Notes from the Underground: Linking Microhabitat and Species Distributions of

Plethodontid Salamanders

A dissertation presented to

the faculty of

the College of Arts and Sciences of Ohio University

In partial fulfillment

of the requirements for the degree

Doctor of Philosophy

Vincent R. Farallo

April 2017

© 2017 Vincent R. Farallo. All Rights Reserved. 2

This dissertation titled

Notes from the Underground: Linking Microhabitat and Species Distributions of

Plethodontid Salamanders

by

VINCENT R. FARALLO

has been approved for

the Department of Biological Sciences

and the College of Arts and Sciences by

Donald B. Miles

Professor of Biological Science

Robert Frank

Dean, College of Arts and Sciences 3

ABSTRACT

FARALLO, VINCENT R., Ph.D., April 2017, Biological Sciences

Notes from the Underground: Linking Microhabitat and Species Distributions of

Plethodontid Salamanders

Director of Dissertation: Donald B. Miles

Environmental characteristics play a pivotal role in delineating species

distributions and, in turn, influence species richness, diversity, and interspecific

interactions. However, species exist at various spatial and temporal scales. Therefore, the scale at which environmental data is collected will influence the interpretations of how it impacts organisms. Most species primarily exist at a small spatial scale making broad- scale climate and habitat data often less biologically relevant to these organisms. Instead microhabitat and microclimate variables are often what species will directly experience, which makes the connection between macro- and micro-scale data important for answering evolutionary and ecological questions.

This dissertation quantifies microhabitat and microclimate use of plethodontid salamanders across a broad geographic area including latitudinal and elevational gradients. First, I review the importance of micro-scale environmental data and how can

bolster studies in evolutionary ecology. Next, I focus on species ability to choose specific

microhabitat and microclimates including direct comparisons to habitat and climates that

is available to captured salamanders, but remains unused. I found that salamanders were

able to choose specific environmental conditions. Furthermore, species used different

microhabitat and microclimates, including closely related species. To further refine our 4 knowledge of microhabitat and microclimate use, I include a study monitoring thermal and hydric properties of various microhabitats in the Great Smoky Mountains National

Park along an elevational gradient. I found that they chose cooler and wetter microhabitats and that they chose microhabitats that maintain their hydric state even between high and low elevation sites, indicating that through behavioral compensation they may be able to extend their ranges beyond what broad-scale climate data predict.

Finally, I tested for phylogenetic signal of these small-scale variables and found as whole there is no indication of phylogenetic signal. However, two individual variables relative humidity and soil temperature, have significant phylogenetic signal which appears driven by differences in the Plethodon cinereus and Plethodon glutinosus species groups.

Together, these results provide a substantial contribution to our knowledge of the evolutionary ecology of plethodontid salamanders, and highlight the importance of incorporating small-scale environmental data, especially when assessing the impacts of climate change.

5

DEDICATION

For my family, especially my grandfathers, Vincent Farallo and Robert E. Brogan, who

taught me a love of learning and nature.

6

ACKNOWLEDGMENTS

This dissertation was possible due in thanks too many people. My advisor, Donald

Miles, who offered academic guidance throughout my dissertation. My dissertation committee, Shawn Kuchta, Willem Roosenburg, and James Dyer have also provided mentorship. Thank you to the Ohio Center for Ecology & Evolutionary Studies (OCEES) graduate students, postdocs, and faculty for the countless hours of discussion and feedback, especially current and former members of the Miles laboratory. The field work required to complete this dissertation would not be possible without help from numerous field assistants including William Ternes, Rebecca Wier, Celeste Wheeler, Morgan

Etheridge, Kaili Boarman, Jessica Mace, and Courtney Thomas. Furthermore, the United

States National Parks and National Forest Service allowed me access to field sites. A special thanks to Paul Super, the research coordinator of the Great Smoky Mountains

National Park, who was an invaluable help throughout my dissertation. Financial support for this work was provided by Ohio University Student Enhancement Award, Ohio

University GSS Original Work Grants, Sigma Xi GIAR, Explorer’s Club Exploration

Fund Grants, Society for the Study of Evolution Rosemary Grant Award, and an

American Philosophical Society Lewis and Clark Fund Award. I was also able to extend my field work in thanks to the OCEES and Ohio University Graduate College

