Ecological and Physiological Effects of Environmental Stressors Across Life-Stages in Amphibians

ECOLOGICAL AND PHYSIOLOGICAL EFFECTS OF ENVIRONMENTAL STRESSORS ACROSS LIFE-STAGES IN AMPHIBIANS

KACEY L. DANANAY

Submitted in partial fulfillment of requirements

For the degree of Doctor of Philosophy

Advisor: Dr. Michael F. Benard

Department of Biology

CASE WESTERN RESERVE UNIVERSITY

August 2018

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

Kacey L. Dananay

candidate for the Doctor of Philosophy degree*.

Committee Chair:

Karen Abbott

Committee Member:

Michael Benard

Committee Member:

Jean Burns

Committee Member:

Mandi Schook

Committee Member:

Mark Willis

Date of Defense: March 19th, 2018

*We also certify that written approval has been obtained for any proprietary material contained therein.

iii

DEDICATION

To my family and friends who have supported me through this journey

TABLE OF CONTENTS

List of Tables ...... viii

List of Figures ...... ix

Acknowledgements ...... x

Abstract ...... 1

Chapter 1: Introduction ...... 3

Environmental stressors ...... 3

Road salt runoff...... 3

Artificial Light At Night (ALAN) ...... 4

Research program and goals ...... 5

Chapter 2: Legacy of Road Salt: Apparent Positive Larval Effects Counteracted by Negative Post Metamorphic Effects in Wood Frogs ...... 7

Abstract ...... 7

Introduction ...... 8

Materials and Methods ...... 11

Field Survey: ...... 11

Statistical Analyses: ...... 12

Experimental Venue: ...... 13

Experiment 1: Salt concentration: ...... 14

Statistical Analyses ...... 16

Experiment 2: Carry-over effects: ...... 16

Statistical Analyses ...... 18

Results ...... 18

Field Survey: ...... 18

Experiment 1: Salt concentration ...... 19

Experiment 2: Carry-over effects ...... 21

Discussion ...... 22

Tables ...... 27

Figures...... 29

Chapter 3: Artificial Light at Night decreases metamorphic duration and increases juvenile growth in a widespread amphibian ...... 35

Abstract ...... 35

Introduction ...... 36

Materials and Methods ...... 38

Study System ...... 38

Experiment 1: Testing for direct and indirect effect of ALAN in larvae ...... 38

Statistical Analyses ...... 42

Experiment 2: Prior and subsequent effects of light on post-metamorphic toads ...... 43 Statistical Analyses ...... 45

Results ...... 46

Experiment 1: Testing for direct and indirect effect of ALAN in larvae ...... 46

Experiment 2: Prior and subsequent effects of light on post-metamorphic toads ...... 47

Discussion ...... 47

Figures ...... 51

Supplementary Tables ...... 53

Supplementary Figures ...... 56

Chapter 4: Artificial Light At Night and Predator-Prey Dynamics: consumptive and non-consumptive effects of predators are not reflected in corticosterone of American toads ...... 60

Abstract ...... 60

Introduction ...... 61

Materials and Methods ...... 64

Experiment 1: Physiological effects of ALAN and predator presence on larval

American toads ...... 64

Statistical Analysis ...... 67

Experiment 2: Effects of larval stage and juvenile stage ALAN on juvenile

Toads ...... 68

Statistical Analysis ...... 69

Hormone extraction and corticosterone radioimmunoassay (RIA) procedures ....70

Results ...... 71

Experiment 1: Physiological effects of ALAN and predator presence on larval

American toads ...... 71

Experiment 2: Effects of larval stage and juvenile stage ALAN on juvenile

Toads ...... 72

Discussion ...... 73

Tables ...... 79

Figures...... 81

Chapter 5: Conclusions: ...... 83

Advancements of my research program ...... 83

Future directions ...... 85

Bibliography: ...... 88

vii

LIST OF TABLES

Chapter 2:

Table 2.1: ...... 27

Chapter 3:

Table S3.1...... 53

Table S3.2...... 54

Table S3.3...... 55

Table S3.4...... 55

Chapter 4:

Table 4.1 ...... 79

Table 4.2 ...... 80

Table 4.3 ...... 80

viii

LIST OF FIGURES

Chapter 2:

Figure 2.1: ...... 30

Figure 2.2: ...... 31

Figure 2.3: ...... 32

Figure 2.4: ...... 33

Chapter 3:

Figure 3.1: ...... 51

Figure 3.2: ...... 51

Figure 3.3: ...... 52

Figure S3.1: ...... 56

Figure S3.2: ...... 57

Figure S3.3: ...... 58

Figure S3.4: ...... 59

Figure S3.5: ...... 59

Chapter 4:

Figure 4.1: ...... 81

Figure 4.2: ...... 82

ACKNOWLEDGEMENTS

Throughout my six years at CWRU, I have had the opportunity to work with many wonderful scholars and friends. This dissertation would not have been possible without all their help and encouragement.

I would first and foremost like to thank my advisor, Dr. Michael Benard. His combination of unwavering support and guidance when I needed it and intellectual freedom to explore exciting new research ideas, was exactly what I needed to become the scientist I am today. His patience, encouragement and insight has in many ways taught me how to advise and mentor my future students. I will be forever proud to be part of the

Benard Lab family.

I would also like to thank my many committee members for their support. I thank

Dr. Mandi Schook for her invaluable help in allowing me to explore new aspects of my research that I have been unable to achieve until now. Her excitement, training and direction were invaluable to this dissertation and to my development as a physiologist. I also thank Dr. Jean Burns for her support both as an instructor and as a mentor. My work benefited from her insightful questions, encouraging suggestions, and brilliant instruction starting my first year. I am also grateful for the help from Dr. Mark Willis. I am appreciative of his helpful feedback, support and enthusiasm for this project.

I would also like to thank the Benard Lab for their steadfast support, laughter in times of need, and long-lasting memories that I will carry with me. In particular, I would like to thank Rebecca Benard, Katherine Krynak, Mimi Guo, Hilary Rollins, Laura x

Dimayuga, Erin Conway, Timothy Nicholas, Catherine Chervenak, Addie Klimek, and

David Dimitrie. This experience truly would not have been the same without all of you.

Each of you has impacted me in profound ways that will never be forgotten.

I also want to thank the Oglebay Grant for financial support for traveling to conferences. I am also grateful for the logistical support provided by CWRU’s Squire

Valleevue and Valley Ridge Farm and Cleveland Metroparks Zoo. In particular, I would like to thank Ana Locci, Shane Brown, and Joe Miller at the CWRU Farm for all their help and support. Your help moving large artificial ponds and digging terrestrial pens was invaluable. Laura Amendolagine at the Cleveland Metroparks Zoo, was also a crucial player in helping me complete my dissertation. Her knowledge, guidance, patience, humor, and friendship made working at the Cleveland Zoo an amazing experience.

Last but certainly not least, I would like to thank my family. A special thank you for Colin Cope, who was always patient when stress got the better of me. Thank you for supporting me when I needed it. I also want to thank my parents and brother who, from the start, instilled in me a love for nature and taught me to follow my dreams. And finally, I want to thank Smokey, Bandit, Cinder and Eva, who kept me sane when reading and writing. Thank you for patiently sitting with me for hours and always knowing the perfect moment for a break.

Ecological and Physiological Effects of Environmental Stressors Across Life-stages in Amphibians

KACEY L. DANANAY

ABSTRACT

Understanding the effects of environmental stressors includes identifying how stressors affect individual physiology and ecological communities. Considering carry- over effects, effects from one life-stage persisting into later life-stages, can further reduce the chances of under- or over-estimating the effects of environmental stressors. I investigated the effects of two environmental stressors on amphibian physiology across life-stages including direct and indirect effects.

I first investigated road salt. Road salt, a de-icing agent used on highways, can spread up to 1km in wetlands during snowmelt. It may be particularly important for early breeding amphibians like wood frogs (Rana sylvatica). Road salt significantly increased larval frog growth and algal biomass which was likely due to an indirect effect of salt decreasing zooplankton abundance, an algal competitor of frogs. A second experiment found despite increased larval growth, exposure to road salt caused juvenile frogs to have higher mortality in low-density terrestrial environments.

The second stressor I investigated was Artificial Light At Night (ALAN). ALAN reduced metamorphic duration of American toads (Anaxyrus americanus) and periphyton biomass but did not affect the colonization of toad predators. These results suggested the effects of ALAN are mediated through direct rather than indirect effects. Extending this 1

experiment found juvenile growth was reduced by juvenile-stage exposure to ALAN.

Increased juvenile activity, specifically the lack of suppressed nocturnal activity, likely reduced juvenile growth of individuals housed with ALAN. Furthermore, carry-over effects were also present; larval-stage ALAN marginally increased juvenile activity.

In the final experiment, I added an additional stressor: predation. Predators reduced toad survival and mass, regardless of ALAN. This suggests ALAN did not increase predator consumption of toads. Neither predators nor ALAN affected corticosterone production in the tadpole or metamorph life-stages but larval-stage ALAN increased corticosterone production in juvenile toads.

These experiments demonstrated environmental stressors can have direct and indirect effects. Furthermore, larval stage stressors can carry-over and affect later life- stages even if that stressor is no longer present. Future environmental stressor studies should investigate direct and indirect effects together and extend experiments beyond a single life-stage. As demonstrated here, failure to do so may under-estimated the effects of these environmental stressors.

Chapter 1: Introduction

In this chapter, I will give a brief overview of 1) effects of environmental stressors, 2) effects of road salt runoff, 3) the effects artificial light at night, and 4) my research program and overarching goals.

Environmental stressors

Organisms must respond to many environmental stressors. Environmental stressors can be caused by natural processes, such as predation or competition, or anthropogenic alterations, such as pollution or habitat fragmentation. Species across the globe are facing drastic environmental changes, many of which caused by anthropogenic disturbances [1]. As a result, the planet is currently facing an elevated global extinction rate [2,3]. Anthropogenic stressors can arise from various sources and have diverse effects individually and synergistically [4]. Two different environmental stressors are focused on here: road salt runoff and artificial light at night.

Road salt runoff

Since 1930, de-icing agents, like road salt, have been applied to roads and walkways world-wide [5]. Approximately 18 million metric tons of road salt are applied across the United States annually; in Ohio alone, over 600,000 metric tons of salt are used each year [6]. Road salt is a freshwater contaminant that can disrupt water consumption and ion balance, alter morphological traits and affect community composition, species density, richness and diversity [7–9]. Furthermore, road salt applied to one area can spread as far as 1 km in wetlands [10]. Most research on road salt has 3

been performed in short-term laboratory trials, often with ecologically unrealistic concentrations. There are also many different deicing agents used in these studies, some of which (e.g. pickling salt [11] or table salt [12]) are likely not being used as road salt or have different chemical composition compared to true de-icers. Due to the design of these experiments, it was unclear if laboratory studies provided accurate estimates of the effect of deicers on natural ecosystems. Furthermore, if relevant concentrations of road salt were used, the effects of road salt could also alter the ecological community [13] or persist and affect later life-stages.

Artificial Light At Night (ALAN)

Another environmental stressor that alters an organism’s ecological and physiological response to the environment is Artificial Light At Night (ALAN). Over twenty percent of the earth’s surface is affected by ALAN [14]. These global ALAN emissions have increased at a rate of 6% each year over recent decades, with some areas increasing as much as 20% per year [15]. These emissions are likely affecting many species, as 30% of all vertebrates and more than 60% of invertebrates are nocturnal [16].

ALAN has brought unprecedented environmental change that is much higher than most natural changes at any time scale, extent and rate of spread, [17] yet ALAN’s effects are still relatively unknown [17]. ALAN can have both indirect and direct effects by altering species interactions, community composition, physiological health or behavior, among others [18–23]. In addition to ecological and physiological effects within a life-stage, these effects can also carry-over and affect later life-stages. These carry-over effects have been found to disrupt circadian rhythms [24]. The main problem with ALAN research is

that direct and indirect effects are often studied separately but it is likely that these two aspects are working together. Furthermore, there is only one study [24] I am aware of that investigates carry-over effects to later life-stages.

Research program and goals

Using an observational field study, indoor and outdoor experimental manipulations, and field measures, I investigated the effects of road salt runoff and

ALAN on amphibian model systems. Amphibians are declining globally; 32% of the

6,600 described amphibian species are currently threatened by extinction [25]. The

Global Amphibian Assessment of 2004 indicates that since 1980 there have been 122 probable amphibian extinctions. This is 2,700 times faster than the background extinction rate [26]. Identifying what factors are causing this rate of decline is crucial and environmental stressors have been identified as a factor in amphibian declines [27,28]. I used amphibians to answer two overarching research questions in respects to environmental stressors: 1) are the effects of stressors under- or over-estimated, and 2) do endocrine responses help us understand the physiological effect of environmental stressors?

In Chapter 2, I investigated the long-term effects of road salt exposure on the wood frog (Rana sylvatica). Using experimental and observational approaches, I found that as conductivity increased in natural wetlands, wood frog larvae were less abundant, but those found were larger. In the first field experiment of this study, I found salt significantly increased larval growth and algal biomass, and decreased the abundance of zooplankton, which may compete with tadpoles for food. In the second experiment, I 5

found larval exposure to road salt caused juvenile frogs to have greater mortality in low- density terrestrial environments.

In Chapter 3, I tested whether ALAN had greater direct or indirect effects on

American toads (Anaxyrus americanus). I found the presence of ALAN significantly reduced the duration of metamorphosis and periphyton biomass. The effects of ALAN appeared to be mediated through direct effects on toad development as no evidence of

ALAN affecting toads indirectly through the ecological community was found. Juvenile- stage ALAN reduced juvenile toad growth and altered activity such that ALAN toads did not suppress activity at night. Carry-over effects of ALAN were also present; juvenile toads that had been exposed to larval ALAN exhibited marginally increased juvenile activity.

In Chapter 4, I expanded upon our research of ALAN by adding an additional stressor, predation. I tested consumptive and non-consumptive effects of predators on toads. In addition to survival and growth, endocrine responses to ALAN and predation were measured by corticosterone. Corticosterone was chosen because it is a key part of the stress response; if for example, ALAN increases prey vulnerability to predators, corticosterone production may increase in ALAN treatments. I tested for the effects of

ALAN and predators on toad corticosterone across life-stages. Predators decreased toad survival and growth but did not affect corticosterone production until the juvenile life- stage. ALAN did not affect larval toads nor did it interact with the predator treatment to increase larval toad predation.

Chapter 2: Legacy of Road Salt: Apparent Positive Larval Effects Counteracted by

Negative Post Metamorphic Effects in Wood Frogs

Published in Environmental Toxicology and Chemistry, Vol. 34, No. 10, pp. 2417-2424, 2015 Authors: Kacey L. Dananay1, Katherine L. Krynak1, Timothy J. Krynak2, and Michael F. Benard1 1:Department of Biology, Case Western Reserve University, Cleveland, Ohio, USA 2:Natural Resource Division, Cleveland Metroparks, Parma, Ohio, USA (Submitted 10 March 2015; Returned for Revision 6 April 2015; Accepted 22 May 2015)

Abstract

Road salt runoff has potentially large effects on wetland communities, but is typically investigated in short-term laboratory trials. We investigated effects of road salt contamination on wood frogs (Rana sylvatica) by combining a field survey with two separate experiments. The field survey tested whether wood frog larval traits were associated with road salt contamination in natural wetlands. As conductivity increased, wood frog larvae were less abundant, but those found were larger. In the first experiment, we raised larvae in outdoor artificial ponds under four salt concentrations and measured larval vital rates, algal biomass and zooplankton abundance. Salt significantly increased larval growth, algal biomass, and decreased zooplankton abundance. In the second experiment, we raised larvae to metamorphosis in the presence and absence of salt contamination and followed resulting juvenile frogs in terrestrial pens at high and low densities. Exposure to road salt as larvae caused juvenile frogs to have

greater mortality in low-density terrestrial environments, possibly due to altered energy allocation, changes in behavior or reduced immune defenses. Our study suggests low concentrations of road salt can have positive effects on larval growth yet negative effects on juvenile survival. These results emphasize the importance of testing for effects of contaminants acting through food webs and across multiple life stages as well as the potential for population-level consequences in natural environments.

Introduction

Identifying mechanisms through which human activities cause species to decline is a key step in protecting biodiversity [1, 2]. Laboratory tests can identify possible causes of decline, and determine the magnitude of a stressor (e.g., parasite or contaminant) necessary to cause mortality or impair reproduction in controlled conditions

[3]; however, the results of laboratory studies cannot directly illustrate whether a specific anthropogenic disturbance is having a negative impact on a species in nature. For example, diseases or contaminants can have a large effect on a single life stage, but that life stage may not contribute strongly to population growth [4]. Similarly, contaminants can have a strong effect in the laboratory, but this effect can be weak or nonexistent in the context of a more complex ecological community or more realistic environmental conditions [1]. Further, the effects of factors causing a species to decline may not be incurred immediately, and thus may not be identified unless studies follow organisms through multiple life stages [5]. A successful approach to overcome these challenges has been to compare the effects of environmental contaminants between patterns observed in the field and laboratory, and across multiple life stages. For example, combinations of

field studies and laboratory experiments convincingly identified causes of limb deformities [6] and hermaphroditism [7] in amphibians. Experiments that link field and laboratory studies and encompass multiple life stages will be critical to estimate the long- term effect of stressors on populations, especially those in decline.

Amphibians are experiencing dramatic declines and extinctions, and thus identifying threats to amphibian populations is of great importance [8]. Chemical contaminants are one of several factors implicated in amphibian declines [8]. Testing how environmental contaminants affect amphibians under natural conditions is essential because complexities of ecological communities can alter the effect of environmental contaminants [9, 10]. Additionally, strong carry-over effects that larval conditions can have on the phenotype and fitness of post-metamorphic individuals [11-13], highlight the importance of investigating contaminants on amphibians across multiple life stages.

However, there are relatively few amphibian studies that test for post-metamorphic fitness effects resulting from exposure to contaminants in the larval stage [14, 15], despite evidence suggesting juvenile survival has a greater effect on population dynamics than egg or larval survival [4]. Thus investigating effects of a contaminant over multiple environments with increased environmental complexity will allow a comprehensive view of the effect contaminants can have in nature.

One contaminant that has been implicated as a potential threat to amphibians is road salt [16-18]. Rock salt or road salt (active ingredient: sodium chloride, NaCl) is widely used to remove snow and ice from roads, but as the snow melts, the minerals are washed off roads and can travel up to a kilometer into wetlands [19]. When amphibians

breed in these wetlands, their eggs and larvae can be exposed to the salt being washed off roads. Laboratory studies investigating direct effects of road salt on amphibians have often found road salt increases egg water loss, decreases larval activity and growth, interferes with development and increases mortality [16, 18, 20, 21]. Road salt, however, can also affect amphibian larvae indirectly, through altering the aquatic food web [22, 23].

Better understanding of the effects of road salt is needed in two key areas. First, despite substantial laboratory evidence for negative effects of road salt contamination, few field studies test for associations between road salt contamination and amphibian presence or health. The results of two existing surveys are mixed; one found a negative correlation between road salt contamination and amphibian species richness [18], while another found no significant correlation between salt concentration and amphibian abundance [24]. Second, to our knowledge no studies have yet tested whether road salt contamination experienced by larvae has negative effects that persist beyond metamorphosis. If such negative effects exist, failure to consider them could cause us to underestimate the effects of road salt contamination. To further our understanding of the impacts of road salt on amphibians, it is necessary to: 1) study effects of road salt at environmentally relevant concentrations, 2) determine mechanisms (i.e. direct vs. indirect) causing changes in survival and growth, 3) link laboratory and experimental results with field studies and 4) expand the analysis to encompass more than one life stage in order to detect any post-metamorphic effects.

