish could swim in one second (swimming speed, SS, Eq. 6) and a semi-circular area equal to the reactive area (RA, Eq. 7). For chironomids, VS calculates

benthic area searched (VSchironomids) as the area of a rectangle a fish could hypothetically

distance (RD, Eq. 8) (Breck, 1993). The rectangle has a length equal to SS and a width of ¥ 2 RD (Breck, 1993).

$ % &' %()

! "# !"#

Eq. 4

: 12 3 456 7

%() $ % )% & %

,- . / 0

! !# "#

!"# 9

Eq. 5

-1

==¥ ¦ >¤¤?==@ The SS; <¥ ), equation is based on field data from

Hergenrader & Hasler (1967), which assesses swimming speed in relation to temperature, and previous swimming speed calculations from Breck (1993), which integrates fish

length into the equation:

ABCDEF G/ HI BDJ

: 7

%NOPQRS $

AB E G -

K L M

Eq. 6

Although the Hergenrader & Hasler (1967) data are from yellow perch, the Breck (1993) inclusion of length into the equation allows us to incorporate some of the differences between our four fish species at the sizes we consider (Table A1).

Reactive area (RA, mm2) represents the total area in which an individual would potentially perceive and attack prey and was calculated as half of the area of a circle with

96 reactive distance (RD, mm, Eq. 8) as the radius, assuming that a fish can only see a half circle of area (Blaxter, 1986):

£¤

¥¦¨

¡¢

Eq. 7

The equation for RD is from Breck & Gitter (1983):

§ )*+,-./0 1 ¢

¡¢

! &

(

% %

"#$ '$

Eq. 8

Chironomid RD uses the same equation multiplied by 0.36, as it is assumed that the RD

for benthic prey is equal to that of pelagic prey that are 36% of the length of benthic prey

86692 5 6:86;<6<;< =865; >?8

(Breck, 1993). The angle of acuity of the predator 2 3 456 7

@6A7>BCA:86;<6<2 CA5 C:;>CA 7 >B6 =8 65; > ?8D<8;A76?EFCBCA:86;?>;@ size (Breck & Gitter, 1983):

O V

! © ©©

I' JK 'JLMN PQRSTU W$' ILMNOPQRSTUV

Eq. 9

Lightfac, determined from a logistic relationship, scales RD based on the available light

-1

¡ ¢ £ ¤¥ ¢ ) at depth d. Lightfac is limited between 0 and 0.5 so that light always strongly influences RD:

!!"# $%&'( **+

)

§¦¨ ©

Eq.

Proportional capture efficiency, Q, accounts for predator preference and prey density at each depth d and for each prey type i through the following equations:

%3)

% %3)

Eq. 11

where CE, capture efficiency, is the product of prey selectivity, prey mass, and prey density at depth d. The values for prey selectivity are dependent on predator and prey type (Table A3).

Appendix A4: Supplemental results on hindcast subdaily and daily variable and GRP responses

September 15, 2002 was chosen as a representative day because it occurred well after stratification and during the hypoxic episode in 2002, which was the representative year chosen because of its moderate temperatures. Throughout September 15, 2002, temperature and DO remained constant, changing with depth only where the warmer, oxygenated epilimnion and cooler, hypoxic hypolimnion meet (Fig. A2a, b). Light was the only physical variable that changed throughout the day (Fig. A2c). The effect of these three factors on prey biomass can be seen most clearly with the large zooplankton, as biomass was higher near the surface in low light periods and began to move deeper

through the water column, but avoids the hypolimnion, as light intensity increases (Fig.

¡¢£¤ ¥¦§¨© ¨ ¢ § ¢ ¦ ¢©¢§ ¢©¨¢ ¨ ¦ bitat preferences (Table 1 main text). Adult yellow perch had somewhat moderate GRP during daylight hours above the hypolimnion, adult rainbow smelt had low to negative GRP in every depth throughout the day, and adult emerald shiner had highly positive GRP during daylight hours above the hypolimnion (Fig. A2g, h, i).

Environmental conditions and GRP exhibited seasonal shifts in their daily patterns. Before June 1, 2002 and after October 1, 2002, temperature and DO were uniform throughout the water column (Fig. A3 row 1, row 2). Hypoxia developed at depths below approximately 16 m. in the beginning of July and continued until October

1. Light gradually increased in intensity both throughout the day and throughout the season, reaching deeper in the water at midday and from May to June, receding at dusk and as summer progresses (Fig. A3 row 3). Small zooplankton primarily responded to

99 daily temperature, as their biomass was evenly distributed throughout the water column at every time step (Fig. A3 row 4). Large zooplankton responded to temperature and light as they increased in biomass in June and performed diel vertical migrations throughout the water column (Fig. A3 row 5). Large zooplankton also responded to DO, with decreased biomass in the hypolimnion compared to higher in the water column in the summer and early fall. The highest biomass of chironomids was observed during the highest temperatures in the benthic layers (Fig. A3 row 6).