Fellowships. 7

TABLE OF CONTENTS

Page

Abstract ...... 3 Dedication ...... 5 Acknowledgments...... 6 List of Tables ...... 10 List of Figures ...... 11 Chapter 1: Seeing the Forest and Missing the Trees: The Role of Scale in the Ecology and Evolution of Species Distributions ...... 13 Introduction ...... 13 The Problem of Scale ...... 16 Spatial-scale and Evolutionary History ...... 30 Case Study: Examining the Habitat of Microendemics ...... 31 Conclusions ...... 38 Figures ...... 39 Chapter 2: The Importance of Microhabitat: A Comparison of Two Microendemic Plethodon Species to the Widespread P. cinereus ...... 44 Introduction ...... 44 Plethodontid Salamanders as a Model System ...... 45 Material and Methods ...... 51 Field Sites ...... 51 Microhabitat Measurements ...... 51 Statistical Analyses ...... 53 Results ...... 54 Variation in Microhabitat Traits ...... 54 GLMM Analysis ...... 55 Temporal Change of Microhabitats ...... 56 Discussion ...... 57 Tables ...... 62 Figures ...... 63 8

Chapter 3: Small-scale Habitat Use of Plethodontid Salamanders along the Appalachian Mountains ...... 68 Introduction ...... 68 Materials and Methods ...... 75 Field Sites ...... 75 Microhabitat Data Collection ...... 75 Statistical Analyses ...... 78 Results ...... 81 Discussion ...... 84 Tables ...... 90 Figures ...... 92 Chapter 4: The Bogert Effect Revisited: Will Behavioral Compensation Save Species in a Changing Climate?...... 95 Introduction ...... 95 Salamanders as a Model System ...... 98 The Interaction Between Thermal and Hydric Ecology ...... 99 Salamanders and the Bogert Effect ...... 100 Materials and Methods ...... 102 Field Sites ...... 102 Agar Model Salamanders ...... 103 Microhabitat Plots ...... 104 Salamander Surveys ...... 105 EWL of Salamanders ...... 106 Statistical Analyses ...... 107 Results ...... 110 Discussion ...... 112 Tables ...... 119 Figures ...... 123 Chapter 5: Phylogenetic Niche Conservatism in Plethodontid Salamanders: Taking a Small-scale Approach ...... 130 Introduction ...... 130 Climate, Habitat, and Spatial Scale of Phylogenetic Niche Conservatism ...... 133 9

Plethodontid Salamanders as a Model System ...... 135 Phylogenetic Niche Conservatism and Plethodontid Salamanders...... 136 Materials and Methods ...... 137 Microhabitat Data Collection ...... 137 Phylogenetic Tree and Phylogenetic Signal of Microhabitats and Microclimates . 139 Phylogenetic Clumping or Overdispersion ...... 140 Results ...... 141 Discussion ...... 142 Tables ...... 146 Figures ...... 148 Chapter 6: Conclusions ...... 149 References ...... 154 Appendix 1: Chapter 3 Field Site List ...... 174 Appendix 2: Chapter 3 Species List ...... 177

10

LIST OF TABLES

Page

Table 1: Correlation of environmental variables to the two NMDS axis ...... 62

Table 2: Results of generalized linear mixed models ...... 62

Table 3: Species captured during this study ...... 90

Table 4: Results of the conditional logistic regression for microclimates and

microhabitats ...... 91

Table 5: Correlation of the environmental variables to the first two NMDS axes...... 91

Table 6: List of species captured during surveys ...... 119

Table 7: Results of the three linear mixed effects models ...... 120

Table 8: Results of the two generalized linear mixed effects models ...... 121

Table 9: Results of the ANCOVA comparing dehydration and rehydration rates ...... 122

Table 10: Values of Lambda and Blomberg’s K for all environmental variables ...... 146

Table 11: Values of NRI and NTI calculated from elevation and latitude categories .... 147

11

LIST OF FIGURES

Page

Figure 1: Modified Stommel diagram...... 39

Figure 2: Diagram of environmental characteristics and their interactions ...... 40

Figure 3: Temperature fluctuation for five microhabitats over two months ...... 41