We conducted one field survey and two experiments that investigated the direct and indirect effects of road salt contamination on wood frogs (Rana sylvatica). Our field survey tested whether frog presence, abundance, size, or developmental stage were associated with road salt contamination across multiple wetlands. Our first experiment tested whether increasing concentrations of road salt affected the survival, growth and development of wood frog larvae, zooplankton abundance, and relative periphyton and phytoplankton biomass. Our second experiment tested whether road salt contamination experienced during the larval stage affected post-metamorphic survival and growth of wood frogs at two different post-metamorphic densities. Together, these studies investigated how road salt contamination affects larvae in realistic environmental conditions and whether road salt effects persist across life stages.

Materials and Methods

Field Survey

We surveyed 30 wetlands in Northeast Ohio to test whether wood frog (Rana sylvatica) presence, abundance, size, or developmental stage were associated with road salt contamination. These wetlands were selected because they met the conditions suitable for wood frog populations to persist [25]: semipermanent or emphemeral and surrounded by forest. We used a timed-dipnet approach to sample wetlands between 25

May and 3 June 2011. We sampled each wetland for a set period of time based on surface area (2 to 5 person-minutes of sampling for wetlands 100 m2, 5 to 10 minutes of sampling for wetlands approximately 100 m2 to 750 m2, and 10 to 20 person-minutes of sampling for wetlands greater than 750 m2 surface area), similar to methods in Werner et 11

al. [26]. We sampled wetlands with dipnets (22 X 27 cm, with 1 X 2 mm mesh size), and recorded the number of amphibians collected. In each wetland, up to 30 larvae of each amphibian species were preserved in 95% ethanol and returned to the laboratory to measure snout-vent length and developmental stage.

We also assessed the environmental conditions of each pond in our survey. We measured conductivity, an estimate of the amount of road salt contamination [27], of each wetland using an ExStik ® II hand-held conductivity meter (Extech Instruments, Nashua,

NH). We estimated the surface area of each wetland because wetland size can affect the overall ecological community [26]. Amphibian diversity exhibits a hump-shaped relationship with wetland area, and the predator community shifts from being invertebrate-dominated to fish-dominated [26]. We measured overhead forest canopy cover (i.e., amount of leaves shading each pond) using a handheld densiometer (Forestry

Suppliers Spherical Crown Densiometer), because increasing canopy cover is associated with reduced larval growth and development [28]. We collected data on distance to nearest paved road from wetland center and amount of surrounding forest cover within 1 km of each wetland using the National Landcover Database analyzed with ArcGIS version 9.0.

Statistical Analyses

We examined four wood frog larval response variables: presence or absence, catch-per-unit effort (CPU), average snout-vent length (SVL), and average Gosner developmental stage [29]. CPU was the number of wood frog larvae counted divided by the sampling effort in person-minutes. We tested whether these four response variables 12

were associated with conductivity, distance to the nearest road, pond area, amount of forest cover within 1 km, and amount of canopy cover. We used a logistic regression to analyze the wood frog presence data, and multiple linear regressions to analyze the CPU, average SVL, and average developmental stage data. To meet assumptions of normality, we natural-log transformed conductivity and pond surface area, square-root transformed distance to the nearest road, and arcsin-transformed the fraction of terrestrial habitat within 1 km that was forest (e.g., not agriculture or suburban).

Experimental venue

We conducted two separate experiments in outdoor mesocosms located at Case

Western Reserve University’s Squire Valleevue and Valley Ridge Farms (hereafter

“University Farm”) in Hunting Valley, Ohio. We filled mesocosms (2 m diameter polyethylene cattle tanks) with approximately 700 liters of pond water on 29 March 2013.

Mesocosms were placed into five spatial blocks. Each block contained one replicate of each treatment Large shade canopies covered each block of mesocosms, to prevent rain from overfilling the mesocosms and to mimic the shading present over many wetlands inhabited by wood frogs. To prevent the colonization of insects or treefrogs, we covered each mesocosm with a tightly fitting lid of 60% shade cloth. We added approximately 94 g of air-dried oak and maple leaf litter to each mesocosm on 01 April 2013. We added road salt (information on the amount of road salt described below for each experiment) to the mesocosms on 15 April 2013. We chose road salt manufactured by Cargill Salt

(95.8-99.8% sodium chloride, 0. 005-0.01% sodium ferrocyanide decahydrate) because it is widely used on Ohio roads

(https://www.dot.state.oh.us/news/Documents/ODOTFY2013SaltContractWinterUsePrici ng.pdf). On 01 April 2013 we collected five wood frog egg masses from a pond at the

University Farm. We placed the eggs in individual containers kept outdoors. Eggs began hatching on 15 April 2013, and hatchlings were distributed into treatments 4-5 days afterward.

Experiment 1: Salt concentration

We tested whether four concentrations of road salt (0, 100, 500, and 900 mg/L of

Cargill Salt) affected larval wood frog vital rates (survival, growth and development), zooplankton abundance, and relative periphyton and phytoplankton biomass. We selected these road salt concentrations to be consistent with commonly observed field concentrations (0.39-1030 mg/L NaCl) [16, 18]. We replicated each road salt concentration five times (for a total of 20 experimental units). On 20 April 2013, we added 70 larvae to each mesocosm; each group of larvae was composed of seventeen or eighteen individuals from four of the five egg masses. We checked mesocosms every 1 to 3 days until the experiment ended. On 28 May 2013 (38 days after the experiment started), we ended the experiment before the stress of metamorphosis occurred by euthanizing all larvae with Tricaine methanesulfonate (MS-222) and preserving them in

70% ethanol.

To relate salt concentrations in our experiment to field conductivity measurements, we measured conductivity on 18 April, 08 May and 21 May 2013 using an Oakton pH/CON 10 Series probe (Oakton Instruments, Vernon Hills, IL, USA). We measured dissolved oxygen (DO) and pH starting at 2 pm on 21 May 2013 using an 14

Oakton DO 300 Series probe and the pH/CON 10 Series probe (Oakton Instruments,

Vernon Hills, IL) at a depth of approximately 25 cm

To assess the effects of road salt concentration on the relative biomass of periphyton and phytoplankton, we measured chlorophyll a in the water column and from periphyton suspensions removed from leaves on 22 May 2013. We measured chlorophyll a using an Aquafluor fluorometer (Turner Designs, Sunnyvale, CA, USA). This determines differences in relative biomass of periphyton and phytoplankton in each treatment, but not absolute biomass. To assess relative phytoplankton biomass in the water column, we collected 2 mL of water near the surface of each mesocosm at four different locations, measured fluorescence and averaged all four water chlorophyll a measures. To assess relative periphyton biomass growing on the leaf litter, we took two leaf punches (6 mm diameter) from each of two oak leaves haphazardly selected from opposite sides of the mesocosm. We placed punches from each leaf into a separate vial and agitated the leaf punches in 2 mL of distilled water for 30 seconds. To ensure our measures were not too opaque for the fluorometer, we diluted our samples by taking one mL from the middle of the cryovial and added it to 1 mL of distilled water. We used mean chlorophyll a measures from each mesocosm for statistical analyses.

To quantify zooplankton abundance, we took four 1-liter subsamples of water from four locations in the mesocosm. We collected each subsample approximately 5 cm below the water surface, and then combined subsamples by straining them through 63- micron Nitex® mesh. We preserved samples in 50 mL of 95% ethanol. We counted the number of zooplankton observed in five subsamples of each sample using a Sedgewick-

Rafter counting cell. We separated zooplankton into two groups: rotifers and arthropods

(i.e., cladocera and copepods).

Statistical Analyses

We tested for effects of salt concentration on most response variables using two multivariate regressions. In the first multivariate regression, we used square-root transformed mean Gosner stage, mean mass and survival as response variables. In the second multivariate regression, we used arthropod abundance, relative periphyton biomass, and relative phytoplankton biomass as our response variables. Salt concentration was used as a continuous explanatory variable in both multivariate regressions. When a multivariate regression was significant, we examined the standardized canonical coefficients to evaluate which variables were contributing to the effect. We analyzed rotifer abundance using a nonparametric correlation test

(Spearman’s rho) because the data could not be transformed to normality (i.e., two treatments had no rotifers in many replicates). We dropped block from the analyses when there was no significant effect. All analyses were done using R [30].

Experiment 2: Carry-over effects

To test whether road salt contamination affected post-metamorphic growth and survival, we manipulated road salt concentration in the larval stage, and assessed post- metamorphic growth and survival in terrestrial pens at two densities. We raised larvae in two treatments: no road salt added and 500 mg/L of Cargill road salt added. On 19 April

2013, 70 larvae were placed in each of ten mesocosms. Each group of 70 larvae was

composed of 14 individuals from each of five egg masses. As larvae approached metamorphosis, floating islands of screen, foam and plastic boxes containing leaves were placed in each mesocosm to allow metamorphs to climb out of the water. Once metamorphs had reabsorbed their tails, they were weighed and immediately placed into a randomly assigned terrestrial pen. All seven metamorphs from each mesocosm were placed in their respective pens between 13 June and 23 June 2013. Terrestrial pens had two treatments: high and low frog density. We placed two metamorphs from each mesocosm into the low-density pen and five metamorphs from the same mesocosm into the high-density pen (total of 20 terrestrial pens; 5 pens/treatment). After the first seven metamorphs from each mesocosm emerged, we euthanized the remaining metamorphs using MS-222.

We constructed 1 m2 terrestrial pens in the forest adjacent to the pond from which all egg masses were collected using 50 cm aluminum flashing stapled to wooden posts and buried approximately 12 cm below the surface. We used aluminum screening as a baffle (i.e., a barrier at a 90° angle to the side of the pen) along the top of the pens. After the pens were constructed, we removed all the leaves inside the pens and replaced them with fresh leaves from a 1 m2 area outside the pen. We did not place any additional food sources in the pens. We checked the pens at least three times each week. The insides of the pens were undisturbed unless no rain had fallen over five days at which time we sprayed each pen with 1L of water using a garden sprayer.

On 22 October 2013, we destructively sampled the first 10 pens. On 24 October

2013, we destructively sampled the remaining 10 pens. We checked all pens daily for

any missed frogs until 6 November 2013. We weighed individuals on the day of collection and immediately euthanized them using MS-222.

Statistical Analyses

We used ANOVA to test whether larval road salt concentration affected average mass at metamorphosis and larval duration using the mean value of the first seven individuals emerging from each mesocosm. We tested whether terrestrial survival and final mass were affected by larval road salt concentration and terrestrial density using a linear mixed-effect model. Models for terrestrial juvenile survival and final mass involved terrestrial density, larval road salt concentration and their interaction as explanatory variables. Mesocosm was used as a random variable. All analyses were completed using R [30].

Results

Field Survey

The distance from each wetland to the nearest road ranged from 5 to 620 m, with a median distance of 170 m. Conductivity of each wetland was negatively correlated with distance to the nearest road (Pearson correlation coefficient, r(28) = -0.40, p <

0.028), consistent with the expectation that road salt contamination is greatest near roads.

Conductivity of the ponds ranged from 30 to 430 µS. Logistic regression indicated that wood frogs were less likely to be found in wetlands with high conductivity

(β = -2.36 ± 1.19, Z = -1.98, p = 0.05). However, wood frog presence was not associated with pond area (β = -1.47 ± 0.89, Z = 1.64 p = 0.10), forest area (β = 0.38 ± 2.67, Z = 18

0.14, p = 0.89), canopy cover (β = 6.1*1020 ± 3.7*1020, Z = 1.63, p = 0.10), or distance to roads (β = -0.21 ± 0.15, Z = -1.38, p =0.17).

For ponds in which wood frog larvae were collected, as conductivity increased, catch-per-unit effort (CPU) declined (Table 1, Fig. 1). Larval SVL and Gosner stage, however, were positively associated with conductivity (Table 1, Fig. 1). As pond area increased, CPU declined, but there was no effect on larval SVL or Gosner stage (Table 1,

Fig. 1). Neither the amount of forest within 1 km of each wetland, nor the amount of tree canopy cover over each wetland was associated with CPU or larval SVL. However, both the amount of surrounding forest and the amount of tree canopy cover over each wetland were negatively associated with Gosner stage (Table 1). The distance to the nearest road were not associated with any of the response variables (Table 1).

Experiment 1: Salt concentration

The four road salt treatments had mean conductivities of 75.2 µS + 1.5 standard error (SE), 275.2 µS + 5.4 SE, 997.6 µS + 43.6 SE, and 1720.6 µS + 66.2 SE, for the control, 100 mg/L, 500 mg/L and 900 mg/L concentrations respectively. Salt concentration and conductivity were highly correlated on all three dates (18 April

(Pearson correlation coefficient = 0.99, t18 = 34.5, p < 0.0001), 08 May (Pearson correlation coefficient = 0.98, t18 = 22.73, p < 0.0001) and 21 May (Pearson correlation coefficient = 0.98, t18 = 22.12, p < 0.0001).

pH ranged from 6.72 to 7.12 across all treatments and averaged 7.02 + 0.05 SE, 7

+ 0.06 SE, 6.8 + 0.05 SE, and 6.8 + 0.01 SE for control, 100, 500, and 900 mg/L NaCl,

respectively. DO ranged from 28.9 to 60.9 % saturation across all treatments and averaged 44.3 % + 4.3 SE, 50.46 % + 3.6 SE, 49 % + 5.4 SE, and 48.1 % + 5.01 SE for control, 100, 500, and 900 mg/L NaCl respectively. Salt concentration was negatively correlated with pH (Pearson correlation coefficient = -0.69, t18 = -3.99, p = 0.0009), but was not correlated with DO (Pearson correlation coefficient = 0.07, t18 = 0.29, p = 0.78).

Mean larval survival across all replicates was 88.9 % (± 4.5% SE). Multivariate regression indicated a significant multivariate effect of salt concentration on larval response variables (Pillai = 0.48, F3,16 = 5.07, p = 0.01). The standardized canonical coefficients were 1.23 for Gosner stage, -1.15 for mass, and -0.22 for survival. The univariate regression indicated no significant relationship between salt concentration and

-5 -5 Gosner stage (β = -0.052 ± 0.03, t18= -1.94, p = 0.07), mass (β = 5.2 X 10 ± 3.7 X10 ,

-6 -5 t18= 1.43, p = 0.17), or survival (β = 6.2X10 ± 1.3 X10 , t16= 0.48, p = 0.64). The lack of significant univariate effects may have occurred because of a multivariate effect of salt concentration acting simultaneously on multiple response variables [31]. The high opposing coefficients of Gosner stage and mass on the canonical variables suggested the response might occur through those variables. Thus, we conducted an additional univariate analysis on the ratio of mass divided by the square root of Gosner stage, and found a significant positive correlation with salt concentration (β = 6.8 X 10-8 ± 2.3 X10-

8 , t18=2.6, p = 0.017). Taken together these results indicate that higher salt concentration caused larvae to be larger but less developed than larvae in lower salt concentrations (Fig.

2).

Multivariate regression also indicated a significant effect of salt concentration on the ecological community response variables (Pillai = 0.49, F3,16 = 5.06, p = 0.01). The standardized canonical coefficients were 0.98 for arthropod abundance, -0.40 for relative periphyton biomass, and -0.68 for relative phytoplankton biomass. The univariate regression indicated a significant negative relationship between arthropod abundance and salt concentration (β = -70 ± 0.26, t18 = -2.7, p = 0.02), and a significant positive relationship between relative periphyton biomass and salt concentration (β = 1.9 X 10-3 ±

-4 8.9 X 10 , t14 = -2.7, p = 0.02). However, there was not a significant relationship between relative phytoplankton biomass and salt concentration (β = 8.0 X 10-4 ± 7.8 X

-4 10 , t18 = 1.0, p = 0.32). The nonparametric correlation indicated that road salt concentration negatively affected rotifer abundance (Spearman’s ρ = -0.79, p < 0.001).

Taken together, these results indicated increased salt concentration caused reduced zooplankton (arthropod and rotifer) abundance, but increased periphyton biomass (Fig.

3).

Experiment 2: Carry-over effects

Road salt concentration positively affected average mass of the first seven metamorphosing frogs that were later placed in the outdoor pens (F1,8 = 7.947, p = 0.023).

Mean mass for newly metamorphosed frogs raised in mesocosms without additional salt was 0.17 g + 0.01 SE, whereas mean mass was 0.21 g + 0.01 SE for frogs raised in 500 mg/L added road salt. Road salt concentration did not affect days to metamorphosis (F1,8

= 1.917, p = 0.2; absence of road salt 59.8 days + 1.04 SE; road salt treatment 57.7 days +

1.1 SE). Terrestrial survivorship of juvenile wood frogs was significantly affected by

larval treatment (F1,8 = 5.64, p = 0.045) and the interaction between larval and terrestrial treatments (F1,8 = 7.37, p = 0.027, Fig. 4A). At high terrestrial densities, juvenile frogs had similar survival regardless of larval salt treatment, but at low terrestrial densities, juvenile frogs exposed to salt as larvae had lower survival than those not exposed to salt.

When the terrestrial portion of the experiment was ended in October, juvenile frogs in the high density treatments were smaller than those emerging from low density treatments

(F1,7 = 21.53, p = 0.002; Fig. 4B), but juvenile mass was apparently not affected by larval road salt exposure (F1,8 = 0.14, p = 0.72; Fig. 4B).

Discussion

In our field survey, we found evidence for seemingly contradictory effects of salt exposure on wood frog traits. While wood frogs were less likely to be found, or were present in lower abundance in wetlands with higher conductivities, they had an overall greater mass in wetlands with higher conductivities. Reduced abundance is consistent with another field survey [18] and predictions from laboratory studies and demographic models that suggest road salt contamination has negative effects on amphibian populations [24]. The positive correlation between mass and conductivity observed in our survey is also consistent with previous work showing increased growth of anuran larvae exposed to salt in outdoor mesocosm experiments [22, 32]. Increased size could indicate a positive effect of road salt on anuran populations, because larger size at metamorphosis is often associated with increased individual fitness and population growth in amphibians [33, 34]. Several mechanisms could potentially account for increased growth of wood frog larvae in wetlands with higher conductivity. Amphibian

larval growth is often density-dependent [35], and thus increased conspecific mortality could lead to increased larval growth through reduced competition. Alternatively, salt contamination can reduce the population size of heterospecific competitors (e.g., snails, insect larvae) that compete with wood frogs, promoting periphyton resources fed on by wood frogs [22, 32]. Our experiments provide insight into the mechanisms that can generate the pattern we observed in the field.

In the salt concentration experiment, we found that increasing salt concentration was associated with greater growth of wood frog larvae and higher relative periphyton biomass, but lower zooplankton abundance. There was no effect of salt on relative phytoplankton biomass. Our results suggest indirect interactions are playing an important role in the patterns we observed in the field study. This conclusion is further supported by other studies that demonstrate the indirect effects (e.g., reduced competition for algal resources due to salt killing invertebrates) of salt on amphibian larvae [22, 23, 32]. The nature of the indirect effects of road salt contamination, however, can vary widely between amphibian clades, depending on the ecological characteristics of those clades.