As with the sub-daily color contour plots (Fig. A2), seasonal consumption and

GRP demonstrated how species preferences and tolerances were integrated into their habitat quality. Adult yellow perch had moderate consumption in the epilimnion which was reflected in their moderate GRP in the epilimnion throughout the summer and fall

(Fig. A4 row 1, row 2). Adult rainbow smelt had negative GRP throughout most of the water column once thermal stratification occurs. Before hypoxia became extreme, they did have positive GRP in the hypolimninon which extended deeper in the water column with greater light at midday. Adult emerald shiner had higher GRP values than either of the other two species in the epilimnion during the warmest parts of the year (Fig. A4 row

4).

Annual average GRP was fairly consistent across years, except for a decline for most species and life-stages in 1998 and 1999 (Fig. A5). Overall, these patterns agreed with those observed with percent positive GRP and GRP-99%.

100

Appendix A5: Supplemental results on prey responses to increasing nutrients

In the nutrient scenarios, both zooplankton and chironomids responded to

increases in nutrients primarily during the times when they typically thrive (Fig. A6).

Large zooplankton increased in biomass in June and chironomid larvae increased during

September to mid-October. Across all temperature and timing years, annual average

-1

-1 L at 1.9 x nutrients. Maximum biomass of large zooplankton

-1 -1

increased from 6.11 (± 0.11) m L at 0.1 x nutrients to 6.55 (± 0.18) L at 1.9 x

-4 -1

nutrients. Chironomid annual average biomass increased from 0.045 (± 8.7 x 10 ) L

-3 -1

at 0.1 x nutrients to 0.051 (± 1.0 x 10 L at 1.9 x nutrients. Chironomid maximum

-1 -1

1.9 x nutrients.

101

Appendix Table A.1 Focal fish species bioenergetics model parameter estimates. The species codes are as follows: YPadult is adult yellow perch, YPyoy is young-of-year yellow perch, RSadult is adult rainbow smelt, RSyoy is young-of-year rainbow smelt, ESadult is adult emerald shiner, RGadult is adult round goby. CA, CB: mass dependence function intercept and slope, respectively for a 1 g fish; CQ: water temperature-dependent coefficient of consumption; CTO, CTM, CTL: optimum, upper, and maximum lethal water temperatures for consumption, respectively; CK1, CK4: fraction of the maximum consumption rate for CQ and CTL, respectively; RA, RB: mass dependence function intercept and slope, respectively for a 1 g fish; RQ: rate of respiration increase over low water temperatures; RTO, RTM: optimum and lethal water temperatures for respiration, respectively; ACT: activity, i.e. a constant multiplier of resting metabolism; FA, UA: proportion of consumption to egestion or excretion, respectively.

YPadulta YPyoya RSadultb RSyoyb ESadultc RGadultd Body Mass (g) 50.00 7.00 6.10 1.00 4.50 10.00 Length (mm) 156.13e 84.94e 106.69f 63.28f 90.52f 92.24g Consumption

-1 -1

£¤£ ¤ ¡ ¢ d ) 0.25 0.25 0.18 0.18 0.254 0.192 CB -0.27 -0.27 -0.275 -0.275 -0.276 -0.256 CQ (°C) 2.3 2.3 3.0 3.0 2.25 5.594 CTO (°C) 23.0 29.0 10.0 16.0 25.0 24.648 CTM (°C) 28.0 32.0 14.0h 21.0 30.0 25.706 CTL (°C) 20.0h 26.0 28.992 CK1 0.4 0.4 0.113 CK4 0.01 0.01 0.419 Respiration

-1 -1 ¤ RA (gO2 ¤ g d ) 0.0108 0.0108 0.0027 0.0027 0.0122 0.00094 RB -0.2 -0.2 -0.216 -0.216 -0.12 -0.157 RQ (°C-1) 2.1 2.1 0.036 0.036 2.35 0.061 RTO (°C) 28.0 32.0 29.0 RTM (°C) 33.0 35.0 33.0 ACT 1.0 1.0 1.0 1.0 1.0 1.0, 6.0i Egestion/Excretion FA 0.158 0.158 0.16 0.16 0.25 0.150 UA 0.0253 0.0253 0.10 0.10 0.10 0.10 Specific Dynamic Action 0.172 0.172 0.175 0.175 0.161 0.175

-1

-1 ¤£ ¢ wet) a Hanson et al. (1997) b Lantry & Stewart (1993) c Means of values reported in Hanson et al. (1997) for dace and Duffy (1998) for fathead minnow d Lee & Johnson (2005) e Mass to length conversion from Willis et al. (1991) f Mass to length conversion from Hartman & Margraf (1992)

102

Table A.1 continued g Mass to length conversion from Marsden et al. (1996) h Used modified values from Arend et al. (2011) i Higher ACT used for RGadult when more than 1.5 m above bottom of water column, described in text

103

-1

Appendix Table A.2 Half saturation constants (k ¡¢£ ¤¥¦ § ¨ © ) for all prey types for each species and age class. The species codes are as follows: YPadult is adult yellow perch, YPyoy is young-of-year yellow perch, RSadult is adult rainbow smelt, RSyoy is young-of-year rainbow smelt, ESadult is adult emerald shiner, RGadult is adult round goby.