Figure 4: Range maps and Maximum Entropy models ...... 42

Figure 5: Results of the NMDS analysis...... 43

Figure 6: Depiction of habitat selection on evolutionary change ...... 63

Figure 7: Distributions maps of Plethodon cinereus, P. hubrichti, and P. sherando ...... 64

Figure 8: Results of NMDS analysis ...... 65

Figure 9: Mean values of microhabitat variables at presence and absence sites ...... 66

Figure 10: Temporal fluctuation in temperature recorded at five different microhabitats

over a two month period ...... 67

Figure 11: Location of field sites...... 92

Figure 12: Results of the NMDS analyses...... 94

Figure 13: Microhabitat variables mapped on the phylogenetic tree...... 94

Figure 14: A photo of an agar model salamander ...... 123

Figure 15: Least square means from a linear mixed effects model with mean air

temperature as a response variable ...... 124

Figure 16: Least-squared means from a linear mixed effects model using temperature

amplitude as a response variable ...... 125 12

Figure 17: Least squared means from a linear mixed effects model using water loss rate as the response variable ...... 126

Figure 18: Least square means from the generalized linear mixed effect model of

salamander occurrence ...... 127

Figure 19: Least square means from the linear mixed effect model of temperature and

water loss rates ...... 128

Figure 20: Mean dehydration and rehydration rates ...... 129

Figure 21: Microhabitat variables mapped on the phylogenetic tree ...... 148 13

CHAPTER 1: SEEING THE FOREST AND MISSING THE TREES: THE ROLE OF

SCALE IN THE ECOLOGY AND EVOLUTION OF SPECIES DISTRIBUTIONS

Introduction

Delineating the factors that structure the geographic distribution and abundance of species is a long standing goal of ecology (Brown 1984). Knowledge of species distributions allows us to quantify aspects of species richness and diversity. Both topics are currently critically important as dramatic levels of anthropogenic changes spurred by activities such as habitat destruction and climate change continue to occur. Generating estimates of predicted species distributions and areas of high species richness provides the opportunity to mitigate the impacts from multiple perturbations (Hooper et al. 2005).

However, species distributions and species richness are complex issues rooted in determining the appropriate scale at which these topics should be addressed. Often the scale that is used is influenced by data availability rather than biological relevance.

Macro-scale environmental data are frequently used because of wide availability and utility for creating inferences about a species or groups of species geographic range. In contrast, the analysis of micro-scale environmental data are often at a small geographic scale and therefore less informative for making broad inferences regarding a species distribution or combining with data on other species to assess environmental drivers of high species richness. Ultimately the issue of scale and species richness can be broken into four questions: 1) do local processes influence local patterns, 2) do local processes influence regional patterns, 3) do regional processes influence local patterns, and 4) do regional process influence regional patterns. Local species richness is partially dictated 14 by the regional species richness, as species present locally are also part of the regional pool by default. However, abiotic and biotic characteristics of the local environment may preclude species from the regional pool from occurring locally (Belmaker and Jetz 2012;

Ricklefs 1989). In this manuscript, I focus on the local environment and why this is a critical component of understanding how species distributions are maintained and this will ultimately impact local species richness. More importantly, as local-scale processes are influenced by anthropogenic alterations, local extirpations may occur, which for some species with small geographic ranges may mean extinction, which will also have regional level impacts.