For example, Ambystoma salamander larvae are indirectly negatively affected by salt contamination because their main food resource, zooplankton, is reduced by salt contamination [23]. In contrast, gray treefrog (Hyla versicolor) and wood frog larvae exhibited greater growth when exposed to salt contamination, likely due to an increase in the periphyton that they feed on [22, 32]. On its own, the increased growth of larval anurans exposed to low (500 mg/L in our study) salt concentrations suggests a potential beneficial effect of salt contamination. This conclusion, however, does not take into account potential effects of salt concentration extending across life stages. 23

Our second experiment demonstrated the importance of testing for carry-over effects of contaminants across life stages. Various environmental factors induce carry- over effects in many taxa including invertebrates, birds, mammals and amphibians [36,

37]. For amphibians, increased size at metamorphosis is often associated with increased growth and survival in later life stages [12, 33, 34]. Despite the increase in mass associated with being raised in road salt, juvenile frogs exposed to salt as larvae actually had reduced survival in low terrestrial densities compared to juvenile frogs not exposed to salt as larvae. An alternative hypothesis is that larger frogs from the salt treatment were more likely to escape. We think the escape hypothesis is unlikely as we never found trespassing wood frogs or toads in the pens, despite high densities of these anurans around the pens. The lack of trespass into pens suggests that our pens were secure. Thus, we believe larval environment reduced post-metamorphic survival demonstrating the importance of investigating the effect of contaminants over multiple life stages.

While we cannot determine the cause of increased mortality in the terrestrial environment, there are several possibilities. For example, from the perspective of

Dynamic Energy Budget Theory, it is possible that larval-stage salt exposure altered how juvenile frogs allocate energy to growth and metabolism in terrestrial environments that vary in per-capita food availability [38], Indeed, environmental variation during the larval stage can carry-over beyond metamorphosis to affect juvenile amphibian metabolism [39], and high densities of post-metamorphic amphibians experience reduced growth and survival, likely due to reduced food resources [40]. An alternative hypothesis is that juvenile immune function was impaired by both larval salt exposure and high conspecific density. Under this hypothesis, the frogs from the no-salt exposure, low 24

density treatment would have had stronger immune defenses and consequently higher survival than the frogs from the other three treatments. Consistent with this hypothesis is the observation of reduced immune defenses in post-metamorphic salamanders exposed to contaminants as larvae [5]. Other studies, however, found no evidence for carry-over effects after larval exposure to atrazine, carbaryl or malathion in the terrestrial life stages of amphibians [15, 41]. Finally, it is worth noting that carry-over effects of contaminants like road salt could occur even if the contaminant does not persist in the tissues of the animals. For example, larval-stage temperature and food can affect post-metamorphic metabolism [39], and larval-stage predation risk can affect post-metamorphic toxin production [13]. It is clear carry-over effects exist and affect individuals, yet we still need data to demonstrate how carry-over effects influence populations.

In addition to affecting individual fitness, the carry-over effect of road salt identified here has the potential to affect population dynamics. While it is logistically challenging to assess contaminant effects across multiple generations in nature, demographic models present an important tool to evaluate population-level effects [4].

Karraker et al. [24] presented the first demographic model that we are aware of to investigate the effects of road salt contamination on wood frogs and spotted salamanders.

They found that road salt’s impact on amphibian populations declined across models as the strength of density-dependence increased. In their model, the larval vital rates were affected by salt exposure and larval density, but the post-metamorphic vital rates were independent of larval salt exposure and post-metamorphic density. Our results (i.e., different effects of salt incurred at different densities) suggest that their model could be expanded upon by varying post-metamorphic vital rates such that they are dependent 25

upon larval salt exposure and post-metamorphic density; this can provide qualitatively different predictions from models that do not include carry-over effects and multistage density dependence. Our present study, however, only investigated the effect of carry- over effects at one salt concentration and two post-metamorphic densities. Robust incorporation of carry-over effects into a demographic model requires a larger experiment to parameterize how vital rates change across a range of larval salt exposures and post-metamorphic densities.

Previous laboratory studies have largely found that salt exposure reduced larval survival, activity, and mass, and caused physical abnormalities [16, 18] whereas field studies like ours have found some positive effects on larval growth [22, 32]. We suggest three possible reasons for the discrepancy among studies: type of salt, salt concentration and experimental venue. First, table salt, pickling salt, and various types of road salt have been previously used to test for effects of salt [16-18]. Different additives in salt could change the effect of salt on amphibians, even though these additives are found in very low concentrations (less than 1%) compared to the salt itself. Second, the majority of the research on road salt has found it to have negative effects on amphibian larvae at concentrations above 1000 mg/L NaCl, whereas we used concentrations ranging from

100 – 900 mg/L NaCl. Indeed, many studies finding direct, negative effects of salt have used concentrations much higher than ours, ranging from 300 mg/L to 26,000 mg/L NaCl

[16-18]. The indirect positive effect on larval growth found in our study would likely be reversed at higher concentrations [32]. Third, laboratory studies drastically reduce the complexity found in nature thereby eliminating indirect effects that likely occur in natural communities. Overall, our study demonstrates road salt can affect wood frog larvae 26

indirectly and carry-over to later life stages when studies are performed in more natural environments at realistic salt concentrations.

Our studies demonstrate that some concentrations of road salt can indirectly benefit larvae but can reduce juvenile survival and therefore adult abundance. Future studies that include additional stressors (e.g. predators or competitors) are necessary to test whether the apparent benefits of salt in the larval stage persist in more complex environments including those with additional stressors. Because carry-over effects are rarely investigated for many contaminants, future studies need to test whether other factors that also cause carry-over effects (e.g. predators) create additive effects with contaminants. Studies investigating multiple stressors will help test the significance of carry-over effects. It is also possible that competitors will reduce the effects of environmental contaminants [41]. Increasing the complexity of experimental mesocosms will enhance our ability to compare field and experimental results. Together this suggests that prolonging experiments to encompass multiple life stages and increasing environmental complexity will likely lead to more accurate estimations of the effect of contaminants in nature.

Tables

Table 1: Multiple regression analysis of pond survey data. Larval catch-per-unit effort

(ln[CPU]), snout-vent length (ln[mm]) and developmental stage (ln[Gosner]) were regressed onto conductivity (ln[µS]), pond area (ln[m2]), forested terrestrial area

(arcsin[m2]), distance to road (square-root [m]), and canopy cover (% cover) over ponds.

The first column for each 27

response variable is the regression coefficient, with standard error listed in parentheses. The coefficient of partial determination

(“partial r2”) is also provided for each explanatory variable.

CPU SVL (mm) Gosner stage

Coef t P partial r2 Coef t P partial r2 Coef t P partial r2

Conductivity -0.96 -2.66 0.017 0.31 0.06 2.27 0.038 0.24 0.042 2.65 0.018 0.30 (µS) (0.36) (0.03) (0.016)

Pond area (m2) -0.68 -2.15 0.047 0.23 1.5*10-4 0.01 0.995 2.5*10-6 -0.010 -0.76 0.460 0.04 (2.4*10-2) (0.31) (0.014)

Forest area (m2) -0.91 -0.63 0.439 0.02 -7.1*10-4 -0.01 0.995 2.6*10-6 -0.141 -2.24 0.040 0.24 (1.1*10-1) (1.45) (0.063)

Canopy cover 1.8*1021 -0.14 0.890 0.09 -5.8*10-22 -0.58 0.569 1.3*10-3 -1.4*10-21 -2.46 0.026 0.03

(1.3*1020) (9.9*10-22) (5.7*10-22)

Road distance -0.006 -1.27 0.222 1.0*10-3 5.5*10-4 0.14 0.887 0.02 -0.002 -0.70 0.497 0.27 (m) (3.8*10-3) (0.005) (0.002)

Figures

Figure 1: Effects of conductivity (µS) on catch per person-minute, snout-vent length

(SVL; mm) and developmental Gosner stage (Gosner). As conductivity increased, (A) 30

fewer wood frog larvae were collected for a given sampling effort, (B) larvae were on average larger, and (C) larvae were on average more developed. All points represent ponds in Northeast Ohio in which wood frog larvae were present. Pond conductivity

(SQRT; µS) was square-root transformed. Catch per person-minute, snout-vent length, and Gosner stage were natural log transformed.

Figure 2: Effects of salt concentration (mg/L) on mean wood frog Gosner stage (square- root[Gosner]) and larval mass (g). Higher salt concentrations caused larvae to be larger but less developed than larvae in lower salt concentrations. Error bars indicate standard errors.

Figure 3: Effects of salt concentration (mg/L) on relative periphyton biomass

(fluorescence, relative fluorescence units) and arthropod abundance (arthropods/L).

Increased salt concentration caused reduced arthropod abundance, but increased periphyton biomass. Error bars indicate standard errors.

Figure 4: Carry-over effects of larval salt exposure on juvenile wood frogs. (A)

Terrestrial survivorship of juvenile wood frogs was significantly affected by the interaction between larval and terrestrial treatments. Regardless of larval salt treatment, high terrestrial densities had low survival. Juvenile frogs exposed to salt as larvae had lower survival than those not exposed to salt as larvae in low density terrestrial

treatments. (B) Juvenile mass (g) was apparently not affected by road salt, but high density terrestrial treatments significantly reduced juvenile frogs mass.

Chapter 3: Artificial Light at Night decreases metamorphic duration and increases

juvenile growth in a widespread amphibian

In Revision at Proceedings of the Royal Society B: Biological Sciences Authors: Kacey L. Dananay1 and Michael F. Benard1 1: Department of Biology, Case Western Reserve University, Cleveland, Ohio, USA

Abstract:

Artificial Light at Night (ALAN) affects over 20% of the earth’s surface and is estimated to increase 6% per year. Most studies of ALAN have focused on a single mechanism or life-stage. We tested for indirect and direct ALAN effects that occurred by altering American toads’ (Anaxyrus americanus) ecological interactions or by altering toad development and growth, respectively. We conducted an experiment over two life- stages using outdoor mesocosms and indoor terraria. In the first phase, the presence of

ALAN reduced metamorphic duration and periphyton biomass. The effects of ALAN appeared to be mediated through direct effects on toad development, and we found no evidence for indirect effects of ALAN acting through altered ecological interactions or colonization. In the second phase, post-metamorphic toad growth was reduced by 15% in the ALAN treatment. Juvenile-stage ALAN also affected toad activity: in natural light, toads retreated into leaf litter at night whereas ALAN toads did not change behaviour.

Carry-over effects of ALAN were also present; juvenile toads that had been exposed to larval ALAN exhibited marginally increased activity. In this time frame and system, our

experiments suggested ALAN’s effects act primarily through direct effects, rather than indirect effects, and can persist across life-stages.

Key words: ecological light pollution, night lighting, direct and indirect effects, legacy effects, American toad, Bufo americanus

Introduction

Artificial light at night (ALAN) is one of the most significant human-induced environmental changes [1,2], yet many of its effects are still poorly understood [3].

Determining ALAN’s effects on organisms’ growth and survival is complicated by the fact that these effects emerge from both how ALAN changes an organism’s interactions with other species (“indirect effects”) and how ALAN alters an individual’s physiology

(“direct effects”) [4]. Further adding to the challenge of studying ALAN is the fact that it may have different effects at different life-stages.

ALAN may indirectly affect individuals through two distinct ecological mechanisms: changing dispersal into and out of an ecological community or altering the outcome of interspecific interactions within a community [4]. Examples of ALAN altering community composition includes changes in the colonization behavior of marine invertebrates [5] or inhibition of drifting behaviour in freshwater invertebrates [6].

Regardless if ALAN changes community composition, ALAN can also indirectly affect individuals by altering the outcome of species interactions [4]. For example, nocturnal illumination similar to that of a full moon facilitates owls’ ability to hunt mice [7]. As a result, nocturnal mice exposed to ALAN may alter the time when they forage, and

consequently increase competition with diurnal congeners [8]. ALAN can also cause asymmetric interspecific competition, such that competition is stronger in lit habitats (e.g. urban) compared to darker habitats (e.g. forest) [9].

ALAN also directly affects individuals by altering their physiology [10–12]. For example, exposure to ALAN can alter the amount or timing of feeding [13,14], accelerate reproductive organ development [15] and slow larval development [16,17]. Altered behaviour has been one of the most studied direct effects of ALAN, and includes improper orientation [18,19], and altered anti-predator behaviours [20,21] or activity periods [22]. Although there are many examples of direct effects of ALAN, the majority of studies are done in the laboratory which limits our ability to relate ALAN’s effects to survival and growth in natural populations [3].

Further limiting our inferences is the fact that ALAN studies typically focus on a single life-stage. Early exposure to a wide range of anthropogenic and natural stressors

[23–26] can affect individuals throughout their lives. To our knowledge, only one study has tested for carry-over effects of early exposure to ALAN. Sixty seven percent of newborn mice exposed to constant light had disrupted circadian rhythms [27].

Subsequent exposure to normal photoperiods, however, restored regular circadian rhythms with 3-5 months [27] demonstrating the potential to reverse carry-over effects.

Carry-over effects of ALAN may be particularly important for wild animals that migrate or utilize different environments at different life-stages. These animals may be carrying the effects of early ALAN exposure, even if their subsequent habitats do not experience

ALAN. As a consequence, studies that are constrained to a single life-stage may under- or over-estimate the effects of ALAN.

Due to the many mechanisms and life-stages through which ALAN may affect an organism, it is important to assess these multiple mechanisms and stages. Here, we describe a study in which we investigated direct and indirect effects of ALAN across two life-stages using a widespread amphibian. Amphibians may be vulnerable to ALAN for three main reasons: (1) amphibians use natural light as a cue for behaviours across both larval and post-metamorphic life-stages (e.g. activity [28]), (2) ALAN may indirectly affect amphibians by influencing other community members that have strong interactions with amphibians, and (3) ALAN may have different effects in each stage, and these effects may carry-over.

Materials and Methods

Study System

We used the American toad (Anaxyrus americanus) because they are geographically widespread and a habitat generalist. They breed in water bodies ranging from permanent lakes to roadside ditches [29], including environments that are exposed to ALAN. Larval toads feed on periphyton, and the growth of periphyton can be affected by ALAN [30]. Larval toads are also preyed upon by a wide range of aquatic predatory invertebrates [29], and thus if ALAN changes the colonization rate of aquatic invertebrates, it can change predation risk for larval toads.

Experiment 1: Testing for direct and indirect effects of ALAN in larvae 38

We conducted an outdoor experiment in 2-m diameter mesocosms to investigate indirect and direct effects of artificial light on American toad larval growth, development and survival. We used a 2 X 2 X 2 experimental design in which we crossed the presence or absence of ALAN with tightly fitting lids (to control colonization) and toad larvae

(Fig. S1, S2). Mesocosms were placed into 10 groups of four mesocosms (40 total mesocosms split into 10 groups). Five groups of four mesocosms (20 mesocosms) were assigned to the ALAN light treatment, and the other five groups were assigned to the natural light treatment. Within each group of mesocosms in the light treatments, half of the mesocosms (2 mesocosms per group) were assigned to a limited-colonization treatment (i.e., lids made of shade cloth are present to reduce colonization) and the other half were assigned to be a free-colonization treatment (i.e., the other two mesocosms did not receive shade cloth lids). Finally, toad presence and absence was manipulated such that half of the mesocosms in each lid group were stocked with 50 larvae (2 mesocosms per group, 1 of each colonization treatment), and the other half of the mesocosms in each lid group did not receive toad larvae. Thus, all eight combinations of light, colonization and toad presence were initially replicated 5 times. Early in the experiment, we removed one group of four mesocosms in the ALAN treatment because heavy shading and leaf input from a nearby tree caused temperature and dissolved oxygen to be very different from the other mesocosms.

We used a single 20-Watt LED Outdoor Flood Light (ZITRADES, China) mounted in the middle of each ALAN mesocosm group approximately 1.5 meters above ground and 0.75 meters above the top of the mesocosms (Fig. S2). These flood lights

measured 835 lux directly underneath with peak wavelengths in the blue-green spectrums

(Fig. S4). Average light levels reaching the water surface was 15.07 lux + 7.42 SE without lids and 3.12 lux + 1.84 SE with lids. Our light intensities were within the range

(4.14-349.2 lux) of ALAN measured near wetlands around greater Cleveland, Ohio (Fig.

S3). We chose LED lights because their use in outdoor lighting is becoming increasingly widespread and they have the potential to become the predominant outdoor light source

[31]. All ALAN flood lights were connected to a timer set to turn on at 8 pm and turn off at 7 am. Data loggers were placed approximately 10 cm below the surface in each mesocosm to monitor temperature hourly from 29 May to 2 July 2015.

All mesocosms (2-m diameter polyethelene cattle tanks) were filled with approximately 700 liters (L) of tap water between 14 and 15 April 2015. Between 17

April and 9 May 2015, we added approximately 11 L of mixed hardwood leaves, 3.8 L of pond water, 1.45 L of pond mud, 0.38 L of zooplankton, 1 water scorpion (Nepidae), 5 snails (Planorbidae), and 8 dragonfly larvae (Libellulidae) into each mesocosm. These organisms and resources were added to make our mesocosms as realistic as possible, thus pond mud, leaves and water were not sterilised before adding it to our mesocosms. We did sift through the mud and water to ensure no large predatory larvae were added to any mesocosm. However, it is likely that other small invertebrates found in the pond mud or water were added to our mesocosms. On 7 May 2015, we collected toad egg strings from five areas of one pond at the Case Western Reserve University’s Squire Valleevue and

Valley Ridge Farm in Hunting Valley, Ohio. By selecting different areas of the pond to collect egg strings from, we expect this resulted in collecting at least five maternal lines

of toads. On 8 May 2015, 50 toad larvae were added to each of the 20 toad treatment mesocosms. These 50 toad larvae consisted of 10 toads from each of the 5 collection areas to increase genetic diversity. On 8 May 2015, we also began the colonization- limited treatment by adding the tightly fitting lids onto the colonization-limited mesocosms. Lids were constructed of 60% shade cloth and were weighed down by sand filled tubing around the edges. All lids were at about 2.5 m in diameter and completely covered the mesocosm. ALAN lights were maintained on the timer from 11 May 2015 until all toads metamorphosed.

To test for treatment effects on toad growth, development and survival, we counted and weighed all metamorphosing toads. Once metamorphs completely resorbed their tails, the first eight metamorphs in each free-colonization mesocosm were transported to the laboratory and used in the second experiment (described below). We weighed, euthanized using Tricaine methanesulfonate (MS-222), and preserved all other metamorphs in 70% ethanol as they emerged.

To test for treatment effects on the relative abundance of periphyton (main food resource for toad larvae) and phytoplankton, we measured relative chlorophyll a using an

Aquafluor fluorometer (Turner Designs, Sunnyvale, CA) [32] following methods found in Dananay et al. 2015. We used mean leaf fluorescence values for each mesocosm in statistical analyses. We measured relative phytoplankton and periphyton biomass on 10

June and 2 July 2015.