YPadult RSadult YPyoy RSyoy ESadult RGadult Small 20,000 20,000 13,000 13,000 15,000 60,000 Zooplankton Medium 350 350 125 100 95 1000 Zooplankton Large 115 160 115 115 100 1000 Zooplankton Chironomids 5.0 6.0 10.0 12.0 400.0 3.95

104

Appendix Table A.3 Prey selectivity values and equations for all prey types for each species and age class. The species codes are as follows: YPadult is adult yellow perch, YPyoy is young-of-year yellow perch, RSadult is adult rainbow smelt, RSyoy is young- of-year rainbow smelt, ESadult is adult emerald shiner, RGadult is adult round goby.

Prey Type YPadult & - RSadult & - ESadult RGadult yoy yoy Prey Prey Selectivity Prey Prey Selectivity Selectivity Selectivity

a a a a -

¡¢£¡¡¤¥ ¦ ¡¢£¡¡¤¥ ¦ ¦ Small ¡¢£¡¡¤¥ ¦ 193499.0 Length Zooplankton Length-7.64 Length-7.64 Length-7.64 7.64

b c c a

Zooplankton 0.3834

Large b 0.276 c 0.0917 c 0.0089 a0.4 / (1.0 + ¤¥ ¦ Zooplankton ((0.4/0.09)

-13.0 0.031 ¦ e ¦ Length)1.0/0.031

d d d

¤¨ ¦ ¤¨ ¦ ¤¨ ¦

Chironomids |1.0 |0.7 0.0 |0.5

(PL/Length)| (PL/Length)| (PL/Length)| For

¥¤§

For For

0.2 0.2

!"# $

a Fulford et al. (2006)

!"# $

b Graeb et al. (2004)

#!' " ( #& ) # "# *+)+, "# #!+( - $ ,!)$#! )+# c % !& Pothoven et al. (2009) d Adjusted equations for flounder prey capture probability of benthic prey from Rose et al. (1996)

112

Appendix B Experimental Data

Appendix Table B.1 Experiment 1 movement and time in treatment data.

Trial Tank Population Treatment Movement Time in Treatment 1 1 Erie Control 0 0.0 1 2 NC Control 0 20.0 1 3 Erie Moderate 0 0.0 1 4 NC Control 1 6.6 1 5 Erie Low 1 9.0 1 6 Erie Low 1 1.9 1 7 Erie Low 1 13.8 1 8 Erie Low 0 15.0 1 10 NC Moderate 0 10.0 1 11 Erie Moderate 0 0.0 1 12 NC Low 1 16.7 1 13 Erie Moderate 0 20.0 1 14 NC Control 1 0.7 1 15 NC Low 1 7.6 1 16 NC Moderate 1 2.3 1 17 NC Control 1 19.1 1 18 NC Moderate 1 14.0 1 19 NC Low 1 13.0 1 20 Erie Control 0 0.0 1 21 Erie Control 0 0.0 1 22 Erie Low 0 10.0 1 23 Erie Moderate 0 0.0 1 24 Erie Control 0 0.0 1 25 Erie Control 1 1.5 1 27 NC Low 1 1.2 1 28 Erie Moderate 1 3.1 1 29 NC Low 1 19.7 1 30 NC Moderate 1 8.6 2 2 Erie Control 0 0.0 2 3 Erie Moderate 0 5.0