The impact that local-scale processes can have on regional processes may also be limited by evolutionary history. If the physiology or ecology of a species is a reflection of ancestry (e.g,, phylogenetic constraint), which is also referred to as phylogenetic niche conservatism (Holt and Gaines 1992; Peterson et al. 1999; Wiens and Graham 2005), then the prevailing local environments will have a limited contribution to species distributions. However, if species are not constrained by their evolutionary history and can locally adapt to unique conditions within a regional pool, then local-scale interactions will play a role in regional processes. Phylogenetic niche conservatism (PNC) has become of increasing interest to ecologist and has therefore been discussed in the literature recently (Losos 2008a; Losos 2008b; Wiens 2008; Wiens et al. 2010; Wiens and Graham 2005). One point of contention is determining evidence for PNC. Many studies use a significant phylogenetic signal as evidence supporting PNC. However as pointed out by Losos (2008a; 2008b), genetic drift or random fluctuation in natural 15 selection may also produce patterns of trait variation that is consistent with phylogenetic signal. Instead PNC needs to be reserved for when traits differ significantly from this expectation and demonstrating this is more complicated than simply documenting phylogenetic signal. Losos (2008a) suggests that PNC should be documented through the identification of old clades which have remained in an ecological stasis or through statistically demonstrating that ecological similarity is greater than is expected through phylogenetic signal alone. This distinction is important because verifying that a phenotypic trait is phylogenetically conserved at a macro-scale may not reflex micro- scale pattern. It is also important to consider both phylogenetic and spatial scale as these may impact the detection of PNC. In contrast to PNC, local adaption would amplify the impacts of local-scale process. Local adaptation would also be difficult to detect at a broad spatial scale, especially for organisms that have small home ranges, and may even be ignored if PNC is thought to be occurring. Local adaptation is possible when there is a heterogeneous environment that opens a door for section to occur on traits within the variable environment. The degree to which local adaption occurs will be dependent on the strength of selection and gene flow (Kawecki and Ebert 2004). One of the best methods to test for local adaptation is to use transplant experiments (Blanquart et al.

2013; Niewiarowski and Roosenburg 1993). However, a first step would be assessing microhabitat and microclimate variation and how it is used by species. If there is sufficient heterogeneity at the micro-scale, the potential for local adaption exists. Both

PNC and local adaption will be dependent on the scale used in analyses, and therefore it 16 is critical to consider the often-overlooked influence of micro-scale environmental characteristics.

The Problem of Scale

Processes Operating at a Local-scale: Microhabitat and Microclimate

The concept of scale is a complex issue that has relevance throughout the sciences. Within the field of biology scale has been discussed in the literature for decades

(Levin 1992; Levin 1993; O'Neil and King 1998; Schumm and Lichty 1965; Stommel

1963; Wiens 1989; Wiens et al. 1986; Wilbanks and Kates 1999); furthermore it is especially important when assessing species distributions (McCreadie and Adler 1998).

As this topic has already been discussed in detail, including numerous reviews, the goal of this manuscript is not to discuss the concept of scale at length, but rather to highlight important aspects of scale focused on the interconnection between micro- and macro- scale habitat and climate. Specifically, how understanding the role of micro-scale habitats and climates will be critical in assessing how organisms will react to habitat alterations especially as a result of climate change by supplementing macro-scale habitat and climate data. As background to the information provided in this manuscript I provide a brief overview of the terminology and key points of contention associated with discussions of scale in an ecological context. The term scale itself is often misused which complicates discussions. As was discussed in the first chapter of Ecological Scale (O'Neil and King

1998), the term scale specifically refers to dimensions, which for the purposes of this manuscript would be in space or time. Often when the term scale is used the term “level” would be more appropriate, which refers to a position within a hierarchically organized 17 group. These groups can include universally recognized levels that are pertinent to all of ecology, e.g. population level. However, some levels may be specific to certain systems in terms of not only the scale that they occur but also in biological relevance. The term microhabitat or microclimate can refer to very different scales and degrees of importance depending on what organisms are under consideration. For example, microhabitat for lizards may include spatial variation through the vertical use of vegetation (Adolph

1990), temporal variation in microhabitat use (Simon and Middendorf 1976), variation in habitat use based on the presence of predators (Downes 2001). These variations in microhabitat use are typically the result of foraging for food, avoiding predators, maintaining body temperature, and performing reproductive activities. However, considering a larger animal such as a Savanna elephant (Loxodonta africana), although it is possible to technically quantify some aspect of microhabitat that they utilize, ultimately they are using the environment at broader scales, often making decisions on habitat use at the habitat or even landscape level (Shrader et al. 2012). In this case, either microhabitat is not important, or the scale at which the data is collected needs to be drastically revised.