To test for treatment effects on invertebrate and amphibian colonization, we performed pipe sampling using plastic garbage cans with the bottoms removed (25 cm 41

long x 16 cm wide). The authors stood on opposite sides of the mesocosm and each pushed the can down into the water column until it hit the bottom. We swept dipnets from the bottom to the top and collected all invertebrates and amphibians until we had 10 empty nets in a row. All invertebrates (except for Nepidae) were immediately preserved in 70% ethanol, and amphibians were euthanized using MS-222 and then preserved in

70% ethanol. Pipe samples were performed on 3 June and 19 June 2015 but combined for statistical analyses.

Statistical Analyses

To distinguish between direct and indirect effects, we tested for effects of the

ALAN and colonization treatments on mean toad survival, mass at metamorphosis, larval duration and metamorphic duration using a general linear mixed model with a random effect of mesocosm nested within light treatment to account for the groups of four mesocosms. Our experimental unit was the mesocosm. Of particular interest were interactions between the colonization and ALAN treatments, because such an interaction would indicate an indirect effect of ALAN acting through colonization. Survival was calculated as the number of metamorphosed toads divided by the initial stocking density, then cube transformed to meet assumptions of normality. Mean mass at metamorphosis was calculated as the average mass of all metamorphosed toads from each mesocosm.

Larval duration was calculated as the average number of days from hatch (8 May) to metamorphosis. Metamorphic duration was the number of days between the first and the last toad to metamorphosis in each mesocosm. Metamorphic duration was reciprocal transformed (1/metamorphic duration) to meet the assumption of normality.

We tested for treatment effects on relative algal biomass using repeated measures

ANOVA with a random effect of light treatment nested within group. For relative periphyton and phytoplankton biomass, we used mean relative biomass from both sampling days as response variables. No interactions between main effects were significant and therefore we eliminated them from our analyses.

We tested for treatment effects on invertebrate abundance and diversity and larval amphibian abundance. Within each mesocosm, pipe samples from both sampling days were combined. All invertebrates were identified to order and categorized as whether or not they were a predator of tadpoles. We tested for treatment effects on total invertebrate abundance, diversity of invertebrate taxonomic orders, and log-transformed invertebrate predator abundance using ANOVA. We also tested for treatment effects on the abundance of non-predatory invertebrates using a general linear model with a poisson distribution. Aside from the intentionally stocked toads, we also found tadpoles from the treefrog family Hylidae, which colonized the mesocosms. We pooled hylid counts from both sampling days and tested for treatment effects on total hylid abundance. To account for the grouping of mesocosms, we used a general linear mixed model with a random effect of light treatment nested within group with a poisson distribution. We performed all analyses in R using packages: car, Hmisc, and MASS [33].

Experiment 2: Prior and subsequent effects of light on post-metamorphic toads

Using toads from Experiment 1, we collected the first eight metamorphosing toads from the free colonization treatments (5 natural and 4 ALAN mesocosms) and transported them back to the laboratory (Fig. S1) to test whether ALAN affected post- 43

metamorphic growth and survival. All toads for this experiment metamorphosed by 6

June 2015. While in the laboratory, toads were randomly assigned to either the natural or

ALAN light treatment, for a total of 18 experimental units (9 per juvenile light treatment), each holding 4 toads.

Toads were kept in plastic terraria (42.5cm X 30.2cm X 17.8cm) with lids. We placed terraria at a slant; the upper portion of the terraria contained dried leaves and the lower portion held 0.25 L of dechlorinated tap water. All juvenile toads were housed in the same room at approximately 23°C with the room lights on a 14:10 light:dark photoperiod. ALAN treatment lights used in the laboratory were 24-Watt LED string lights (LE, China; 61 lux at the top of the terraria) placed 50.8 cm above the terraria. The average of four measurements at the cardinal directions taken at the terraria surface was

16.9 + 3.04 lux. Similar to the outdoor LED lights used previously, laboratory LED lights had peak wavelengths in the blue-green spectrums (Fig. S4). On 17 June 2015, our experiment started and the ALAN lights were set on a timer to turn on at 9 pm and turn off at 6 am, coinciding with the times in which the main laboratory lights turn off and on, respectively. We fed toads fruit flies or crickets ad libitum daily. Coinciding with weekly water and leaf replacement, we weighed each toad to determine growth. We kept toads in their treatments until 2 September 2015 at which time we weighed and euthanized all toads.

On 16 July 2015, we measured toad activity during the day and night. During each observation period, we observed toads for 30 seconds, after a 10 second acclimation period. We approached the terraria, waited 10 seconds, and then counted how many toads

were visible for a thirty second period. Two observations were made during the day (all terraria lit) and three observations were made at night (only ALAN terraria lit). We made all night observations using a dimly lit red headlamp. While it was easier to see into the terraria with ALAN in the absence of the red headlamp, the red headlamp sufficiently lit the natural light treatment terraria such that we could clearly see into the terraria in order to count visible toads.

Statistical Analyses:

We tested for effects of larval and juvenile light on toad survival, mass and activity. To determine if there were effects of ALAN on survival, we used ANOVA with larval light, juvenile light and their interaction as explanatory variables and mean survival as the response variable.

We used repeated measures ANOVA to test for treatment effects on toad growth on the following dates: 24 June, 9 July, 24 July, 8 August, 20 August and 2 September.

This statistical model included larval and juvenile light treatments and initial mass (17

June) as between subject test variables and time, the interaction between time and juvenile light, larval light and initial mass as within subject test variables.

For activity, we used repeated measures ANOVA using the proportion of visible toads (number of toad visible divided by total number of toads present) as the response variable and larval light, juvenile light, observation time and their interactions as explanatory variables. We performed all analyses in R using packages: car, ggplot2,

Hmisc, MASS, and Rmisc [33].

Results

Experiment 1: Testing for direct and indirect effects of ALAN in larvae

Mean temperature was affected by the presence of lids (F1,30 = 15.47, p <0.001) but not light (F1,30 = 0.86, p = 0.36) or their interaction (F1,30 = 2.67, p =0.11). Free- colonization mesocosms had mean temperatures of 22.55°C + 0.09 standard error (SE) whereas colonization-limited mesocosms were 21.62°C + 1.22 SE. Mean survival

(78.67% + 0.03 SE) and mass at metamorphosis (0.16 grams (g) + 0.01 SE) were not affected by any treatment or interaction (Table S1). Larval duration was marginally reduced by 1 day in ALAN treatments (F1,7 = 5.04, p = 0.06) and reduced by 8 days in the free-colonization treatments (F1,7 = 179.99, p <0.001). Metamorphic duration was reduced by approximately 30% in the ALAN treatments (F1,7 = 6.39, p = 0.04).

Relative periphyton biomass was reduced by ALAN (F1,32= 7.43, p = 0.01; Table

S2). Mean periphyton abundance was 79.8 + 10.62 SE in mesocosms under natural light conditions and 47.02 + 6.26 SE for mesocosms in ALAN treatments. Relative phytoplankton biomass was increased in colonization-limited mesocosms (F1,32= 4.17, p

= 0.05; Table S2). Mean phytoplankton biomass was 23.85 + 3.59 SE for mesocosms with lids present and 19.7 + 1.77 SE for mesocosms with lids absent.

While the total number of invertebrates was similar across treatments, the colonization-limited treatment reduced the diversity of invertebrates in the mesocosms

(F1,28= 6.94, p = 0.01; Table S3). Mesocosms with free colonization had on average 2.1 +

0.2 SE taxonomic orders of invertebrates, while colonization-limited mesocosms had 1.3

+ 0.2 SE orders of invertebrates. Hylid frogs (spring peepers (Pseudacris crucifer) and gray treefrogs (Hyla versicolor)) readily colonized 24 out of 36 mesocosms (range: 4-481 total individuals collected within all pipe samples, median: 43 individuals across all pipe samples). Many Hylidae tadpoles captured during dipnet sampling had recently hatched and were too small for us to distinguish between species. However, we found breeding H. versicolor adults and mostly larger H. versicolor tadpoles later in the season suggesting the majority of the hylid tadpoles were H. versicolor. Hylid abundance was significantly affected by the interaction between ALAN and toad presence (F1,28 = -3.24, p = 0.004;

Fig. 1). Tukey post-hoc tests (p-value < 0.05) indicate that hylid abundance was significantly higher when toads were present under natural light conditions compared to the other treatments.

Experiment 2: Prior and subsequent effects of light on post-metamorphic toads

While juvenile toad survival was high across all treatments (83% + 5 SE), juvenile toad growth was reduced by exposure to juvenile-stage ALAN (F1,14 = 6.46, p =

0.02; Fig. 2; Table S4). Juvenile toads in the laboratory were marginally more active if they had been raised with ALAN as tadpoles (F1,14 = 4.17, p = 0.06; Fig. S5). Toad activity was also significantly affected by the interaction of juvenile light treatment and observation time (F1,14=14.56, p = 0.002; Table S5). Tukey post-hoc test indicated juvenile toads in the natural light treatment were visible during the day but went under the leaves at night (p = 0.002); juvenile toads in the ALAN light treatment, however, were equally visible in the day and night (p = 0.99) (Fig. 3).

Discussion 47

We were surprised that we did not find evidence for indirect effects of ALAN in our study, despite these effects occurring in other systems (e.g.,[5,34,35]. There were two ways through which we expected indirect effects. First, ALAN might have increased periphyton growth [30] which would in turn affect food availability for toad larvae.

However, while ALAN actually reduced periphyton growth, this did not cause food limitation for the toads: there was no effect of toad presence on periphyton growth, which we would have expected if periphyton was a limiting resource for toads. The second way we could have detected an indirect effect was if ALAN affected colonization of invertebrates, particularly if ALAN increased the abundance of predators on larval toads.

We predicted that ALAN would alter invertebrate immigration and thus community composition through differential attraction of invertebrates to ALAN [5,6,36]. We found no interaction between the light and colonization treatments, which indicated that ALAN did not affect colonization by invertebrates in a way that affected larval toads.

Although there was no effect of ALAN-mediated colonization on toad larvae, hylid treefrogs preferentially oviposited in mesocosms under natural light treatments when toads were present. Previous studies found female hylids avoid ovipositing in wetlands with certain types of predators and conspecifics but did not respond to heterospecifics or nutrients [37–40]. The presence of toad larvae may signal suitable habitat quality for breeding hylids, although mesocosm experiments have found that toad larvae can have a strong competitive effect on hylids [41]. Regardless of the mechanism that caused hylid colonization to be biased towards mesocosms without ALAN and with toads, we found no evidence that hylid colonization affected the toads.

The primary direct effect of ALAN in the larval stage was increased metamorphic synchrony (i.e., reduced metamorphic duration). In ALAN treatments, toads from a single mesocosm typically metamorphosed over six days compared to nine days in natural light.

Increased synchrony of life history events has previously been shown to occur as an anti- predator response [42–44]. Perceived predation risk associated with light, natural or artificial, can alter predation rates and thus prey behaviour in mammals [7,45], fish [46], amphibians [47], and birds [48]. Therefore, the perceived risk associated with light may have induced the metamorphic synchrony in toads under ALAN even in the absence of a predator.

We did not find evidence for strong carry-over effects of ALAN exposure, despite carry-over effects being widespread for other stressors[23–26]. Our experiment found

ALAN had a marginally significant carry-over effect: larval-stage ALAN exposure led to marginally more active juvenile toads. We also did not find larval-stage exposure effects on juvenile growth. There were direct effects of ALAN exposure during the juvenile- stage; ALAN reduced juvenile growth by 15% and eliminated the transition from diurnal to nocturnal behavior. The increased nocturnal activity in ALAN treatments may have resulted in greater energy expenditure and therefore significantly slower juvenile growth despite being fed ad libitum. In contrast, laboratory mice raised in ALAN altered behaviour such that the mice ate during normal times of inactivity which increased weight gain even when the quantity of food consumed between groups remained constant

[14]. There are many negative consequences of reduced growth including delayed reproductive maturity, lower fecundity, and reduced survival [49–52]. Thus, toads

inhabiting terrestrial habitats with ALAN may ultimately have lower fitness, which could scale up to affect survival, fecundity or population growth rates.

Natural light is a critical part of circadian and circannual rhythms which affect ecological systems and individual physiology [53]. As a result, ALAN can have wide reaching effects across taxa and habitats [3]. Amphibians in particular are rapidly declining and this decline has been linked to multiple anthropogenic factors. Despite the many ecological and physiological effects of ALAN, ALAN has not been investigated as a potential contributor to amphibian decline. As a result of this research focusing on multiple mechanisms and life-stages, we detected a particularly troubling finding: reduced juvenile toad growth in response to ALAN. This reduced growth may suggest toads exposed to ALAN may incur reduced fitness later in life, especially if additional stressors (e.g. predators, competitors, desiccation risk) are present and alter the effect of

ALAN. Thus, our work extends our understanding of the scope of ALAN’s effects across ecological contexts and life-stages.

Figures

Figure 1: Effects of light and toad treatments on hylid colonization. Hylid abundance was significantly affected by the interaction between ALAN and toad presence. Tukey post- hoc tests indicated that hylid abundance was significantly higher when toads were present under natural light conditions compared to all other treatments. Letters represent significant (p < 0.05) differences between treatments. Error bars represent standard error.

Figure 2: Effects of juvenile stage light exposure on juvenile toad mass. Juvenile toad growth was reduced by exposure to juvenile-stage ALAN. Error bars represent standard error.

Figure 3: The proportion of visible toads in the ALAN treatment did not differ between day and night observations, whereas toads in natural light treatments were more active during the day than at night. Triangles represent ALAN treatments and circles represent natural light treatments. Error bars represent standard error.

Supplementary Tables

Table 1: Repeated measures analysis of relative algal biomass.

Explanatory Num Den Periphyton Phytoplankton

Variable df df F value p value F value p value

Between Subject Tests:

Light Treatment 1 32 7.43 0.01 0.032 0.9

Colonization Treatment 1 32 0.87 0.4 4.17 0.05

Toad Treatment 1 32 0.82 0.4 0.39 0.5

Within Subject Tests:

Time 1 32 0.4 0.5 0.11 0.7

Light: Time Interaction 1 32 0.17 0.7 0.005 0.9

Colonization: Time Interaction 1 32 0.44 0.5 0.5 0.5

Toad: Time Interaction 1 32 0.8 0.4 1.1 0.3

Table 2: ANOVA and general linear model analysis of invertebrate and hylid colonization

Total Invert Invertebrate Predator Non-Predator Hylid Explanatory Num Den Colonization Orders Presence Presence Colonization Variable df df F value p value F value p value F value p value z value p value t value p value Light 1 28 1.3 0.26 0.3 0.59 0.32 0.58 0.5 0.62 0.85 0.42 Treatment Colonization 1 28 3.32 0.08 6.94 0.01 2.15 0.15 -1.36 0.17 -0.0001 1.0 Treatment Toad 1 28 0.001 0.97 0.14 0.71 0.09 0.76 1.05 0.29 2.94 0.008 Treatment Light* 1 28 0.01 0.9 1.11 0.3 0.03 0.86 -1.57 0.14 <-0.0001 1.0 Colonization Light *Toad 1 28 0.03 0.85 0.002 0.97 0.7 0.41 0.68 0.5 -3.24 0.004 Colonization 1 28 0.46 0.5 0.14 0.71 0.18 0.67 1.41 0.18 0.0001 1.0 * Toad Light* 1 28 0.15 0.7 0.64 0.43 0.03 0.87 -1.46 0.14 -0.00005 1.0 Colonization *Toad

Table 3: Repeated measures analysis of post-metamorphic toad growth

Explanatory Variable Num df Den df F-value p-value Between Subject Tests: 1 14 6.46 0.02 Juvenile Light Treatment Larval Light Treatment 1 14 0.93 0.3 Initial Mass 1 14 11.34 0.005 Within Subject Tests: Time 5 10 4.12 < 0.03 Juvenile Light Treatment: Time 5 10 1.3 0.3 Larval Light Treatment: Time 5 10 0.44 0.8 Initial Mass: Time 5 10 3.67 0.04

Table 4: ANOVA analysis of juvenile toad activity

Explanatory Variable Num df Den df F-value p-value Between Subject Tests: Juvenile Light Treatment 1 14 0.44 0.5 Larval Light Treatment 1 14 4.17 0.06 Juvenile: Larval Light Treatment 1 14 0.44 0.5 Within Subject Tests: Time 1 14 16.6 0.001 Juvenile Light Treatment: Time 1 14 14.56 0.002 Larval Light Treatment: Time 1 14 0.41 0.5 Juvenile Light: Larval Light: Time 1 14 3.09 0.1

Supplementary Figures:

Figure 1: Diagram of experimental design. In experiment 1 (larval stage), we employed a

2x2x2 factorial design manipulating the presence and absence of artificial light (ALAN vs. natural), colonization (lidded vs. unlidded mesocosms to affect colonization) and toad presence or absence. Several of the comparisons we made are indicated by letters. A: comparing mesocosms with ALAN vs natural light tested for the main effect of light at night. B: comparing mesocosms with lids present or absent tested for main effects of colonization. C: comparing mesocosms with toads present or absent determined if toads affect the growth of periphyton, their primary food source. D: Evaluating an interaction

effect between light and colonization tested if light changed the community composition through differential colonization, and if those differences in colonization affected toad growth, development and survival. In experiment 2 (post-metamorphic stage) we took the first eight metamorphosing toads from mesocosms with lids absent and moved them into laboratory terraria to determine direct and carry-over effects of ALAN. We tested for direct effects of ALAN on juvenile toads and asked whether larval or post-metamorphic exposure to ALAN affected post-metamorphic toad growth or activity in laboratory terraria. If there were carry-over effects of ALAN, we expected there to be a larval light or larval and juvenile light interaction effect on juvenile survival, growth or activity.

Figure 2: Photos of experimental set-up. a) Natural light treatment group shown in the daytime. b) Natural light treatment group shown at nighttime. c) ALAN light treatment group shown in the daytime. d) ALAN light treatment group shown at nighttime. Lids made of 60% shadecloth completely covered all mesocosms in the colonization-limited

treatments. Lights in ALAN treatments were mounted in the middle of each ALAN treatment group.

Figure 3: Measurement of light (lux) near wetlands around greater Cleveland, Ohio.

Measurements were taken by walking around wetlands and measuring illuminance.

Average light levels for our experimental ponds measured at water surface was 15.07 lux

+ 7.42 SE without lids and 3.12 lux + 1.84 SE with lids.

Figure 4: Spectrometer readings for outdoor mesocosm and laboratory terraria LED lights. Both lights have peak wavelengths in the blue-green spectrums.

Figure 5: Effects of larval stage light on juvenile toad activity. Effects of larval-stage

ALAN exposure on proportion of visible toads (i.e., activity). Toads raised in ALAN treatments as larvae were marginally more active than toads raised under natural light treatments. 60

Chapter 4: Artificial Light At Night and predator-prey dynamics: Consumptive and non-consumptive effects of predators are not reflected in corticosterone of American toads

Authors: Kacey L. Dananay1, Mandi Schook2, and Michael F. Benard1 1: Department of Biology, Case Western Reserve University, Cleveland, Ohio, USA 2: Animals, Science and Environment, Walt Disney World, Orlando, Florida, USA

Abstract

Artificial Light at Night (ALAN) is an environmental stressor that can disrupt individual physiology and ecological interactions. Hormones such as corticosterone are often responsible for mediating an organism’s response to environmental stressors. We investigated whether ALAN directly caused a stress response and exacerbated the effects of a common stressor, predation. We tested for consumptive, non-consumptive and physiological effects of ALAN and predator (dragonfly larvae) presence on a widespread amphibian, the American toad (Anaxyrus americanus). We found predators had consumptive (decreased survival) and non-consumptive (decreased growth) effects on larval toads. ALAN did not affect larval toads nor did it interact with the predator treatment to increase larval toad predation. Despite the consumptive and non- consumptive effects of predators, neither predators nor ALAN affected corticosterone production in the larval and metamorph life-stages. The lack of a predator effect on larval and metarmorph corticosterone may be due to habituation to predation or decreased

competition in the predator treatments. Our study did not find any evidence to suggest

ALAN alters predator-prey interactions between dragonfly larvae and toads, in contrast to other studies demonstrating ALAN alters predator-prey interactions. As juveniles, larval- stage exposure to ALAN increased corticosterone production in juvenile toads. Our results suggest the physiological effects of ALAN may not be demonstrated until the later life-stages.