113

2 4 Erie Low 0 0.0 2 5 Erie Low 0 5.0 2 7 Erie Low 0 5.0 2 8 Erie Control 0 5.0 2 9 NC Control 1 14.1 2 10 NC Moderate 0 10.0 2 11 NC Moderate 0 5.0 2 13 NC Control 1 4.6 2 15 NC Control 0 0.0 2 16 Erie Moderate 0 0.0 2 17 Erie Control 0 0.0 2 18 Erie Control 0 20.0 2 19 NC Low 0 0.0 2 20 NC Low 1 5.7 2 21 Erie Moderate 0 0.0 2 22 NC Moderate 0 10.0 2 23 Erie Moderate 0 0.0 2 24 Erie Control 0 5.0 2 25 NC Low 1 7.7 2 26 NC Moderate 1 12.8 2 27 NC Low 0 15.0 2 28 NC Control 0 0.0 2 29 Erie Low 0 0.0 2 30 Erie Low 0 0.0 3 1 NC Moderate 0 15.0 3 2 NC Low 0 0.0 3 3 NC Control 0 10.0 3 4 Erie Control 0 20.0 3 6 Erie Low 0 0.0 3 7 Erie Low 0 10.0 3 8 Erie Moderate 0 0.0 3 9 NC Low 1 5.3 3 10 NC Control 0 20.0 3 11 NC Control 0 5.0 3 12 Erie Moderate 0 0.0 3 13 Erie Control 0 15.0 3 14 Erie Control 0 10.0 3 16 Erie Control 0 10.0 3 17 NC Control 0 5.0 3 18 NC Control 1 5.3 3 19 Erie Moderate 1 4.2

114

3 20 NC Low 0 0.0 3 21 Erie Moderate 0 10.0 3 22 Erie Moderate 1 5.6 3 23 NC Moderate 0 0.0 3 24 Erie Low 0 0.0 3 26 NC Moderate 0 0.0 3 27 NC Moderate 1 5.2 3 28 NC Moderate 0 5.0 3 29 NC Low 0 0.0 3 30 NC Low 0 0.0 4 1 NC Low 0 0.0 4 2 NC Low 0 0.0 4 3 Erie Low 0 15.0 4 4 Erie Moderate 0 10.0 4 6 NC Moderate 0 15.0 4 7 Erie Low 0 15.0 4 8 NC Low 1 6.5 4 9 NC Low 0 0.0 4 11 NC Moderate 0 0.0 4 12 Erie Low 1 6.8 4 13 Erie Low 0 0.0 4 14 Erie Moderate 0 15.0 4 15 NC Low 0 5.0 4 17 NC Control 0 0.0 4 18 NC Control 1 10.0 4 19 NC Moderate 0 10.0 4 20 Erie Control 0 0.0 4 21 Erie Moderate 0 10.0 4 22 NC Moderate 0 10.0 4 23 Erie Moderate 0 10.0 4 24 NC Control 0 15.0 4 25 NC Moderate 0 5.0 4 26 NC Control 0 5.0 4 27 Erie Control 0 0.0 4 28 NC Control 1 13.0 4 29 Erie Control 1 16.8 4 30 Erie Control 0 5.0

115

Appendix Table B.2 Experiment 2 (Acute Exposure) fold data.

¢ Tissue Tank Population Treatment hif- ¡ igf- cyp19a1a socs3 hsp70 Brain 3 Erie Control 1.01 0.92 1.09 1.65 1.22 Brain 4 Erie Control 0.72 0.57 0.41 0.78 0.63 Brain 5 Erie Control 0.95 1.34 1.06 1.42 0.96 Brain 6 Erie Control 1.09 1.48 1.03 1.05 1.00 Brain 11 Erie Control 1.25 1.84 1.18 1.37 1.32 Brain 14 Erie Control 1.26 1.93 1.44 1.31 1.34 Brain 7 Erie Exposure 1.26 1.32 3.10 1.98 1.51 Brain 15 Erie Exposure 0.97 0.99 1.02 1.55 0.79 Brain 18 Erie Exposure 1.27 1.85 2.46 3.00 1.56 Brain 19 Erie Exposure 1.45 2.25 3.58 2.93 2.38 Brain 20 Erie Exposure 0.88 0.71 0.33 0.98 0.69 Brain 22 Erie Exposure 1.47 2.19 1.58 1.80 1.13 Brain 1 NC Control 1.00 1.00 1.00 1.00 1.00 Brain 2 NC Control 1.00 0.85 0.78 1.31 0.80 Brain 10 NC Control 0.96 0.89 0.95 1.49 1.02 Brain 13 NC Control 1.23 1.38 3.46 2.34 1.48 Brain 16 NC Control 0.56 0.22 0.54 0.71 0.43 Brain 17 NC Control 1.03 0.68 0.70 0.48 0.61 Brain 8 NC Exposure 1.10 1.10 1.24 1.04 0.92 Brain 9 NC Exposure 0.86 0.84 0.98 1.37 0.93 Brain 12 NC Exposure 1.86 2.60 2.01 2.27 1.61 Brain 21 NC Exposure 1.20 1.25 2.03 3.02 1.50 Brain 23 NC Exposure 0.93 0.75 2.51 1.38 1.18 Brain 24 NC Exposure 0.65 0.40 0.45 1.31 0.90 Gill 3 Erie Control 1.29 1.44 0.22 0.38 0.87 Gill 4 Erie Control 1.14 0.59 1.29 0.54 0.76 Gill 5 Erie Control 1.17 0.27 0.31 0.03 0.06 Gill 6 Erie Control 1.05 1.21 0.86 0.31 0.37 Gill 11 Erie Control 3.01 10.42 0.63 0.60 1.11 Gill 14 Erie Control 1.26 3.94 0.94 0.52 1.08 Gill 7 Erie Exposure 1.71 0.65 0.70 0.06 0.08 Gill 15 Erie Exposure 2.59 10.67 1.61 0.99 1.08 Gill 18 Erie Exposure 1.57 2.25 0.23 0.17 0.29 Gill 19 Erie Exposure 1.16 3.31 1.03 0.74 0.78 Gill 20 Erie Exposure 1.54 4.48 0.54 0.26 0.35 Gill 22 Erie Exposure 1.88 4.44 1.42 0.69 0.75 Gill 1 NC Control 1.00 1.00 1.00 1.00 1.00 Gill 2 NC Control 1.75 5.31 1.14 0.59 0.83