In terms of scientific research, this essentially becomes semantics, as the definitions of levels are typically considered because they highlight the scale at which data are collected to ensure biological relevance to research questions. Therefore, attempting to study microhabitat for a Savanna elephant would be comparable to studying at the habitat or landscape level for many other organisms. In contrast, macrohabitat data is typically collected at a relatively consistent scale. Data generated from weather stations such as

WorldClim are typically at resolutions as high as 1 km2. Although macrohabitat variables 18 can be used at lower resolutions, this is not normally associated with taxa level biological differences, but rather on the geographic scale of the question being addressed. For these reasons, I think it also is important to clarify what is meant by micro-scale habitat and climatic variables. When trying to apply this term in a general framework the temporal and spatial scales need to be standardized. A modified version of a figure from Stommel

(1963) provides a good way to explore possible definitions of micro-scale environmental characteristics for the context of this manuscript (Fig. 1). This modified figure presents two axes, time and space, both on a log scale. A third axis is not included but would be used to represent some measure of biological or geophysical variation. Instead I have highlighted areas where high levels of variation would be expected to occur for typical climate measurements, however this could easily be modified to include habitat measurements as well. For example, precipitation will vary more as spatial scale increases, variation within a 1 cm2 will be non-existent, but may be substantial over

10000 km2. Temporally precipitation is expected to vary most at the month scale, as

monthly precipitation tends to vary in seasonal temperate habitat, and tropical habitats

often also have a rainy season. However, within both space and time variation in

precipitation will be dependent on the specific point being quantified. For example, if one

were to examine monthly precipitation in the Atacama Desert in Chile at a 100 km2 resolution they would except to see minimal variation as the yearly rainfall is typically 15 mm, and even at a yearly or century scale variation in precipitation would be low.

Assuming a seasonal temperate location, animal behaviors would be expected to vary, at least in part, with variation in precipitation, as this variation in rainfall will have an 19 impact on the habitat and environmental conditions. However, precipitation will impact hydric conditions available to forest floor organisms via moisture in leaf litter and the soil. Unlike precipitation, variation of leaf litter and soil moisture will occur at smaller spatial and temporal scales as it will be further effected by temperature and structural aspects of the habitat such as vegetation and soil type. These differences between seemingly interconnected environmental characteristics are why addressing microhabitat variables can be so important and why defining what a microhabitat or microclimate entails is critical comparing between studies.

One approach for defining these micro-scale environmental characteristics is to set a spatial and temporal scale, which would be consistent across all organisms. For this definition, microhabitat and microclimate would be strictly defined by scale. In this case, microhabitat would virtually never be relevant to large organisms or when discussing the long-term ecology of a system. Another approach is for microhabitat to refer to the habitat directly experienced by an organism. This definition is more complicated as the term loses consistency. Microhabitat of a small insect would be dramatically different than the microhabitat of a large mammal. However, this definition makes the most biological sense. The prefix micro- means very small, but this is relative. Very small- scale habitat is going to be dependent on the organism using that habitat or on the questions that are being addressed. One complication with this definition is that when developing research questions researchers inherently collect data at some defined scale(s) and the scale(s) they chose will have an impact on how results are interpreted. For these reasons, I think both definitions should be combined, 1) is it at a small spatial scale (Fig. 20

1), and 2) does it directly impact the study organism? If the answer to both questions is yes, then it is appropriate to us the term microhabitat.

Microhabitat, Microclimate, and the Ecological Niche

When thinking about research that incorporates microhabitat, the history of ecology needs to be considered, as the onset of ecological thinking is what will spur questions that necessitate the consideration of microhabitat importance. Ecology started to become a prominent field of study in the early 20th century with the work of Clements

and Gleason who both offered contrasting views on ecological succession (Clements

1916; Gleason 1926). At this time Joseph Grinnell (1917) introduced ecology to the

concept of the niche and began thinking about habitat requirements, while Charles Elton

(1927) focused on biotic interactions in his interpretation of the niche. Alfred Lotka and

Vito Volterra developed models of population dynamics (Lotka 1925; Volterra 1926) a

key component in understanding how species persist and interact. Finally, G. Evelyn

Hutchinson (1957) expanded on the niche concept, combining both biotic and abiotic

factors into an n-dimensional hyper volume; he has even been referred to as the father of

modern ecology (Slack and Wilson 2010). By the middle of the 20th century a base of

ecological research has been done which opens the door for thinking about specific

aspect of species ecology.