Introduction

Artificial Light At Night (ALAN) can disrupt many physiological and ecological interactions [1]. Hormones associated with the neuroendocrine stress axis, like glucocorticoids, are often responsible for mediating an organism’s physiological response to environmental stressors [2]. One such glucocorticoid that is used to assess the physiological effect of many environmental stressors is corticosterone. Corticosterone mobilizes energy by altering metabolism and suppressing immune responses, among others [3]. Corticosterone is most known, however, for its role in regulating stress [4].

Independent of additional environmental stressors, ALAN can affect an individual’s physiology including altered body mass [5] sometimes due to the timing in which foraging takes place [6], slowed larval development [7,8], and various changes in behavior (Dananay and Benard, In review) [9–12]. ALAN induced changes in growth, development and behavior are likely mediated by corticosterone production [4,13]. The few studies investigating ALAN and corticosterone indicate physiological stress is often

[14–17] but not always [18] increased by ALAN. Increased stress as a result of ALAN in birds can increase cholesterol [15] and activity [16], or induce myoclonus [17], but these

effects are not always directly linked to fitness [14].

In natural systems, ALAN is not the only environmental stressor organisms must respond to, however, ALAN can alter how organisms respond to other stressors. One type of environmental stressor that may be susceptible to disruption by ALAN is predation.

For example, ALAN can attract individuals therefore altering the abundance of both predator and prey species [19]. Furthermore, ALAN can also change the strength of species interactions such that predators are more effective at consuming prey [20,21]. If

ALAN strengthens the effects of predators, through what mechanisms can this effect be detected? Predators alone can affect prey in two ways: 1) by killing and eating prey which reduces abundance (i.e. consumptive effects), or 2) by inducing a stress response or behavioral change (i.e. non-consumptive effects) which can ultimately affect individual fitness later in life [22]. When ALAN is also present, it is possible that ALAN can exacerbate these predator effects by acting on one or both mechanisms.

Both consumptive and non-consumptive effects are important aspects of predator- prey dynamics [22,23]. With consumptive effects, predators reduce the abundance of prey; fewer prey increases the per-capita resource availability for surviving individuals

[24,25]. Across different systems, ALAN can sometimes increase and sometimes decrease the consumptive effects of predators. For example, ALAN altered predator consumption by decreasing prey capture by orb-weaving spiders [26] but greatly increased prey consumption by light-tolerant bats, birds, and owls [27–29]. Far fewer studies have investigated predator consumption outside of the terrestrial environment but find ALAN can benefit aquatic predators by increasing consumption [12].

Predators also cause non-consumptive effects, which can have a large effect on prey [22,30,31]. Non-consumptive effects in vertebrates include changes in morphology

[32,33], physiology [34,35] or behavior [36,37], all of which are likely mediated by corticosterone. When ALAN and predators are present, ALAN changes prey behavior.

Many of these behavioral changes are associated with the assessment of risk, as ALAN often leads to an increased perceived predation threat [29,31,38–40]. This can lead to reduced foraging, restricted activity or timing of activity, reduced individual health, low reproductive success, and misorientation, among others [12,41,42].

Here we describe an experiment that investigated the consumptive, non- consumptive and physiological effects of ALAN and/or predator presence on a widespread amphibian, the American toad (Anaxyrus americanus). We investigated the effects of multiple stressors in the larval environment by manipulating the presence and absence of predators (dragonfly larvae) and ALAN. Furthermore, we investigated the effects of ALAN across three life-stages (larval, metamorph and juvenile). We aimed to determine how ALAN and predators affect toads by measuring survival, growth and corticosterone production. While little is known about amphibian responses to ALAN, amphibians have many strategies to increase their survival when predators are present.

These include altering their phenotype, body shape or size as well as the timing in which they metamorphose [43]. Amphibian responses to predation are often mediated by hormones like corticosterone. Amphibians generally respond to predators by initially decreasing corticosterone production [44], however, after the threat of predation becomes chronic, corticosterone will increase and remain high [13]. To account for these diverse

responses, we measured toad mortality, growth, and corticosterone production. We hypothesized that ALAN or predators will decrease toad survival and growth but increase toad corticosterone production. After metamorphosis, we also hypothesized that ALAN would continue to affect individuals by reducing growth and increasing corticosterone production in the terrestrial environment.

We also aimed to determine if predator consumption increases in aquatic environments when ALAN was also present by determining if there are interaction effects between treatments in which both ALAN and predators are present. Regarding amphibians, it is unclear how predation risk, particularly in the aquatic environment, is altered in habitats with ALAN. Amphibians may use light as a cue to forage [45,46]. If light is important, ALAN may increase toad vulnerability to predators at night by increasing toad visibility. Additionally, some natural predators of toads, such as the dragonfly larvae used in this experiment, are visual predators that forage more under daylight conditions [47]. This suggests ALAN may increase the ability of predators to consume tadpoles at night. We hypothesize that ALAN and predators will interact, suppressing toad survival and growth, whereas larval or metamorphic toad corticosterone production will be further increased in treatments with both stressors compared to treatments with a single stressor.

Materials and Methods

Experiment 1: Physiological effects of ALAN and predator presence on larval American toads

We designed an experiment investigating the effects of ALAN and predator presence on the survival, growth, development and hormone production of our focal species, American toads (Anaxyrus americanus), over two life stages: tadpoles and metamorphs. American toads are an ideal model system because they are widespread habitat generalists. Larval American toads are often predated upon by aquatic invertebrates, therefore we used dragonflies as the predator in the larval stage experiments [48].

We used a 2 x 2 experimental design with outdoor mesocosms (2-m diameter polyethelene cattle tanks) to investigate the effects of the presence or absence of ALAN and predators on American toad growth, survival, development and hormone production.

All mesocosms were covered with tightly fitting lids and placed into 10 groups of four mesocosms. Half of these groups were assigned to an ALAN treatment (i.e., nights lit by artificial light); the remaining five groups were assigned to a natural light treatment (i.e., no additional light at night). Within each ALAN or natural light treatment, two of the four mesocosms in each group were randomly assigned to a predator treatment (i.e., 10 total mesocosms per light treatment for a total of 20 mesocosms). Mesocosms assigned to the predator present treatment were stocked with two aeshnid dragonfly larvae.

Mesocosms assigned to the predator absent treatment were not stocked with any predators. This design allowed for all four combinations of ALAN and predator presence to be replicated 10 times.

For the ALAN treatment mesocosms, we used a single 20-Watt LED outdoor flood light (Zitrades, China; 293 nm peak wavelength). We chose LEDs because of their

growing use as indoor and outdoor lighting options. It should be noted different light types produce different wavelengths, thus it possible that different bulb types may have distinct effects on living organisms [49,50]. Despite their energy efficiency, the blue- green wavelengths that LED lights emit can be more disruptive to various organisms including plants, invertebrates and even humans compared to other bulb types [51,52].

Each LED light was mounted in the middle of each ALAN light treatment group of four mesocosms approximately 0.75 meters above the top of the mesocosms. All flood lights were on for 24 hours. Measuring the light reaching the top of the water column averaged

3.12 lux + 1.84 SE. Our LED flood lights were within the range of light levels found at wetlands around the Greater Cleveland area (Fig. S1). Similar to what organisms would experience in natural habitats with ALAN where illuminance decreases with distance, there was a gradient of light present across the mesocosms that ranged from 0.03 to 8.35 lux at water level.

Mesocosms were filled with tap water between 04 and 12 April 2016. Between

12 and 25 April 2016, we added approximately 10 L of mixed hardwood leaf litter, 3.8 L of pond water, 0.3 L of concentrated zooplankton, floating islands of screen for shallow water habitat, and 15 g of rabbit chow to all mesocosms. Mesocosms were left to acclimate until 27 May 2016 when American toad larvae were added to each mesocosm.

On 26 May 2016, we collected toad egg strings from five different areas of one pond at the University Farm. On 27 May 2016, we added 90 toad larvae to each mesocosm and turned on ALAN lights until the end of the experiment (i.e., all toads metamorphosed).

Toads were checked every 1-3 days throughout the experiment.

On 10 June 2016 starting at 3:00 am, 6 tadpoles were collected from each mesocosm for hormone analysis. Samples were collected using a red headlamp. Once samples were collected, they were immediately euthanized using MS-222, weighed and flash frozen with distilled water at 1:1 g/mL ratio on dry ice for hormone analysis

(described below). Samples were stored at -80°C until tissues were processed. To test for treatment effects on toad growth, development and survival, we counted and weighed all metamorphosing toads. On 17 June 2016, the first metamorphs emerged. Between 19 and 24 June 2016 starting at 1:00 am, 6 metamorphs were collected from each mesocosm for hormone analysis (described below). These 6 metamorphs (removed after the first 10 individuals metamorphosed) were collected, weighed, euthanized (MS-222), flash frozen and stored at -80°C until hormone processing occurred. We weighed, euthanized and preserved all other metamorphs in 70% ethanol as they emerged.

Statistical Analysis

We tested for the effects of ALAN and predator presence on toad survival, growth, development, and corticosterone production. Larval survival was tested using a generalized linear mixed model using a penalized quasi-likelihood model that included a poisson distribution and a random effect of mesocosms nested within light treatment.

This model was chosen because our survival data was bimodal due to treatment effects and the random effect was to account for the groups of four mesocosms. We tested for effects of ALAN and predators on mass at metamorphosis, larval duration, metamorphic duration, and corticosterone production using a general linear mixed effect model with a random effect of mesocosm nested within light treatment. Larval duration, the length of

time between hatching and metamorphosis, was inverse transformed (1/larval duration).

Metamorphic duration, the time between the first and last metamorph to emerge from each mesocosm, was log transformed. Both tadpole corticosterone and metamorph corticosterone were log transformed to meet the assumption of normality.

To further examine the effects of our treatments on metamorphosed toads’ corticosterone, we tested for effects of survival to metamorphosis (i.e. toad survival), predator treatment and light treatment using an ANCOVA. Our rationale for this additional analysis was to account for density, which is known to affect corticosterone

[53]. Our experimental unit was mesocosm. All analyses were completed using the statistical program R [54].

Experiment 2: Effects of larval stage and juvenile stage ALAN on juvenile toads

After collecting the first 10 toads to metamorphose from all predator absent mesocosms in Experiment 1, they were transported back to the laboratory. Predator treatments were not carried to the post-metamorphic life-stage because there was not enough laboratory space for all 80 terraria. Five of the ten toads in each predator absent mesocosm were randomly assigned to the ALAN treatment; the remaining five toads were assigned to the natural light treatment. We had a total of 40 experimental units (20 per juvenile light treatment), each starting with a density of 5 toads per terraria. Thus, all four combinations of larval and juvenile light treatments were replicated 10 times.

We manipulated ALAN in both the larval stage as explained in Experiment 1 using outdoor mesocosms and in the juvenile stage using lidded plastic terraria (42.5cm

X 30.2cm X 17.8cm) as described in Dananay and Benard, In review. Briefly, four sets of shelves were used with two shelves back-to-back with plastic foam sheets on 3 sides to block light from entering or keep light from leaving the treatment shelves. ALAN lights used in the laboratory were 24-Watt LEDs (LE, China; peak wavelength 320 nm, 16.9 lux + 3.04 SE) placed 50.8 cm above the terraria set on a timer to turn on at 9 pm and turn off at 6 am such that experimental terraria always had lit conditions. We fed toads fruit flies or crickets ad libitum 3-7 times per week. We completely replaced the water and leaves weekly. We weighed each toad biweekly to determine growth.

On 3 August 2016, we measured toad activity during the day and night according to Dananay and Benard, In review. Two observations were made during the day (all terraria lit) and two observations were made at night (only ALAN terraria lit). We made all night observations using a dimly lit red headlamp. We kept toads in their respective treatments until the experiment ended on 6 September 2016 at 1 am when 2 individuals were immediately euthanized, weighed and flash frozen with distilled water at 1:1 g/mL ratio on dry ice for hormone analysis (see below). Samples were stored at -80°C until tissues were processed. All other individuals were euthanized and preserved in 70% ethanol.

Statistical Analysis

We tested for the effects of larval and juvenile stage ALAN on juvenile toad survival, growth, behavior, and corticosterone production. For mean survival, we used an

ANOVA model with larval light, juvenile light and their interaction as explanatory variables. For growth and behavior, we used repeated measures ANOVA. The model for 70

toad growth included mass measure on 30 June, 16 July, 25 July, 17 August and 6

September. This statistical model included larval and juvenile ALAN treatments and initial mass at metamorphosis as between subject test variables and time, the interaction between time and juvenile light, larval light and initial mass as within subject test variables. For activity, we used repeated measures ANOVA using the proportion of visible toads (number of toad visible divided by total number of toads present) as the response variable and larval light, juvenile light, observation time and their interactions as explanatory variables. Juvenile corticosterone was log transformed to meet the assumption of normality. Our experimental units were the terraria.

Hormone extraction and corticosterone radioimmunoassay (RIA) procedures

We extracted corticosterone on whole bodied amphibians as described and validated by Bennett et al. [35]. Briefly, tissues prefrozen in water were defrosted just enough to allow homogenization with a, tissue homogenizer (Omni International, Tissue

Master 125, United States). Methanol was added at a ratio of 0.5ml methanol for every

0.5 g of pre-weighed tissue, and vortexed for 1 minute. Six mL of dichloromethane was then added to each sample and placed on a shaker for 15 minutes. Samples were then frozen at -80°C for 1 hour and centrifuged at 1°C for 15 mins at 3500 X g. The supernatant was removed, placed in a 15-mL test tube, and dried down overnight under air in a warm water bath at 35°C. Dried test tubes were then capped and frozen at -80°C until RIA analyses were completed.

We used a corticosterone radioimmunoassay from MP Biomedicals to analyze samples for corticosterone concentrations. Samples were reconstituted with 600 µL of 71

corticosterone steroid diluent, diluted at a ratio of 1:3 for larval and metamorphic life- stages and 1:4 for the juvenile life-stage. This RIA was conducted according to manufacturer’s directions with the exception that we halved all manufacturer recommended ingredients in each tube. All samples were run in duplicate, thus the average between the two samples were taken and used in our analyses. Only samples with lower than 12% CV were used in the analysis. Toad dilutions of extracts were parallel to the standard curve of the RIA. Recovery of samples spiked with high and low controls averaged 92% and 89.5%, respectively. Whole-body corticosterone content was measured in pg/g body mass.

Results

Experiment 1: Physiological effects of ALAN and predator presence on larval American toads

Predator presence significantly reduced toad survival (t28 = -4.37, p = 0.002).

Mean survival in predator absent treatments was 79% + 0.03 standard error (SE) compared to 50% + 0.04 SE in the predator present treatment (Fig. 1a). Neither ALAN (t8

= -0.52, p = 0.61) nor the interaction between ALAN and predator treatments (t28 = 0.85, p = 0.40) had an effect on toad survival.

Predator treatment also significantly reduced toad mass at metamorphosis (F1,28 =

85.08, p < 0.001). Mean toad mass in predator absent treatments was 0.19 g + 0.01 SE compared to 0.15 g + 0.01 SE in the predator present treatment (Fig. 1b). Neither ALAN

(F1,8 = 2.77, p = 0.13) nor the interaction between ALAN and predator treatments (F1,28 =

0.01, p = 0.96) had an effect on toad mass.

Larval duration was not affected by predator treatment (F1,28 = 2.90, p = 0.10),

ALAN treatment (F1,8 = 0.84, p = 0.39), or their interaction (F1,28 = 1.75, p = 0.20).

Metamorphic duration was affected predator treatment (F1,28 = 0.02, p = 0.89), ALAN treatment (F1,8 = 2.07, p = 0.19), or their interaction (F1,28 = 0.38, p = 0.54).

Tadpole corticosterone production was not affected by predator treatment (F1,28 =

0.71, p = 0.41), ALAN treatment (F1,8 = 0.94, p = 0.36), or their interaction (F1,28 = 0.40, p = 0.53; Fig. 1c). Metamorph corticosterone production was not affected by predator treatment (F1,28 = 0.14, p = 0.72), ALAN treatment (F1,8 = 0.31, p = 0.59), or their interaction (F1,28 = 0.09, p = 0.77; Fig. 1d). Further analysis of the metamorph corticosterone data that included survival as a covariate determined that there was a significant interaction between survival and predator treatment (X2 = 5.64; p = 0.02) and survival and ALAN treatment (X2 = 4.53; p = 0.03) (Table 1, Fig. 2). When predators were absent, corticosterone decreased with increasing survival whereas predator present treatments increased corticosterone with increasing survival (Fig. 2a). While ALAN treatments had a greater decrease, both light treatments decreased corticosterone production with increasing survival (Fig. 2b).

Experiment 2: Effects of larval stage and juvenile stage ALAN on juvenile toads

Mean juvenile toad survival in the laboratory was 96% + 1 SE and was not affected by larval or juvenile stage ALAN. Toad growth increased over time but was not

affected by any treatment (Table 2). Toad behavior was significantly affected by observation time: toads were more active during the day than the night (Table 3). There were no treatment effects on behavior (Table 3). Carry-over effects were present in toad corticosterone production. Corticosterone in juvenile toads was significantly increased by larval ALAN exposure (F1,35 = 7.7, p = 0.009). Corticosterone production of toads raised with ALAN averaged 5.72 ng/g body mass (BM) + 0.11 SE whereas natural light was

5.17 ng/g BM + 0.16 SE. Juvenile ALAN (F1,35 = 1.8, p = 0.19) and the interaction between larval and juvenile ALAN treatments did not significantly affect corticosterone

(F1,35 = 0.6, p = 0.4).

Discussion

We did not find an effect of ALAN on corticosterone production until the juvenile life-stage where we found a carry-over effect of being exposed to larval-stage ALAN. We used white LEDs throughout the experiment. The use of LED and fluorescent lights as the source of ALAN can increase corticosterone production in songbirds [14,18] and cortisol in mammals [53]. Both white and red LED light increased corticosterone production in birds [14]. In contrast, another study found no effect of green, red or blue

LED lights on corticosterone production on laboratory raised fish [54], but did not measure the effect of white light. The illuminance levels of our light may be the reason we did not see a physiological effect of ALAN on larval toads. The illuminance levels used in the Ouyang et al. [14] study on songbirds were almost 3 times as high as the average illuminance level used here (8.2 lux vs. 3.12 lux). However, our values across the mesocosm ranged from 0.03 to 8.35 lux at water level. Our illuminance levels were

chosen to mimic natural conditions found around Greater Cleveland Ohio wetlands

(Chapter 3, Fig. S2). While higher illuminance levels were present in our survey, many of these measures were not taken at the water surface due to the inability to access the water

(e.g. fences, difficult terrain). If measurements were taken at the water surface, the illuminance levels would have decreased. Although our levels of illuminance were low, they are realistic to natural habitats in this area and no effects of ALAN on larval toads were found.