116

Gill 10 NC Control 0.90 0.51 0.76 0.58 0.49 Gill 13 NC Control 1.36 3.73 1.64 0.74 0.68 Gill 16 NC Control 2.46 6.37 1.08 0.67 1.00 Gill 17 NC Control 1.76 5.10 0.35 0.23 0.80 Gill 8 NC Exposure 2.36 7.86 2.12 0.80 1.11 Gill 9 NC Exposure 2.11 10.27 1.24 0.67 1.06 Gill 12 NC Exposure 2.10 7.17 0.28 0.36 0.87 Gill 21 NC Exposure 1.07 3.35 1.06 0.07 0.13 Gill 23 NC Exposure 0.68 0.95 0.08 0.13 0.16 Gill 24 NC Exposure 2.71 10.32 2.20 1.18 1.07 Liver 3 Erie Control 1.77 0.57 0.11 0.17 0.29 Liver 4 Erie Control 1.54 3.48 1.66 1.67 1.98 Liver 5 Erie Control 2.08 3.94 2.26 2.44 2.18 Liver 6 Erie Control 0.83 2.01 0.25 0.75 0.98 Liver 11 Erie Control 1.33 1.75 1.28 0.82 0.90 Liver 14 Erie Control 1.43 1.15 0.28 0.53 0.49 Liver 7 Erie Exposure 2.08 6.54 2.43 1.53 1.75 Liver 15 Erie Exposure 1.51 2.50 1.76 1.25 1.33 Liver 18 Erie Exposure 0.95 1.49 0.79 0.43 0.35 Liver 19 Erie Exposure 1.31 1.63 3.48 2.14 1.01 Liver 20 Erie Exposure 0.97 0.71 0.18 0.21 0.17 Liver 22 Erie Exposure 4.38 1.34 2.06 1.24 1.33 Liver 1 NC Control 1.00 1.00 1.00 1.00 1.00 Liver 2 NC Control 1.45 2.02 1.70 1.45 1.49 Liver 10 NC Control 9.78 1.41 0.62 0.28 0.79 Liver 13 NC Control 1.86 6.23 2.48 2.77 2.74 Liver 16 NC Control 2.20 2.94 7.91 1.81 3.10 Liver 17 NC Control 0.95 1.14 0.73 0.57 0.54 Liver 8 NC Exposure 1.07 2.49 1.92 0.59 0.96 Liver 9 NC Exposure 1.42 1.72 1.00 0.65 0.93 Liver 12 NC Exposure 1.23 1.53 0.53 0.46 0.59 Liver 21 NC Exposure 1.54 1.09 1.60 0.60 0.53 Liver 23 NC Exposure 0.97 2.76 1.29 0.86 0.85 Liver 24 NC Exposure 1.55 1.65 6.93 3.27 4.77 Muscle 3 Erie Control 1.43 2.08 3.14 1.85 1.47 Muscle 4 Erie Control 1.24 0.79 1.45 0.87 0.93 Muscle 5 Erie Control 1.01 1.36 1.54 1.03 1.22 Muscle 6 Erie Control 0.96 1.88 1.24 1.10 0.83 Muscle 11 Erie Control 1.12 1.96 1.92 1.88 1.10 Muscle 14 Erie Control 1.22 1.52 1.58 0.61 1.24 Muscle 7 Erie Exposure 1.20 1.71 2.07 1.23 1.23