The term microhabitat was first used in biological literature as late as the early

1930’s (Allee 1931). Earlier mentions of the term or discussions of the general concept

are certainly likely, either way researchers have been thinking about it since the early

stages of ecological research. In these early mentions of microhabitat, the term seems to 21 be taken at face value, as no definition is provided and it appears the authors simply mean to imply small-scale habitat (Allee 1931; Ide 1935; Patterson 1940). During this time, studies used microhabitat as a potential explanation for why coarse grain habitat data do not fit with expectations, but there was little to no direct study of microhabitat variables.

However, in the early 20th century there are numerous researcher who developed methods, intended to quantifying plant communities (Braun-Blanquet 1932; Daubenmire

1959), that eventually contribute to quantifying microhabitat use. Vegetation is an obvious aspect of microhabitat for many animal species, so improvements in methods have made collecting these types of data more accessible.

In the late 1960’s through the 1980’s studies began to focus explicitly on individual selection of microhabitats (Baust 1976; Douglass 1976; Hughes 1966; Lemen and Rosenzweig 1978; Loman 1978; McShea and Francq 1984; Murua and Gonzalez

1982; Simon and Middendorf 1976; Waldorf 1976). This includes a study by Hughes

(1966) which showed that mayfly nymphs selected microhabitats based on habitat lighting conditions. Studies such as this demonstrate the connection of an intermediate- scale habitat variable, which would be influenced by canopy cover and water condition, to microhabitat use. Douglass (1976) found that a species of vole (Microtus montanus) used different microhabitats when in the presence of the sympatric M. pennsylanicus.

Through this time period most studies maintain this theme of using microhabitat in a general context, primarily considering what types of microhabitat that species are using.

These studies are considering questions at a small-scale but they are not always including small-scale variables that allow us to quantify differences between microhabitats. 22

Understanding differences between microhabitats is ideal because the aspects of the microhabitat allow species to persist within these environments can then be quantified.

This trend continues, with most studies that include information on microhabitat typically addressing what microhabitats are being used, but do not quantifying detailed characteristics of the microhabitats themselves. There are obviously exceptions to this trend, especially when considering organisms that are sensitive to their surrounding environments such as amphibians and invertebrates. Researchers began to consider the importance of specific small-scale environmental variables such as Keen (1984) who demonstrated the importance of soil moisture on the ecology of plethodontid salamanders. As invertebrates also typically use habitat at a small-scale more studies emerge to assess the impacts of change on their populations. For example, arthropod abundance decreased in areas of timber harvest when organic matter was low as it resulted in temperatures in the top layer of soil increasing to lethal levels due to canopy removal (Seastedt and Crossley 1981).

For recent examples of studies incorporating microhabitat data I will focus on forest floor animals, primarily plethodontid salamanders because they are highly susceptible to environmental change because they use cutaneous respiration making them susceptible to changes in moisture and temperature, both key components of microhabitat that will be impacted by a wide array of other habitat characteristics (Fig. 2).

Plethodontid salamanders are also an ideal group for monitoring impacts on biodiversity as a whole (Welsh and Droege 2001). An increase in microhabitat research may have partially been spurred by studies documenting the impacts of timber harvests, especially 23 on amphibian populations (Ash and Bruce 1994; Bury 1983; Corn and Bury 1989;

Petranka et al. 1994). Obviously, studies of microhabitat use are also heavily dependent on understanding an organism’s physiology which dictates how microhabitat conditions will impact performance and ultimately survivorship. Important early physiological work was conducted by Brattstrom (1963) and Spotila (1972) who studied thermal and hydric requirements of amphibians which provides a basis for understanding the impact microhabitat conditions can have on activity time. As our knowledge of physiological requirements increased and more studies addressed the negative impacts of habitat alteration on populations it became apparent that understanding the small-scale details behind populations declines would be the best way to mediate these population reductions. Recent studies focusing on plethodontid salamanders have shown that microhabitat use can be impacted by invasive species (Ziemba et al. 2016), macro-scale environmental characteristics such as precipitation (O'Donnell et al. 2014), and that microhabitat use can vary between closely related species (Farallo and Miles 2016).