While we did not see short-term effects of larval ALAN on the larval or metamorphic life-stages, we did find larval-stage exposure to ALAN affected toads later in life. Regardless of their juvenile light treatment, toads raised with ALAN as larvae had increased corticosterone production as juveniles suggesting the effects of larval-stage

ALAN carried over to affect juveniles. The few studies that have investigated carry-over effects of ALAN have found evidence to suggest ALAN can cause carry-over effects

Dananay and Benard, In Review, [55]. One such study found negative effects resulting from ALAN could be reversed after a few months of exposure to natural photoperiods

(i.e. dark nights) [55]. However, our study suggests the negative carry-over effects of increased corticosterone production was not mitigated within the time frame of this study

(about 12 weeks). While increases in corticosterone can increase survival in the short- term, chronic increases in corticosterone are often associated with negative long-term effects such as the suppression of reproduction [56,57] and immune function [58,59].

Therefore, it is possible that the long-term increase in corticosterone production may reduce fitness later in life for individuals that develop in environments with ALAN.

Despite ALAN not altering the strength of predator-prey interactions, larval dragonflies themselves influenced the toads by decreasing growth and increasing mortality. However, the reduction of survival and growth were not associated with changes in corticosterone production. This contrasts with other studies that have shown predators increase corticosterone production in amphibians [13,35,60]. In tadpoles, corticosterone production typically shows an initial spike of corticosterone within minutes of detecting a predator [61], then a decrease after a few hours [44].

Corticosterone will increase and remain high for days to weeks with chronic exposure to predation [13] before returning to baseline [35]. While our larval toads were experiencing chronic exposure, our results do not show elevated corticosterone production. We have three hypotheses as to why predator effects on larval toad corticosterone in our study may have been lost or silenced: 1) timing of hormone collection, 2) habituation to predators, and 3) predators altering intraspecific competition.

We collected hormone samples at night. Corticosterone has a circadian rhythm such that production is highest during the day and at baseline at night [62–64]. However, because we collected all samples during the same timeframe, we do not believe the circadian rhythm was a factor in our lack of a stress response. So, while we collected at a time where production is lowest, if production was increased overall (i.e. a chronic response), baseline values would have been different thus an effect would have still been seen regardless of timing [63,64].

Habituation to predator presence is also a possible explanation for the lack of an increase in corticosterone when predators were present. Wood frogs (Lithobates

sylvatica) reared for three weeks with chronic exposure to predator cue caused a dampened corticosterone response to predation threats [61]. This significant but smaller corticosterone response of wood frogs to predators was hypothesized to be due to tadpole body size. While our sampling took place approximately one week before the start of metamorphosis, our toads were small and thus would likely still be experiencing high predation.

Another possibility is that the effects of competition countered the effects of predators. Our experiment demonstrated predators reduced survival through consumptive effects. Fewer toads in the predator treatments would reduce the intraspecific competitive effects on surviving toads. Stronger competition resulting from high densities can sometimes [65] but not always [66] increase corticosterone production. As discussed previously, predators can increase corticosterone production [35,60,67]. Predators also increase mortality which decreases intraspecific competition. Lower intraspecific competition can decrease corticosterone production [65]. Thus, decreased competition in predator treatments may counteract the predator-induced increase in corticosterone.

These antagonistic effects between predators and competition could then explain why there were no differences in corticosterone between predator treatments. Our additional

ANCOVA analysis of the effects of predators on metamorph corticosterone shows corticosterone production slightly decreases as survival increased when predators were present but this slope is not significantly different than zero. Studies on predator-induced changes in corticosterone production use either caged predators (i.e. non-lethal) [13,68] or predator cue water [35,60] rather than the free-ranging predators used here. Decreased

competition from direct consumption is not present in any previous studies we are aware of. Using free-ranging predators is a more realistic environment compared to caged predators and may result in having different physiological effects.

Despite the lack of predator effects on corticosterone, predatory dragonfly larvae did have consumptive, and non-consumptive effects on toads; toads reared with predators had reduced survival and reduced growth regardless of ALAN treatments. Predator- induced reductions in survival and growth are common responses of toads to predators

[48,69,70]. While activity was not measured here, the reduction of toad growth has been associated with reduced activity [48,71]. Reduced activity decreases the visibility of toads to predators but also reduces their foraging time; reduced foraging ultimately reduces their growth. While this is a common response for toads, not all amphibians respond to predators by altering their size at metamorphosis (i.e. growth) [43]. For example, when predators are present but food is limited, predator-induced reductions in activity could prevent prey from wasting energy looking for food that is not present [72] or by releasing tadpole food resources from herbivory thus allowing them to build-up

[73,74]. Both situations can increase amphibian size at metamorphosis. Prey species can also suppress activity until they reach a body size that releases them from the threat of predation if that predator is gape-limited; this strategy can increase larval duration but also causes them to metamorphose at a larger size [75]. For this experiment, we used free-ranging predators, but caged predators have similar effects [22] suggesting non- consumptive effects are important regardless if the predators is free-ranging or caged.

While we were not surprised to see strong effects of predators, we were surprised that

ALAN did not change the strength of these predator-prey interactions.

ALAN can alter predator-prey dynamics, but the direction of those alterations vary between taxa and environments [1,41,76]. There are cases in which predators benefit from the addition of light at night and have better foraging success [27,29]. Alternatively, predators may be avoiding light themselves [77], or may simply not increase foraging success from the addition of light [26]. The reason we hypothesized ALAN would benefit the predators in our experiment was two-fold: 1) the potential effect of light on toad behavior [45], and 2) the highly visual mode of prey capture used by dragonflies [47]. If toads respond to ALAN like natural light, toads in the ALAN treatments would continue to be active at night. This increased nocturnal activity could then increase their predation risk. In addition to dragonflies being visual predators, dragonflies sometimes [47], but not always [78] forage more during the day than night. If predators could consume more prey due to ALAN, we would have expected the ALAN and predator combined treatments to further decrease the survival of our toads. This predicted effect, however, was not seen in our experiment and suggests that predatory dragonflies do not consume more toads in artificially lit habitats.

In the larval stage, we did not find any evidence to suggest ALAN alters predator- prey interactions between larval dragonflies and toads, in contrast to other studies of

ALAN’s effects on predator-prey interactions. In the juvenile stage, we found further evidence to suggest the effects of ALAN can persist in later life-stages even if those effects are not seen earlier, as demonstrated by corticosterone production. This chronic increase in production of corticosterone could lead to decreased fitness later in life.

Future work on the physiological effects of environmental stressors should encompass multiple life-stages, as the larger time-frame used here helped shed light on the long-term physiological effects of ALAN.

Tables

Table 1: ANCOVA results of Metamorph Corticosterone Production

Response Estimate SE df t-value p-value Intercept 12.62 2.29 23 5.51 0.0 Percent Survival -7.07 2.76 23 -2.56 0.02 Light Treatment -5.42 2.44 8 -2.22 0.06 Predator Treatment -5.78 2.35 23 -2.46 0.02 Survival: Light 6.31 2.96 23 2.13 0.04 Survival: Predator 6.98 2.94 23 2.37 0.03 Light: Predator 4.86 2.59 23 1.88 0.07 Survival: Light: Predator -5.33 3.36 23 -1.59 0.13

Table 2: Repeated measures ANOVA analysis of juvenile toad growth

Explanatory Variable Num df Den df F-value p-value

Between Subject Tests:

Larval Light Treatment 1 36 0.8 0.39

Juvenile Light Treatment 1 36 3.5 0.07

Initial Mass 1 36 0.7 0.42

Within Subject Tests:

Time 4 33 344.2 <0.001

Larval Light Treatment: Time 4 33 0.1 0.99

Juvenile Light Treatment: Time 4 33 1.4 0.26

Initial Mass: Time 4 33 1.4 0.25

Table 3: Repeated measures ANOVA analysis of juvenile toad activity

Explanatory Variable Num df Den df F-value p-value

Between Subject Tests:

Larval Light Treatment 1 36 2.72 0.11

Juvenile Light Treatment 1 36 0.42 0.52

Larval: Juvenile Light Treatment 1 36 0.64 0.43

Within Subject Tests:

Time 1 36 21.76 >0.001

Larval Light Treatment: Time 1 36 2.06 0.16

Juvenile Light Treatment: Time 1 36 1.16 0.29

Juvenile Light: Larval Light: Time 1 36 1.67 0.20

Figures

Figure 1: Treatment effects on survival, growth and corticosterone production. a) Tadpole survival to metamorphosis. Predators significantly reduced survival regardless of light treatment. b) Mass at metamorphosis. Predators significantly reduced metamorph mass regardless of light treatment. c) Corticosterone production in tadpoles. Neither predators

nor ALAN treatment affected corticosterone production in tadpoles. d) Corticosterone production in metamorphs. While production was increased compared to tadpoles, there were no treatment effects on metamorph corticosterone production. Corticosterone values are log transformed and corrected for body mass (BM). Open bars represent

ALAN treatments, filled bars represent natural light treatments. Error bars represent standard error.

Figure 2: Treatment effects on survival and metamorph corticosterone production. a)

Metamorph corticosterone production decreased with increasing survival in predator present treatments; corticosterone production increased with increasing survival in predator absent treatments. Black points are predator present treatments, gray points are predator absent treatments. b) Metamorph corticosterone production decreased as survival increased in both light treatments but ALAN treatments had a larger negative slope. Black squares are ALAN light treatments, gray squares are natural light treatments.

Chapter 5: Conclusions

In this chapter, I will: 1) explain how my research has advanced the field of environmental stressor research, and 2) explain what future research should focus on.

Advancements of my research program

Overall, my research highlighted the importance of extending experiments or creating realistic environments to help reduce the likelihood of under- or over-estimating the effects of stressors. This was done by focusing on three important aspects of environmental stressor research that is currently lacking: 1) ecologically relevant exposures to stressors, 2) focus on direct and indirect effects together, and 3) long-term experiments spanning multiple life-stages.

The amount of a stressor organisms come in contact with is a key part of designing experiments such that they can be related back to the natural environment. For the research presented here, field measures of each stressor were quantified around the greater Cleveland area. These field measures directly informed my research by providing relevant measures of stressors organisms are likely to experience. In particular, Chapter 2 highlighted previous research that pointed to direct negative effects of road salt yet the salt concentrations used in these studies ranged from 300 to 26,000 mg/L of salt [1–3]; this compares to our measures from North East Ohio wetlands which were between 20 and 700 mg/L. Studies using these high concentrations of salt found increased mortality associated with salt. My study used concentrations up to 900 mg/L and found high survival. Unlike road salt, aside from a few exceptions of light levels between 400 and

1000 lux [4], Artificial Light At Night (ALAN) research is investigated with relevant levels of light.

In addition to relevant stressor exposure, road salt and ALAN research more often focuses on direct rather than indirect effects. Environmental stressors can alter how organisms interact with other species (indirect effects) as well as an individual organism’s physiology (direct effects). Indirect effects include altered community composition, spreading of invasive species, and altering predator-prey interactions [5–

11]. Direct effects include altered endocrine production, development and behavior, and decreased immune function and reproductive output [12–19]. Without measuring indirect effects, environmental stressors may appear to have direct effects when in fact indirect effects may be at work. This occurred in Chapter 2 when it appeared road salt increased larval growth. However, it was likely an indirect effect of salt reducing the survival of an algal competitor, zooplankton. The effect of salt on zooplankton abundance then increased the abundance of algal food resources for tadpoles. As shown here, measuring direct and indirect effects together can increase our knowledge of the effects of environmental stressors. When designing my research program, I aimed to determine both indirect and direct effects. I determined the indirect effects, by measuring changes in food abundance (Chapters 2, 3) or colonization (Chapter 3), and direct effects, by measuring changes in growth (Chapters 2, 3, 4), development (Chapters 2, 3, 4) or corticosterone production (Chapter 4). Particularly in chapter 3, I was able to determine the relative importance each of these effects. While direct effects were shown to have a larger impact compared to indirect effects originating from the ecological community,

altering our experiment to increase the chances of colonization should be investigated first.

The final aspect environmental stressor research needs are experiments that extend beyond a single life-stage. The effects of early exposure to stressors such as herbicides [20], predators [21], and competitors [22] can persist beyond a single life- stage. Furthermore, because juvenile survival has a large effect on population dynamics

[23], studies identifying carry-over effects will help estimate the long-term effect of stressors. Thus, studies investigating solely a single life-stage are likely under- or over- estimating the effect of those stressors on their model system. The research presented here encompassed multiple life-stages. This allowed us to determine the effects of that stressor in each life-stage but also determine if the effects of that stressor carry-over and affect later life-stages. In Chapters 2 and 3, I found carry-over effects were present for road salt and ALAN. The carry-over effects seen in Chapter 3 demonstrated ALAN had a marginal effect on juvenile activity. In Chapter 2, however, carry-over effects were particularly important. Generally increased mass at metamorphosis is associated with greater fitness [24,25]. If I had stopped this experiment at metamorphosis, I would have concluded that low concentrations of road salt could be beneficial to wood frogs because of this increased size at metamorphosis. However, extending this experiment found juveniles exposed to salt as larvae had increased mortality, particularly in low-density terrestrial environments. Therefore, the apparent positive effects of road salt in the larval stage were counteracted by negative effects found in the juvenile stage.

Future directions

While my research program focused on amphibian model systems, the effects of stressors are far-reaching across taxa. Both stressors studied here can affect aquatic and terrestrial systems as well as the many organisms within each of these habitats. While I have established that there are many effects of these stressors and designing experiments to encompass these diverse effects are needed, their still remains one question: what can be done to mitigate the effects of these stressors? For both stressors, human safety is a main concern. Road salt ensures safer roadways or walkways. Light is also used for safety but is also important for general recreation and shift-working. Thus, these environmental stressors will likely not be eliminated from our environment but there are still ways to mitigate their effects.

For road salt, many alternatives have been suggested such as beet juice, pickle brine, cheese brine, potato juice, and sand. However, no research I am aware of has been published to determine the effectiveness of these alternatives as deicers, nor of their effect on the environment. Beet juice appears to be the most popular choice but the concern would be how the added sugar would affect the environment. Until research is performed to determine the effectiveness and impact of these alternatives, the best strategy to mitigate the effects of road salt on the environment are to use salts sparingly.

This will help road salt be used as intended, without placing excessive quantities of salt into the environment unnecessarily.

For ALAN, there are no alternatives to light, but there are still potential ways to mitigate the effect of that light. In order to mitigate light, more detailed research is needed. These studies should: 1) investigate ALAN at multiple levels rather than binary

presence/absence, and 2) determine if there are bulb types or colors that are less disruptive to the environment. In addition to studies on the impact of different levels or types of light, the general design of our outdoor lighting could also be modified. For example, lights should have shades to angle the light down rather than up into the environment where it is no longer useful for its intended purpose. My final suggestion would be to understand the ecology of local species and have a plan to mitigate the effects of light at particular times. This strategy has been successfully used in Florida who has implemented a “Lights-Out” program to mitigate the effect of lights during turtle breeding and hatchling seasons. This program can be used as a model for species that are particularly prone to ALAN. Turning off lights, shielding lights, or using less disruptive types of lights during the peak of activity, may be able to mitigate the effects of that light on species and also reduce the impact of that light on other non-target species.

BIBLIOGRAPHY

Chapter 1:

1. Millennium Ecosystem Assessment. 2005. Ecosystems and Human Well-being: Synthesis. Ecosystems. 5, Island Press, Washington, DC, USA. 2. Isaac NJB, Turvey ST, Collen B, Waterman C, Baillie JEM. 2007. Mammals on the EDGE: Conservation priorities based on threat and phylogeny. PLoS One. 2:e296. 3. Barnosky AD, Matzke N, Tomiya S, Wogan GOU, Swartz B, Quental TB, Marshall C, McGuire JL, Lindsey EL, Maguire KC, Mersey B, Ferrer E A. 2011. Has the Earth’s sixth mass extinction already arrived? Nature. 471:51–57. 4. Holmstrup M, Bindesbøl AM, Oostingh GJ, Duschl A, Scheil V, Köhler HR, Loureiro S, Soares AMVM, Ferreira ALG, Kienle C, Gerhardt A, Laskowski R, Kramarz PE, Bayley M, Svendsen C, Spurgeon DJ. 2010. Interactions between effects of environmental chemicals and natural stressors: A review. Sci. Total Environ. 408:3746– 3762. 5. D’Itri FM. 1992. Chemical Deicers and the Environment. Lewis Publishing, Boca Raton, Florida. 6. Kostick DS. 2011. Salt. 2010 Miner. Yearb. 7. Cañedo-Argüelles M, Grantham TE, Perrée I, Rieradevall M, Céspedes-Sánchez R, Prat N. 2012. Response of stream invertebrates to short-term salinization: A mesocosm approach. Environ. Pollut. 166:144–151. 8. Hart BT, Bailey P, Edwards R, Hortle K, James K, Andrew M, Meredith C, Swadling K. 1991. A review of the salt sensitivity of the Australan freshwater biota. Hydrobiologia. 210:105–144. 9. Trombulak SC, Frissell C A. 2000. Review of Ecological Effects of Roads on Terrestrial and Aquatic Communities. Conserv. Biol. 14:18–30. 10. Richard T, Forman T, Deblinger RD. 2000. The ecological road-effect zone of a Massachusetts (U.S.A.) suburban highway. Conserv. Biol. 14:36–46. 11. Sanzo D, Hecnar SJ. 2006. Effects of road de-icing salt (NaCl) on larval wood frogs (Rana sylvatica). Environ. Pollut. 140:247–56. 12. Collins SJ, Russell RW. 2009. Toxicity of road salt to Nova Scotia amphibians. Environ. Pollut. 157:320–4.

13. Petranka JW, Francis R A. 2013. Effects of road salts on seasonal wetlands: Poor prey performance may compromise growth of predatory salamanders. Wetlands. 33:707– 715. 14. Cinzano P, Falchi F, Elvidge CD. 2001. The first World Atlas of the artificial night sky brightness. Mon. Not. R. Astron. Soc. 328:689–707. 15. Hölker F, Moss T, Griefahn B, Kloas W, Voigt CC, Henckel D, Hänel A, Kappeler PM, Völker S, Schwope A, Franke S, Uhrlandt D, Fischer J, Klenke R, Wolter C, Tockner K. 2010. The Dark Side of Light: A Transdisciplinary Research Agenda for Light. Ecol. Soc. 15. 16. Hölker F, Wolter C, Perkin EK, Tockner K. 2010. Light pollution as a biodiversity threat. Trends Ecol. Evol. 25:681–682. 17. Gaston KJ, Visser ME, Hölker F. 2015. The biological impacts of artificial light at night: The research challenge. Philos. Trans. R. Soc. B. 370:20140133. 18. Navara KJ, Nelson RJ. 2007. The dark side of light at night: Physiological, epidemiological, and ecological consequences. J. Pineal Res. 43:215–224. 19. Gaston KJ, Davies TW, Nedelec SL, Holt LA. 2017. Impacts of Artificial Light at Night on Biological Timings. Annu. Rev. Ecol. Evol. Syst. 48:49–68. 20. Davies TW, Duffy JP, Bennie J, Gaston KJ. 2014. The nature, extent, and ecological implications of marine light pollution. Front. Ecol. Environ. 12:347–355. 21. Gaston KJ, Bennie J, Davies TW, Hopkins J. 2013. The ecological impacts of nighttime light pollution: a mechanistic appraisal. Biol. Rev. 88:912–927. 22. Longcore T, Rich C. 2004. Ecological Light Pollution. Front. Ecol. Environ. 2:191–198. 23. Rich C, Longcore T. 2005. Ecological Consequences of Artificial Night Lighting. Island Press, Washington, DC, USA. 24. Ohta H, Mitchell AC, McMahon DG. 2006. Constant light disrupts the developing mouse biological clock. Pediatr. Res. 60:304–308. 25. IUCN. 2017. The IUCN Red List of Threatened Species. Version 2017-3. Available from http://www.iucnredlist.org/initiatives/amphibians/analysis. 26. Roelants K, Gower DJ, Wilkinson M, Loader SP, Biju SD, Guillaume K, Moriau L, Bossuyt F. 2007. Global patterns of diversification in the history of modern amphibians. Proc. Natl. Acad. Sci. 104:887–892. 27. Collins JP, Storfer A. 2003. Global amphibian declines: sorting the hypotheses. Divers. Distrib. 9:89–98.