117

Muscle 15 Erie Exposure 1.28 1.38 0.60 0.80 1.14 Muscle 18 Erie Exposure 1.77 1.01 1.91 1.47 1.35 Muscle 19 Erie Exposure 1.21 1.01 2.05 1.31 0.90 Muscle 20 Erie Exposure 1.01 2.55 0.79 0.66 1.22 Muscle 22 Erie Exposure 1.21 1.62 2.67 1.64 1.42 Muscle 1 NC Control 1.00 1.00 1.00 1.00 1.00 Muscle 2 NC Control 1.32 1.28 1.25 1.20 1.68 Muscle 10 NC Control 1.39 1.83 1.25 1.43 0.75 Muscle 13 NC Control 1.12 0.61 1.34 1.46 0.80 Muscle 16 NC Control 0.99 0.81 1.38 0.85 1.25 Muscle 17 NC Control 1.16 1.15 1.81 0.86 0.95 Muscle 8 NC Exposure 1.56 2.25 1.55 1.09 1.28 Muscle 9 NC Exposure 1.16 1.15 1.01 0.94 1.07 Muscle 12 NC Exposure 1.62 2.19 1.83 1.22 1.15 Muscle 21 NC Exposure 1.17 0.61 1.22 0.95 0.94 Muscle 23 NC Exposure 1.13 1.17 0.38 0.38 1.11 Muscle 24 NC Exposure 1.31 1.08 2.06 1.68 1.11

118

Appendix Table B.3 Experiment 3 (Chronic Exposure) growth, consumption, and [Hb] data.

Consumption

-1 -1

¡ ¢£ Tank Treatment Population Growth (mm) ¡ ¢ ) [Hb] 1 Control NC 0.33 0.00 47.6 -4

2 Control NC 0.67 3.62 ¢¤¥ 45.4 -5

4 Low NC 0.00 2.79 ¢¤¥ 42.3 -4

5 Control NC 0.33 4.4¥¢¤¥ 46.3 -4

6 Low NC -0.33 4.38 ¢¤¥ 34.8 -4

7 Moderate Erie 0.33 6.81 ¢¤¥ 50.7 -5

8 Moderate Erie 0.00 9.40 ¢¤¥ 49.6 -4

9 Moderate NC 1.00 3.53 ¢¤¥ 37.8 -5

10 Control Erie 0.33 7.49 ¢¤¥ 52.9 -3

11 Moderate NC 0.33 1.513 ¢¤¥ 43.0 -5

12 Moderate Erie 0.00 8.18 ¢¤¥ 46.1 -4

13 Low Erie -0.33 6.75 ¢¤¥ 46.2 -4

14 Low NC 0.33 2.6¥¢¤¥ 53.2 15 Moderate NC -0.33 0.00 45.1 16 Control Erie 0.00 0.00 34.9 17 Moderate NC 0.00 0.00 44.1 -4

18 Control NC -0.33 9.01 ¢¤¥ 44.3 -4

19 Control Erie 0.00 5.41 ¢¤¥ 36.9

-3

§¨ ¢¤¥ 20 Low Erie 0.67 1 45.9 -4

21 Control Erie 0.33 1.21 ¢¤¥ 47.5 22 Low Erie 0.33 0.00 50.4 23 Moderate Erie 0.00 0.00 49.8 -3

24 Low NC -0.67 1§¥¥ ¢ ¤¥ 35.3

119

Appendix Table B.4 Experiment 3 (Chronic Exposure) fold data.

¢ Tissue Tank Population Treatment hif- ¡ igf- cyp19a1a socs3 hsp70 Brain 10 Erie Control 1.03 1.65 1.22 0.91 1.22 Brain 16 Erie Control 1.12 1.44 0.77 1.46 1.10 Brain 19 Erie Control 0.99 1.61 1.20 1.41 1.38 Brain 21 Erie Control 1.00 1.60 1.23 1.27 1.13 Brain 7 Erie Moderate 0.98 1.18 0.67 0.74 0.73 Brain 8 Erie Moderate 0.91 1.51 0.77 1.06 0.85 Brain 12 Erie Moderate 0.99 1.34 1.33 1.70 1.04 Brain 23 Erie Moderate 1.02 0.80 0.75 1.29 1.02 Brain 3 Erie Low 0.97 0.65 0.65 1.31 0.74 Brain 13 Erie Low 0.92 0.96 1.36 1.24 1.19 Brain 20 Erie Low 1.06 1.81 1.07 1.31 1.43 Brain 22 Erie Low 0.96 1.23 1.23 1.92 1.10 Brain 1 NC Control 0.85 1.11 0.83 0.95 0.90 Brain 2 NC Control 1.00 1.42 1.16 0.72 0.82 Brain 5 NC Control 0.99 1.29 1.30 1.09 1.34 Brain 18 NC Control 1.00 1.00 1.00 1.00 1.00 Brain 9 NC Moderate 1.00 1.04 0.95 0.89 1.01 Brain 11 NC Moderate 1.05 0.84 1.23 1.31 1.02 Brain 15 NC Moderate 0.99 0.79 0.80 0.91 0.74 Brain 17 NC Moderate 0.89 0.57 0.91 0.92 0.99 Brain 4 NC Low 1.10 0.95 1.41 1.72 1.02 Brain 6 NC Low 0.90 0.36 1.13 1.17 1.13 Brain 14 NC Low 1.03 1.40 1.31 1.85 1.32 Brain 24 NC Low 0.81 0.41 0.55 0.86 0.88 Gill 10 Erie Control 0.71 1.00 0.71 0.52 0.60 Gill 16 Erie Control 1.21 1.43 0.50 0.61 0.43 Gill 19 Erie Control 1.18 1.24 0.88 1.00 1.44 Gill 21 Erie Control 0.52 0.17 0.10 0.40 0.76 Gill 7 Erie Moderate 0.77 0.93 0.91 0.48 0.84 Gill 8 Erie Moderate 0.75 0.68 0.83 0.55 0.63 Gill 12 Erie Moderate 0.71 0.27 0.81 0.85 0.60 Gill 23 Erie Moderate 0.71 0.11 0.49 0.27 0.77 Gill 3 Erie Low 0.84 1.09 0.83 0.68 0.87 Gill 13 Erie Low 0.70 0.46 1.37 0.44 0.52 Gill 20 Erie Low 0.83 0.69 0.85 0.44 0.48 Gill 22 Erie Low 0.70 0.31 0.41 0.68 0.67 Gill 1 NC Control 0.64 0.22 0.57 0.52 0.64 Gill 2 NC Control 0.73 0.43 0.43 0.46 0.71

120

Gill 5 NC Control 0.68 0.51 0.87 0.59 1.02 Gill 18 NC Control 1.00 1.00 1.00 1.00 1.00 Gill 9 NC Moderate 0.71 0.38 0.35 0.42 0.42 Gill 11 NC Moderate 0.94 1.39 1.26 0.67 1.00 Gill 15 NC Moderate 0.86 0.66 0.84 0.56 1.40 Gill 17 NC Moderate 0.93 0.50 1.53 1.17 1.42 Gill 4 NC Low 1.35 2.33 1.20 1.01 0.92 Gill 6 NC Low 0.77 1.08 0.79 0.97 1.07 Gill 14 NC Low 0.78 1.06 0.88 0.71 0.76 Gill 24 NC Low 0.53 0.53 0.14 0.30 0.39 Liver 10 Erie Control 0.72 1.11 0.66 1.73 0.94 Liver 16 Erie Control 0.86 1.06 0.65 1.48 1.41 Liver 19 Erie Control 0.86 1.16 0.82 1.31 1.09 Liver 21 Erie Control 0.94 0.97 0.48 0.82 0.83 Liver 7 Erie Moderate 0.87 1.23 0.44 0.64 0.93 Liver 8 Erie Moderate 1.02 1.07 0.50 1.10 1.43 Liver 12 Erie Moderate 1.63 1.01 1.41 1.45 1.20 Liver 23 Erie Moderate 1.15 1.51 1.33 0.89 1.24 Liver 3 Erie Low 0.85 1.22 0.42 0.89 1.07 Liver 13 Erie Low 1.09 1.59 0.79 1.00 1.69 Liver 20 Erie Low 1.06 1.03 0.35 1.62 1.49 Liver 22 Erie Low 0.92 1.51 0.50 1.12 1.46 Liver 1 NC Control 0.99 1.19 0.77 2.93 1.41 Liver 2 NC Control 0.51 0.12 0.36 0.67 0.26 Liver 5 NC Control 1.00 1.16 1.00 2.10 1.80 Liver 18 NC Control 0.46 1.00 1.99 1.00 1.00 Liver 9 NC Moderate 0.73 1.27 0.83 1.25 1.09 Liver 11 NC Moderate 0.98 1.32 0.77 0.77 1.19 Liver 15 NC Moderate 0.92 1.30 1.32 1.57 1.11 Liver 17 NC Moderate 0.84 1.04 1.81 0.66 0.86 Liver 4 NC Low 0.90 1.20 1.79 1.22 1.22 Liver 6 NC Low 0.96 1.13 0.56 0.80 1.43 Liver 14 NC Low 0.87 1.29 1.13 2.72 1.43 Liver 24 NC Low 0.90 1.03 0.89 1.01 1.36 Muscle 10 Erie Control 1.38 3.52 0.53 0.38 0.53 Muscle 16 Erie Control 0.90 2.77 0.75 0.48 0.28 Muscle 19 Erie Control 1.41 0.83 1.52 0.56 0.71 Muscle 21 Erie Control 1.17 0.72 2.22 0.88 0.62 Muscle 7 Erie Moderate 1.46 1.94 1.87 0.80 0.61 Muscle 8 Erie Moderate 1.37 1.33 1.74 0.67 0.57 Muscle 12 Erie Moderate 1.18 2.46 3.27 1.19 0.52