28. Collins JP. 2010. Amphibian decline and extinction: What we know and what we need to learn. Dis. Aquat. Organ. 92:93–99.

Chapter 2:

1. Blaustein AR, Han BA, Relyea RA, Johnson PTJ, Buck JC, Gervasi SS, Kats LB. 2011. The complexity of amphibian population declines: Understanding the role of cofactors in driving amphibian losses. Ann N Y Acad Sci 1223:108-119. 2. Pereira HM, Leadley PW, Proenca V, Alkemade R, Scharlemann JPW, Fernandez-Manjarres JF, Araujo MB, Balvanera P, Biggs R, Cheung WWL, Chini L, Cooper HD, Gilman EL, Guenette S, Hurtt GC, Huntington HP, Mace GM, Oberdorff T, Revenga C, Rodrigues P, Scholes RJ, Sumaila UR, Walpole M. 2010. Scenarios for global biodiversity in the 21st century. Science 330:1496-1501. 3. Kerby JL, Richards-Hrdlicka KL, Storfer A, Skelly DK. 2010. An examination of amphibian sensitivity to environmental contaminants: Are amphibians poor canaries? Ecol Lett 13:60-67. 4. Vonesh JR, De la Cruz O. 2002. Complex life cycles and density dependence: Assessing the contribution of egg mortality to amphibian declines. Oecologia 133:325- 333. 5. Rohr JR, Raffel TR, Halstead NT, McMahon TA, Johnson SA, Boughton RK, Martin LB. 2013. Early-life exposure to a herbicide has enduring effects on pathogen- induced mortality. Proc R Soc Lond B Biol Sci 280:20131502. 6. Johnson PTJ, Lunde KB, Ritchie EG, Launer AE. 1999. The effect of trematode infection on amphibian limb development and survivorship. Science 284:802-804. 7. Hayes T, Haston K, Tsui M, Hoang A, Haeffele C, Vonk A. 2003. Atrazine- induced hermaphroditism at 0.1 ppb in American leopard frogs (Rana pipiens): Laboratory and field evidence. Environ Health Perspect 111:568-575. 8. Collins JP, Storfer A. 2003. Global amphibian declines: Sorting the hypotheses. Divers Distrib 9:89-98. 9. Relyea RA. 2005. The impact of insecticides and herbicides on the biodiversity and productivity of aquatic communities. Ecol Appl 15:618-627. 10. Relyea RA, Schoeppner NM, Hoverman JT. 2005. Pesticides and amphibians: The importance of community context. Ecol Appl 15:1125-1134. 91

11. Rohr JR, Palmer BD. 2005. Aquatic herbicide exposure increases salamander desiccation risk eight months later in a terrestrial environment. Environ Toxicol Chem 24:1253-1258. 12. Scott DE. 1994. The effect of larval density on adult demographic traits in Ambystoma opacum. Ecology 75:1383-1396. 13. Benard MF, Fordyce JA. 2003. Are induced defenses costly? Consequences of predator-induced defenses in western toads, Bufo boreas. Ecology 84:68-78. 14. Boone MD. 2005. Juvenile frogs compensate for small metamorph size with terrestrial growth: overcoming the effects of larval density and insecticide exposure. J Herpetol 39:416-423. 15. Webber NR, Boone MD, Distel CA. 2010. Effects of aquatic and terrestrial carbaryl exposure on feeding ability, growth, and survival of American toads. Environ Toxicol Chem 29:2323-2327. 16. Sanzo D, Hecnar SJ. 2006. Effects of road de-icing salt (NaCl) on larval wood frogs (Rana sylvatica). Environ Pollut 140:247-256. 17. Harless ML, Huckins CJ, Grant JB, Pypker TG. 2011. Effects of six chemical deicers on larval wood frogs (Rana sylvatica). Environ Toxicol Chem 30:1637-1641. 18. Collins SJ, Russell RW. 2009. Toxicity of road salt to Nova Scotia amphibians. Environ Pollut 157:320-324. 19. Forman RTT, Deblinger RD. 2000. The ecological road-effect zone of a Massachusetts (USA) suburban highway. Conserv Biol 14:36-46. 20. Karraker NE, Gibbs JP. 2011. Road deicing salt irreversibly disrupts osmoregulation of salamander egg clutches. Environ Pollut 159:833-835. 21. Hua J, Pierce BA. 2013. Lethal and sublethal effects of salinity on three common Texas amphibians. Copeia:562-566. 22. Van Meter RJ, Swan CM, Leips J, Snodgrass JW. 2011. Road salt stress induces novel food web structure and interactions. Wetlands 31:843-851. 23. Petranka JW, Francis RA. 2013. Effects of road salts on seasonal wetlands: Poor prey performance may compromise growth of predatory salamanders. Wetlands 33:707- 715. 24. Karraker NE, Gibbs JP, Vonesh JR. 2008. Impacts of road deicing salt on the demography of vernal pool-breeding amphibians. Ecol Appl 18:724-734. 25. Mazerolle MJ, Desrochers A, Rochefort L. 2005. Landscape characteristics influence pond occupancy by frogs after accounting for detectability. Ecol Appl 15:824- 834. 92

26. Werner EE, Skelly DK, Relyea RA, Yurewicz KL. 2007. Amphibian species richness across environmental gradients. Oikos 116:1697-1712. 27. Novotny EV, Murphy D, Stefan HG. 2008. Increase of urban lake salinity by road deicing salt. Sci Total Environ 406:131-144. 28. Werner EE, Glennemeier KS. 1999. Influence of forest canopy cover on the breeding pond distributions of several amphibian species. Copeia:1-12. 29. Gosner KL. 1960. A simplified table for staging anuran embryos and larvae with notes on food. Herpetologica 16:183-190. 30. R Core Team. 2013. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 31. Scheiner SM. 2001. MANOVA: Multiple response variables and multispecies interactions. In Scheiner SM, Gurevitch J, eds, Design and Analysis of Ecological Experiments. Oxford University Press, New York, NY, USA, p. 415. 32. Petranka JW, Doyle EJ. 2010. Effects of road salts on the composition of seasonal pond communities: can the use of road salts enhance mosquito recruitment? Aquat Ecol 44:155-166. 33. Berven KA. 1990. Factors affecting population fluctuations in larval and adult stages of the wood frog (Rana sylvatica). Ecology 71:1599-1608. 34. Semlitsch RD, Scott DE, Pechmann JHK. 1988. Time and size at metamorphosis related to adult fitness in Ambystoma talpoideum. Ecology 69:184-192. 35. Burley LA, Moyer AT, Petranka JW. 2006. Density of an intraguild predator mediates feeding group size, intraguild egg predation, and intra- and interspecific competition. Oecologia 148:641-649. 36. Pechenik JA. 2006. Larval experience and latent effects - metamorphosis is not a new beginning. Integr Comp Biol 46:323-333. 37. Harrison XA, Blount JD, Inger R, Norris DR, Bearhop S. 2011. Carry-over effects as drivers of fitness differences in animals. J Anim Ecol 80:4-18. 38. Martin BT, Jager T, Nisbet RM, Preuss TG, Hammers-Wirtz M, Grimm V. 2013. Extrapolating ecotoxicological effects from individuals to populations: a generic approach based on Dynamic Energy Budget theory and individual-based modeling. Ecotoxicology 22:574-583. 39. Beck CW, Congdon JD. 2000. Effects of age and size at metamorphosis on performance and metabolic rates of southern toad, Bufo terrestris, metamorphs. Funct Ecol 14:32-38.

40. Benard MF, Middlemis Maher J. 2011. Consequences of intraspecific niche variation: Phenotypic similarity increases competition among recently metamorphosed frogs. Oecologia 166:585-592. 41. Distel CA, Boone MD. 2010. Effects of aquatic exposure to the insecticide carbaryl are species-specific across life stages and mediated by heterospecific competitors in anurans. Funct Ecol 24:1342-1352.

Chapter 3:

1. Gaston KJ, Bennie J. 2014. Demographic effects of artificial nighttime lighting on animal populations. Environ. Rev. 22:1–8.

2. Navara KJ, Nelson RJ. 2007. The dark side of light at night: Physiological, epidemiological, and ecological consequences. J. Pineal Res. 43:215–224.

3. Gaston KJ, Visser ME, Hölker F. 2015. The biological impacts of artificial light at night: The research challenge. Philos. Trans. R. Soc. B. 370:20140133.

4. Gaston KJ, Duffy JP, Gaston S, Bennie J, Davies TW. 2014. Human alteration of natural light cycles: Causes and ecological consequences. Oecologia. 176:917–931.

5. Davies TW, Coleman M, Griffith KM, Jenkins SR. 2015. Night-time lighting alters the composition of marine epifaunal communities. Biol. Lett. 11:20150080.

6. Perkin EK, Hölker F, Tockner K, Richardson JS. 2014. Artificial light as a disturbance to light-naïve streams. Freshw. Biol. 59:2235–2244.

7. Clarke JA. 1983. Moonlight’s influence on predator/prey interactions between short-eared owls (Asio flammeus) and deermice (Peromyscus maniculatus). Behav. Ecol. Sociobiol. 13:205–209.

8. Rotics S, Dayan T, Kronfeld-Schor N. 2011. Effect of artificial night lighting on temporally partitioned spiny mice. J. Mammal. 92:159–168.

9. Petren K, Bolger DT, Case TJ. 1993. Mechanisms in the competitive success of an invading sexual gecko over an asexual native. Science (80). 259:354–358.

10. Beier P. 2006. Effects of Artificial Night Lighting on Terrestrial Mammals. In Rich, C and Longcore, T, eds, Ecol. consequences Artif. night Light. Island Press, Washington, DC, USA., pp 19–42.

11. Inger R, Bennie J, Davies TW, Gaston KJ. 2014. Potential biological and ecological effects of flickering artificial light. PLoS One. 9:e98631.

12. Perry G, Buchanan BW, Fisher RN, Salmon M, Wise SE. 2008. Effects of artificial night lighting on amphibians and reptiles in urban environments. Urban Herpetol. 3:239–256.

13. Pan J, Yang Y, Yang B, Yu Y. 2014. Artificial polychromatic light affects growth and physiology in chicks. PLoS One. 9:113595.

14. Fonken LK, Workman JL, Walton JC, Weil ZM, Morris JS, Haim A, Nelson RJ. 2010. Light at night increases body mass by shifting the time of food intake. Proc. Natl. Acad. Sci. 107:18664–18669.

15. Dominoni D, Quetting M, Partecke J. 2013. Artificial light at night advances avian reproductive physiology. Proc. Biol. Sci. 280:20123017.

16. Gutierrez P, Delgado MJ, Alonso-Bedate M. 1984. Influence of photoperiod and melatonin administration on growth and metamorphosis in Discoglossus pictus larvae. Comp. Biochem. Physiol. 79A:255–260.

17. Edwards MLO, Pivorun EB. 1991. The effects of photoperiod and different dosages of melatonin on metamorphic rate and weight gain in Xenopus laevis tadpoles. Gen. Comp. Endocrinol. 81:28–38.

18. Salmon M. 2006. Protecting sea turtles from artificial night lighting at Florida’s oceanic beaches. In Rich, C and Longcore, T, eds, Ecol. Consequences Artif. Night Light. Island Press, Washington, DC, USA., pp 141–168.

19. Gauthreaux SA, Belser CG. 2006. Effects of artificial night lighting on migrating birds. Ecol. Consequences Artif. Night Light. Island Press, Washington, DC, USA., pp 67–93.

20. Frank KD. 2006. Effects of Artificial Night Lighting on Moths. Ecol. Consequences Artif. Night Light. Island Press, Washington, DC, USA., pp 305–344.

21. Yorzinski JL, Chisholm S, Byerley SD, Coy JR, Aziz A, Wolf JA, Gnerlich AC. 2015. Artificial light pollution increases nocturnal vigilance in peahens. PeerJ. 3:e1174.

22. Da Silva A, Valcu M, Kempenaers B. 2015. Light pollution alters the phenology of dawn and dusk singing in common European songbirds. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 370:20140126-.

23. Dananay KL, Krynak KL, Krynak TJ, Benard MF. 2015. Legacy of road salt: Apparent positive larval effects counteracted by negative post-metamorphic effects in wood frogs. Environ. Toxicol. Chem. 34:2417–2424.

24. Rohr JR, Raffel TR, Halstead NT, McMahon TA, Johnson SA, Boughton RK, Martin LB. 2013. Early-life exposure to a herbicide has enduring effects on pathogen- induced mortality. Proc. R. Soc. B. 280:20131502. 95

25. Beck CW, Congdon JD. 2000. Effects of age and size at metamorphosis on performance and metabolic rates of southern toad, Bufo terrestris, metamorphs. Funct. Ecol. 14:32–38.

26. Benard MF, Fordyce JA. 2003. Are induced defenses costly? Consequences of predator-induced defenses in Western toads, Bufo boreas. Ecology. 84:68–78.

27. Ohta H, Mitchell AC, McMahon DG. 2006. Constant light disrupts the developing mouse biological clock. Pediatr. Res. 60:304–308.

28. Beiswenger RE. 1977. Diel patterns of aggregative behavior in tadpoles of Bufo americanus, in relation to light and temperature. Ecology. 58:98–108.

29. Brune CR. 2013. Eastern American toad. In Pfingsten, RA, Davis, JG, Matson, TO, Lipps, GJ, Wynn, D and Armitage, BJ, eds, Amphib. Ohio. Ohio Biological Survey, Columbus, OH, pp 467–482.

30. Schulze PSC, Barreira LA, Pereira HGC, Perales JA, Varela JCS. 2014. Light emitting diodes (LEDs) applied to microalgal production. Trends Biotechnol. 32:422– 430.

31. Schubert EF, Kim JK. 2005. Solid-state light sources getting smart. Science (80). 308:1274–1278.

32. Chase JM, Burgett AA, Biro EG. 2010. Habitat isolation moderates the strength of top-down control in experimental pond food webs. Ecology. 91:637–643.

33. R Core Team. 2013. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

34. Gaston KJ, Bennie J, Davies TW, Hopkins J. 2013. The ecological impacts of nighttime light pollution: A mechanistic appraisal. Biol. Rev. 88:912–927.

35. Gaston KJ, Gaston S, Bennie J, Hopkins J. 2015. Benefits and costs of artificial nighttime lighting of the environment. Environ. Rev. 23:14–23.

36. Davies TW, Bennie J, Gaston KJ. 2012. Street lighting changes the composition of invertebrate communities. Biol. Lett. 8:764–767.

37. Binckley CA, Resetarits Jr. WJ. 2003. Functional equivalence of non-lethal effects: Generalized fish avoidance determines distribution of gray treefrog, Hyla chrysoscelis, larvae. Oikos. 102:623–629.

38. Resetarits Jr. WJ, Wilbur HM. 1989. Choice of oviposition site by Hyla chrysoscelis: Role of predators and competitors. Ecology. 70:220–228.

39. Binckley CA, Resetarits Jr. WJ. 2008. Oviposition behavior partitions aquatic landscapes along predation and nutrient gradients. Behav. Ecol. 19:552–557.

40. Resetarits Jr. WJ. 2005. Habitat selection behaviour links local and regional scales in aquatic systems. Ecol. Lett. 8:480–486.

41. Lawler SP, Morin PJ. 1993. Temporal overlap, competition, and priority effects in larval anurans. Ecology. 74:174–182.

42. Sweeney BW, Vannote RL. 1982. Population synchrony in mayflies: A predator satiation hypothesis. Evolution (N. Y). 36:810–821.

43. Ims RA. 1990. The ecology and evolution of reproductive synchrony. Trends Ecol. Evol. 5:135–140.

44. DeVito J. 2003. Metamorphic synchrony and aggregation as antipredator responses in American toads. Oikos. 103:75–80.

45. Bird BL, Branch LC, Miller DL. 2004. Effects of coastal lighting on foraging behavior of beach mice. Conserv. Biol. 18:1435–1439.

46. Scheuerell MD, Schindler DE. 2003. Diel vertical migration by juvenile sockeye salmon: Empirical evidence for the antipredation window. Ecology. 84:1713–1720.

47. Buchanan BW. 2006. Observed and potential effects of artificial night lighting on anuran amphibians. In Rich, C and Longcore, T, eds, Ecol. Consequences Artif. Night Light. Island Press, Washington, DC, USA., pp 192–220.

48. Mougeot F, Bretagnolle V. 2000. Predation risk and moonlight avoidance in nocturnal seabirds. J. Avian Biol. 31:376–386.

49. Cabrera-Guzmán E, Crossland MR, Brown GP, Shine R. 2013. Larger body size at metamorphosis enhances survival, growth and performance of young cane toads (Rhinella marina). PLoS One. 8:e70121.

50. Tarvin RD, Silva Bermúdez C, Briggs VS, Warkentin KM. 2015. Carry-over effects of size at metamorphosis in red-eyed treefrogs: Higher survival but slower growth of larger metamorphs. Biotropica. 47:218–226.

51. Berven KA. 1990. Factors affecting population fluctuations in larval and adult stages of the wood frog (Rana sylvatica). Ecology. 71:1599–1608.

52. Taylor BW, Anderson CR, Peckarsky BL. 1998. Effects of size at metamorphosis on stonefly fecundity, longevity, and reproductive success. Oecologia. 114:494–502.