121

Muscle 23 Erie Moderate 1.03 1.27 0.88 0.41 0.31 Muscle 3 Erie Low 1.33 0.78 2.06 0.74 0.49 Muscle 13 Erie Low 1.67 3.51 2.06 0.91 0.56 Muscle 20 Erie Low 1.02 0.95 0.43 0.47 0.63 Muscle 22 Erie Low 1.21 1.20 2.08 0.54 0.47 Muscle 1 NC Control 1.13 0.80 0.83 0.45 0.87 Muscle 2 NC Control 1.49 1.67 1.74 0.74 0.61 Muscle 5 NC Control 1.00 0.90 1.00 0.47 0.39 Muscle 18 NC Control 1.33 1.00 2.73 1.00 1.00 Muscle 9 NC Moderate 0.96 0.47 1.97 0.83 0.61 Muscle 11 NC Moderate 1.01 2.45 1.05 0.48 0.67 Muscle 15 NC Moderate 1.25 0.96 1.10 0.45 0.72 Muscle 17 NC Moderate 1.17 2.69 1.68 0.49 0.46 Muscle 4 NC Low 1.08 0.35 0.92 0.57 0.30 Muscle 6 NC Low 1.31 2.12 1.50 0.81 0.77 Muscle 14 NC Low 1.03 0.78 2.93 0.80 0.95 Muscle 24 NC Low 0.99 1.57 1.61 0.57 0.36

122

(a)

(b)

Appendix Figure B.1 Differences in initial total length for Erie and NC populations in (a)

¢£¤ ¥ ¦ §¤ ¢£¨ © § §¤ ¢£¨ © § Experiment 2 (acute) control and ¡ treatments and (b) Experiment 3 (chronic) control, moderate, and low treatments. Center line in box plots shows median, and upper and lower edges of the box are the 25th and 75th percentiles, and error bars are the 10th and 90th percentiles.

123

Appendix C Supplementary References

Andersen T, Hessen DO (1991) Carbon, nitrogen, and phosphorus content of freshwater

zooplankton. Limnology and Oceanography, 36, 807 814.

Arend KK, Beletsky D, Depinto J V. et al. (2011) Seasonal and interannual effects of

hypoxia on fish habitat quality in central Lake Erie. Freshwater Biology, 56, 366

383.

Baek MJ, Yoon TJ, Bae YJ (2012) Development of Glyptotendipes tokunagai (Diptera:

Chironomidae) under different temperature conditions. Environmental Entomology,

41, 950 958.

Blaxter JHS (1986) Development of sense organs and behavior of teleost larvae with

special reference to feeding and predator avoidance. Transactions of the American

Fisheries Society, 115, 98 114.

Breck JE (1993) Hurry up and wait: Growth of young bluegills in ponds and in

simulations with an individual-based model. Transactions of the American Fisheries

Society, 122, 467 480.

Breck JE, Gitter MJ (1983) Effect of fish size on the reactive distance of bluegill

(Lepomis macrochirus) sunfish. Canadian Journal of Fisheries and Aquatic

Sciences, 40, 162 167.

124

Duffy WG (1998) Population dynamics, production, and prey consumption of fathead

minnows (Pimephales promelas) in prairie wetlands: a bioenergetics approach.

Canadian Journal of Fisheries and Aquatic Sciences, 55, 15 27.

Fulford RS, Rice JA, Miller TJ, Binkowski FP, Dettmers JM, Belonger B (2006)

Foraging selectivity by larval yellow perch (Perca flavescens): Implications for

understanding recruitment in small and large lakes. Canadian Journal of Fisheries

and Aquatic Sciences, 63, 28 42.

Goedkoop W, Demandt M, Ahlgren G (2007) Interactions between food quantity and

quality (long-chain polyunsaturated fatty acid concentrations) effects on growth and

development of Chironomus riparius. Canadian Journal of Fisheries and Aquatic

Sciences, 64, 425 436.

Graeb BDS, Dettmers JM, Wahl DH, Cáceres CE (2004) Fish size and prey availability

affect growth, survival, prey selection, and foraging behavior of larval yellow perch.

Transactions of the American Fisheries Society, 133, 504 514.

Hanson PC, Johnson TB, Schindler DE, Kitchell JF (1997) Fish bioenergetics 3.0 for

Windows, Vol. WISCU-T-97.

Hartman KJ, Margraf FJ (1992) Effects of prey and predator abundances on prey

consumption and growth of walleyes in western Lake Erie. Transactions of the

American Fisheries Society, 121, 245 260.

125

Hergenrader GL, Hasler AD (1967) Seasonal changes in swimming rates of yellow perch

in Lake Mendota as measured by sonar. Transactions of the American Fisheries

Society, 96, 373 382.