53. Bradshaw WE, Holzapfel CM. 2010. What season is it anyway? Circadian tracking vs. photoperiodic anticipation in insects. J. Biol. Rhythms. 25:155–165. 97

Chapter 4:

1. Gaston KJ, Visser ME, Hölker F. 2015. The biological impacts of artificial light at night: The research challenge. Philos. Trans. R. Soc. B. 370:20140133. 2. Denver RJ. 2009. Structural and functional evolution of vertebrate neuroendocrine stress systems. Ann. N. Y. Acad. Sci. 1163:1–16. 3. Nelson RJ, Kriegsfeld LJ. 2017. An Introduction to Behavioral Endocrinology. Sinauer Associates, Inc., Sunderland, MA. 4. Denver RJ. 2009. Stress hormones mediate environment-genotype interactions during amphibian development. Gen. Comp. Endocrinol. 164:20–31. 5. Pan J, Yang Y, Yang B, Yu Y. 2014. Artificial polychromatic light affects growth and physiology in chicks. PLoS One. 9:113595. 6. Fonken LK, Workman JL, Walton JC, Weil ZM, Morris JS, Haim A, Nelson RJ. 2010. Light at night increases body mass by shifting the time of food intake. Proc. Natl. Acad. Sci. 107:18664–18669. 7. Gutierrez P, Delgado MJ, Alonso-Bedate M. 1984. Influence of photoperiod and melatonin administration on growth and metamorphosis in Discoglossus pictus larvae. Comp. Biochem. Physiol. 79A:255–260. 8. Edwards MLO, Pivorun EB. 1991. The effects of photoperiod and different dosages of melatonin on metamorphic rate and weight gain in Xenopus laevis tadpoles. Gen. Comp. Endocrinol. 81:28–38. 9. Bird BL, Branch LC, Miller DL. 2004. Effects of coastal lighting on foraging behavior of beach mice. Conserv. Biol. 18:1435–1439. 10. Lorne J, Salmon M. 2007. Effects of exposure to artificial lighting on orientation of hatchling sea turtles on the beach and in the ocean. Endanger. Species Res. 3:23–30. 11. Silva A Da, Valcu M, Kempenaers B. 2015. Light pollution alters the phenology of dawn and dusk singing in common European songbirds. Philos. Trans. R. Soc. B. 370:20140126. 12. Davies TW, Duffy JP, Bennie J, Gaston KJ. 2014. The nature, extent, and ecological implications of marine light pollution. Front. Ecol. Environ. 12:347–355. 13. Maher JM, Werner EE, Denver RJ, B PRS. 2013. Stress hormones mediate predator-induced phenotypic plasticity in amphibian tadpoles. Proc. R. Soc. B Biol. Sci. 280:1–9.

14. Ouyang JQ, de Jong M, Hau M, Visser ME, van Grunsven RHA, Spoelstra K. 2015. Stressful colours: Corticosterone concentrations in a free-living songbird vary with the spectral composition of experimental illumination. Biol. Lett. 11:20150517. 15. Lalitha R, Suthanthirarajan N, Namasivayam A. 1988. Effect of flickering light stress on certain biochemical parameters in rats. Indian J. Physiol. Pharmacol. 32:182– 186. 16. Evans JE, Smith EL, Bennett ATD, Cuthill IC, Buchanan KL. 2012. Short-term physiological and behavioural effects of high- versus low-frequency fluorescent light on captive birds. Anim. Behav. 83:25–33. 17. Smith EL, Evans JE, Párraga CA. 2005. Myoclonus induced by cathode ray tube screens and low-frequency lighting in the European starling (Sturnus vulgaris). Vet. Rec. 157:148–150. 18. Maddocks SA, Goldsmith AR, Cuthill IC. 2001. The influence of flicker rate on plasma corticosterone levels of European starlings, Sturnus vulgaris. Gen. Comp. Endocrinol. 124:315–320. 19. Stone EL, Harris S, Jones G. 2015. Impacts of artificial lighting on bats: A review of challenges and solutions. Mamm. Biol. - Zeitschrift für Säugetierkd. 80:213–219. 20. Bennie J, Davies TW, Cruse D, Inger R, Gaston KJ. 2015. Cascading effects of artificial light at night: Resource-mediated control of herbivores in a grassland ecosystem. Philos. Trans. R. Soc. B Biol. Sci. 370:20140131. 21. Davies TW, Duffy JP, Bennie J, Gaston KJ. 2014. The nature, extent, and ecological implications of marine light pollution. Front. Ecol. Environ. 12:347–355. 22. Preisser EL, Bolnick DI, Benard MF. 2005. Scared to death? The effects of intimidation and consumption in predator – prey interactions. Ecology. 86:501–509. 23. Werner EE, Peacor SD. 2003. A review of trait- mediated indirect interactions in ecological communities. Ecology. 84:1083–1100. 24. Chase JM, Abrams PA, Grover JP, Diehl S, Chesson P, Holt RD, Richards SA, Nisbet RM, Case TJ. 2002. The interaction between predation and competition: A review and synthesis. Ecol. Lett. 5:302–315. 25. Gurevitch J, Morrison JA, Hedges L V. 2000. The interaction between competition and predation: A meta‐analysis of field experiments. Am. Nat. 155:435–453. 26. Yuen SW, Bonebrake TC. 2017. Artificial night light alters nocturnal prey interception outcomes for morphologically variable spiders. PeerJ. 5:e4070. 27. Minnaar C, Boyles JG, Minnaar I a., Sole CL, McKechnie AE. 2015. Stacking the odds: Light pollution may shift the balance in an ancient predator-prey arms race. J. Appl. Ecol. 52:522–531. 99

28. Titulaer M, Spoelstra K, Lange CYMJG, Visser ME. 2012. Activity patterns during food provisioning are affected by artificial light in free living great tits (Parus major). PLoS One. 7:5–8. 29. Clarke JA. 1983. Moonlight’s influence on predator/prey interactions between short-eared owls (Asio flammeus) and deermice (Peromyscus maniculatus). Behav. Ecol. Sociobiol. 13:205–209. 30. Werner EE, Peacor SD. 2003. A review of trait- mediated indirect interactions in ecological communities. Ecology. 84:1083–1100. 31. Lima SL, Dill LM. 1990. Behavioral decisions made under the risk of predation: A review and prospectus. Can. J. Zool. 68:619–640. 32. McCollum SA, Buskirk J Van. 1996. Costs and benefits of a predator-induced polyphenism in the gray treefrog Hyla chrysoscelis. Evolution. 50:583–593. 33. Dodson SI, Havel JE. 1988. Indirect prey effects: Some morphological and life history responses of Daphnia pulex exposed to Notonecta undulata. Limnol. Oceanogr. 33:1274–1285. 34. Creel SR, Christianson D, Liley S, Winnie JA. 2007. Predation risk affects reproductive physiology and demography of elk. Science (80). 315:960. 35. Bennett AM, Longhi JN, Chin EH, Burness G, Kerr LR, Murray DL. 2016. Acute changes in whole body corticosterone in response to perceived predation risk: A mechanism for anti-predator behavior in anurans? Gen. Comp. Endocrinol. 229:62–66. 36. Resetarits Jr. WJ, Wilbur HM. 1989. Choice of oviposition site by Hyla chrysoscelis: Role of predators and competitors. Ecology. 70:220–228. 37. Crespi EJ, Denver RJ. 2004. Ontogeny of corticotropin-releasing factor effects on locomotion and foraging in the Western spadefoot toad (Spea hammondii). Horm. Behav. 46:399–410. 38. Brown JS, Kotler BP, Smith RJ, Ii WOW, Brown JS, Kotier BP, Smith RJ, Ii WOW. 1988. The effects of owl predation on the foraging behavior of heteromyid rodents. Oecologia. 76:408–415. 39. Lockard RB, Owings DH. 1974. Moon-related surface activity of bannertail (Dipodomys spectabilis) and fresno (D. Nitratoides) kangaroo rats. Anim. Behav. 22:262–273. 40. Yorzinski JL, Chisholm S, Byerley SD, Coy JR, Aziz A, Wolf JA, Gnerlich AC. 2015. Artificial light pollution increases nocturnal vigilance in peahens. PeerJ. 3:e1174. 41. Gaston KJ, Duffy JP, Gaston S, Bennie J, Davies TW. 2014. Human alteration of natural light cycles: Causes and ecological consequences. Oecologia. 176:917–931.

100

42. Gaston KJ, Bennie J. 2014. Demographic effects of artificial nighttime lighting on animal populations. Environ. Rev. 22:1–8. 43. Benard MF. 2004. Predator-Induced phenotypic plasticity in organisms with complex life histories. Annu. Rev. Ecol. Evol. Syst. 35:651–673. 44. Fraker ME, Hu F, Cuddapah V, McCollum SA, Relyea RA, Hempel J, Denver RJ. 2009. Characterization of an alarm pheromone secreted by amphibian tadpoles that induces behavioral inhibition and suppression of the neuroendocrine stress axis. Horm. Behav. 55:520–529. 45. Beiswenger RE. 1977. Diel patterns of aggregative behavior in tadpoles of Bufo americanus, in relation to light and temperature. Ecology. 58:98–108. 46. Warkentin KM. 1992. Effects of temperature and illumination on feeding rates of green frog tadpoles (Rana clamitans). Copeia. 1992:725–730. 47. Norma-Rashid Y, Saleeza SNR. 2014. Eco-friendly control of three common mosquito larvae species by Odonata nymphs. Basic Appl. Asp. Biopestic. Springer India, New Delhi, pp 235–243. 48. Skelly DK, Werner EE. 1990. Behavioral and life-historical responses of larval American toads to an Odonate predator. Ecology. 71:2313–2322. 49. Gaston KJ, Davies TW, Bennie J, Hopkins J. 2012. Reducing the ecological consequences of night-time light pollution: Options and developments. J. Appl. Ecol. 49:1256–1266. 50. Davies TW, Bennie J, Inger R, De Ibarra NH, Gaston KJ. 2013. Artificial light pollution: Are shifting spectral signatures changing the balance of species interactions? Glob. Chang. Biol. 19:1417–1423. 51. Holzman DC. 2010. What’s in a color? The unique human health effects of blue light. Environ. Health Perspect. 118:A22–A27. 52. Pawson SM, Bader MK-F. 2014. LED lighting increases the ecological impact of light pollution irrespective of color temperature. Ecol. Appl. 24:1561–1568. 53. Bedrosian TA, Galan A, Vaughn CA, Weil ZM, Nelson RJ. 2013. Light at night alters daily patterns of cortisol and clock proteins in female Siberian hamsters. J. Neuroendocrinol. 25:590–596. 54. Brüning A, Hölker F, Franke S, Preuer T, Kloas W. 2015. Spotlight on fish: Light pollution affects circadian rhythms of European perch but does not cause stress. Sci. Total Environ. 511:516–522. 55. Ohta H, Mitchell AC, McMahon DG. 2006. Constant light disrupts the developing mouse biological clock. Pediatr. Res. 60:304–308.

101

56. Moore FL, Miller LJ. 1984. Stress-induced inhibition of sexual behavior: Corticosterone inhibits courtship behaviors of a male amphibian (Taricha granulosa). Horm. Behav. 18:400–410. 57. Tsai PS, Lunden JB, Jones JT. 2003. Effects of steroid hormones on spermatogenesis and GnRH release in male leopard frogs, Rana pipiens. Gen. Comp. Endocrinol. 134:330–338. 58. Kaiser K, Devito J, Jones CG, Marentes A, Perez R, Umeh L, Weickum RM, McGovern KE, Wilson EH, Saltzman W. 2015. Reproductive and immune effects of chronic corticosterone treatment in male White’s treefrogs, Litoria caerulea. Conserv. Physiol. 3:1–8. 59. Martin LB, Hopkins WA, Mydlarz LD, Rohr JR. 2010. The effects of anthropogenic global changes on immune functions and disease resistance. Ann. N. Y. Acad. Sci. 1195:129–148. 60. Marino J a., Holland MP, Middlemis Maher J. 2014. Predators and trematode parasites jointly affect larval anuran functional traits and corticosterone levels. Oikos. 123:451–460. 61. Bennett AM, Longhi JN, Chin EH, Burness G, Kerr LR, Murray DL. 2016. Acute changes in whole body corticosterone in response to perceived predation risk: A mechanism for anti-predator behavior in anurans? Gen. Comp. Endocrinol. 229:62–66. 62. Wright ML, Duffy JL, Guertin CJ, Alves CD, Szatkowski MC, Visconti RF. 2003. Developmental and diel changes in plasma thyroxine and plasma and ocular melatonin in the larval and juvenile bullfrog, Rana catesbeiana. Gen. Comp. Endocrinol. 130:120–128. 63. Barriga C, Martin M, Tabla R, Ortega E, Rodriguez AB. 2001. Circadian rhythm of melatonin, corticosterone and phagocytosis: Effect of stress. J. Pineal Res. 30:180– 187. 64. Butte JC, Kakihana R, Noble EP. 1976. Circadian rhythm of corticosterone levels in rat brain. J. Endocrinol. 68:235–239. 65. Glennemeier KA, Denver RJ. 2002. Role for corticoids in mediating the response of Rana pipiens tadpoles to intraspecific competition. J. Exp. Zool. 292:32–40. 66. Belden LK, Kiesecker JM. 2005. Glucocorticosteroid hormone treatment of larval treefrogs increases infection by Alaria sp. trematode cercariae. J. Parasitol. 91:686–688. 67. Maher JM, Werner EE, Denver RJ, B PRS. 2013. Stress hormones mediate predator-induced phenotypic plasticity in amphibian tadpoles. Proc. R. Soc. B Biol. Sci. 280:1–9. 68. Burraco P, Duarte LJ, Gomez-Mestre I. 2013. Predator-induced physiological responses in tadpoles challenged with herbicide pollution. Curr. Zool. 59:475–484. 102

69. Wilbur HM, Fauth JE. 1990. Experimental aquatic food webs: Interactions between two predators and two prey. Am. Nat. 2:176–204. 70. Van Buskirk J. 1988. Interactive effects of dragonfly predation in experimental pond communities. Ecology. 69:857–867. 71. Relyea R. 2001. Morphological and behavioral plasticity of larval anurans in response to different predators. Ecology. 82:523–540. 72. Babbitt KJ. 2001. Behaviour and growth of southern leopard frog (Rana sphenocephala) tadpoles: Effects of food and predation risk. Can. J. Zool. 79:809–814. 73. Altwegg R. 2002. Predator-induced life-history plasticity under time constraints in pool frogs. Ecology. 83:2542–2551. 74. Peacor SD. 2002. Positive effect of predators on prey growth rate through induced modifications of prey behaviour. Ecol. Lett. 5:77–85. 75. Van Buskirk J, Schmidt BR. 2000. Predator-induced phenotypic plasticity in larval newts: Trade-offs, selection, and variation in nature. Ecology. 81:3009–3028. 76. Longcore T, Rich C. 2004. Ecological light pollution. Front. Ecol. Environ. 2:191–198. 77. Lacoeuilhe A, Machon N, Julien JF, Le Bocq A, Kerbiriou C. 2014. The influence of low intensities of light pollution on bat communities in a semi-natural context. PLoS One. 9. 78. Luttbeg B, Hammond JI, Sih A. 2009. Dragonfly larvae and tadpole frog space use games in varied light conditions. Behav. Ecol. 20:13–21.

Chapter 5:

1. Sanzo D, Hecnar SJ. 2006. Effects of road de-icing salt (NaCl) on larval wood frogs (Rana sylvatica). Environ. Pollut. 140:247–56. 2. Harless ML, Huckins CJ, Grant JB, Pypker TG. 2011. Effects of six chemical deicers on larval wood frogs (Rana sylvatica). Environ. Toxicol. Chem. 30:1637–41. 3. Collins SJ, Russell RW. 2009. Toxicity of road salt to Nova Scotia amphibians. Environ. Pollut. 157:320–4. 4. Gaston KJ, Bennie J, Davies TW, Hopkins J. 2013. The ecological impacts of nighttime light pollution: A mechanistic appraisal. Biol. Rev. 88:912–927. 5. Hölker F, Moss T, Griefahn B, Kloas W, Voigt CC, Henckel D, Hänel A, Kappeler PM, Völker S, Schwope A, Franke S, Uhrlandt D, Fischer J, Klenke R, Wolter 103

C, Tockner K. 2010. The Dark Side of Light: A transdisciplinary research agenda for light. Ecol. Soc. 15. 6. Forman RTT, Alexander LE. 1998. Roads and their major ecological effects. Annu. Rev. Ecol. Syst. 29:207–231. 7. Trombulak SC, Frissell CA. 2000. Review of ecological effects of roads on terrestrial and aquatic communities. Conserv. Biol. 14:18–30. 8. Mitchell SE, Rogers ES, Little TJ, Read AF. 2005. Host-parasite and genotype- by-environment interactions: Temperature modifies potential for selection by a sterilizing pathogen. Evolution. 59:70–80. 9. Rodewald AD, Kearns LJ, Shustack DP. 2011. Anthropogenic resource subsidies decouple predator-prey relationships. Ecol. Appl. 21:936–943. 10. Bennie J, Davies TW, Cruse D, Inger R, Gaston KJ. 2015. Cascading effects of artificial light at night: Resource-mediated control of herbivores in a grassland ecosystem. Philos. Trans. R. Soc. B Biol. Sci. 370:20140131–20140131. 11. Gaston KJ, Davies TW, Bennie J, Hopkins J. 2012. Reducing the ecological consequences of night-time light pollution: Options and developments. J. Appl. Ecol. 49:1256–1266. 12. Norris DO, Donahue S, Dores RM, Lee JK, A T, Ruth T, Woodling JD. 1999. Impaired adrenocortical response to stress by brown trout. Gen. Comp. Endocrin. 113:1– 8. 13. Romero LM. 2004. Physiological stress in ecology: Lessons from biomedical research. Trends Ecol. Evol. 19:249–255. 14. Scott GR, Sloman KA. 2004. The effects of environmental pollutants on complex fish behaviour: Integrating behavioural and physiological indicators of toxicity. Aquat. Toxicol. 68:369–392. 15. Acevedo-Whitehouse K, Duffus ALJ. 2009. Effects of environmental change on wildlife health. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 364:3429–3438. 16. Noyes PD, McElwee MK, Miller HD, Clark BW, Van Tiem LA, Walcott KC, Erwin KN, Levin ED. 2009. The toxicology of climate change: Environmental contaminants in a warming world. Environ. Int. 35:971–986. 17. Gaston KJ, Davies TW, Nedelec SL, Holt LA. 2017. Impacts of artificial light at night on biological timings. Annu. Rev. Ecol. Evol. Syst. 48:49–68. 18. Gaston KJ, Bennie J. 2014. Demographic effects of artificial nighttime lighting on animal populations. Environ. Rev. 22:1–8.

104

19. Bennie J, Davies TW, Cruse D, Inger R, Gaston KJ, Gaston KJ. 2015. Cascading effects of artificial light at night: Resource-mediated control of herbivores in a grassland ecosystem. Philos. Trans. R. Soc. B. 370:20140131. 20. Rohr JR, Raffel TR, Halstead NT, McMahon TA, Johnson SA, Boughton RK, Martin LB. 2013. Early-life exposure to a herbicide has enduring effects on pathogen- induced mortality. Proc. R. Soc. B. 280:20131502. 21. Benard MF, Fordyce JA. 2003. Are induced defenses costly? Consequences of predator-induced defenses in western toads, Bufo boreas. Ecology. 84:68–78. 22. Scott DE. 1994. The effect of larval density on adult demographic traits in Ambystoma opacum. Ecology. 75:1383–1396. 23. Vonesh J, De la Cruz O. 2002. Complex life cycles and density dependence: Assessing the contribution of egg mortality to amphibian declines. Oecologia. 133:325– 333. 24. Berven KA. 1990. Factors affecting population fluctuations in larval and adult stages of the wood frog (Rana sylvatica). Ecology. 71:1599–1608. 25. Semlitsch RD, Scott DE, Pechmann JHK. 1988. Time and size at metamorphosis related to adult fitness in Ambystoma talpoideum. Ecology. 69:184–192.

105