These studies highlight the importance of microhabitat, but do not explicitly address how microhabitat and macrohabitat characteristics will differ and consequently how these differences will impact organisms.

One of the best examples of the relevance of microhabitat in addressing questions at appropriate spatial scale and specifically focusing on the differences in temperature based on habitat characteristics is provided by the recent study of Suggitt et al. (2011).

One of the key points they make is that the majority of organisms on the planet occur at small-scales (< 1 m2), macro-level analyses typically use data at a resolution of a km2 or 24 greater, especially when trying to assess the impacts of climate change. They found that temperatures varied greatly based on the habitat (grassland, heathland, and deciduous forest) and height of measurements (0 – 1.6 m). Specifically, maxima and minima varied, with a difference in temperature range of up to 10 °C. Farallo and Miles (2016) showed an even greater contrast in temperature between above and below ground microhabitats with temperatures at 50 cm above ground spanning 29.6 °C compared to 30 cm below ground spanning only 5 °C. Similarly, microhabitat temperatures and hydric conditions also varied when comparing cover object types (Chapter 4). There is even noticeable variation in temperature as a result of heterogeneous microtopography that can mimic differences in temperature seen over large elevational and latitudinal gradients (Scherrer and Körner 2010). Variation in small-scale habitat can provide a refuge from ambient conditions that can impact whether an organism, within small-spatial scales, can persist.

This small-scale environmental variation is also, at least in part, decoupled from macro- scale environmental conditions. Species are even able to buffer against yearly variation in temperatures (Suggitt et al. 2012) by using different habitat types, and even buffer against daily thermal and hydric variation through changes in activity time and microhabitat use

(Chapter 4). Intermediate habitat structure (Fig. 2) can impact the variation in thermal and hydric microclimates. Microhabitats, as a form of microrefugia, have also been suggested as a mechanism for populations of species to persist in otherwise inhospitable climatic regions both currently (Hampe and Jump 2011), and during previous climatic shifts (Dobrowski 2010). 25

Scaling from Microhabitat to Macrohabitat Data

Quantifying habitat is typically done at a specific spatial scale. Within this specific scale either as many aspects of habitat as possible are incorporated or specific habitat characteristics are isolated while keeping others constant. Typically, the former is a method for making climatic niche models or other methods of assessing habitat use in the field. Discussing the use of macrohabitat data is important, because often they are used in lieu of microhabitat data as they are easier to collect over larger spatial scales.

When considering microhabitat data, obviously if one were to make a model that only included a single variable, such as precipitation, it would be ignoring other aspects of the environment that would influence the impacts of precipitation levels. High temperatures in one location may result in greater evaporation rates resulting in less water available for animals compared other locations with cooler temperatures even if the warmer location experienced greater precipitation rates. For this reason, when broad-scale environmental data are used, a wide range of biologically relevant variables are incorporated.

Techniques are available that account for missing values, non-linear relationships, and interactions which allows for the inclusion of a maximum number of biologically relevant environmental variables when making comparisons between species (De'Ath 2002). Even when correlative ecological niche modeling was first becoming readily available to researchers in 2007, models still incorporated numerous environmental variables in model creation (Arif et al. 2007; Chen et al. 2007). However, using multiple environmental variables does not automatically equate with a rigorous assessment of a species niche. Rodda et al. (2009) attempted to predict areas of potential invasion by the 26

Burmese Python (Python molurus bivittatus) in the United States using primarily mean monthly temperature and precipitation. These data were used specifically because additional data were not available for the snake’s native range in Southeast Asia.

However, their results seem to indicate that much more is involved than simply temperature and precipitation as they predicted suitable habitat in areas that are clearly not biologically appropriate for Burmese Pythons, such as west Texas. Again, in 2009, the data available were more limited than it is today, but a study like this could have benefited from incorporating knowledge of the species microhabitat use. Recent advancements in computing technology along with a wealth of environmental data that is now freely available provides new challenges. Specifically, increasingly complex models that account for a multitude of interactions with environmental gradients can underperform less complex models unless there is a specific ecological justification for how the model is constructed (Bell and Schlaepfer 2016).

This is also true when assessing differences between habitat use at small-scales, but studies less frequently incorporate a suite of small-scale environmental variables, but rather at small-scales, microhabitats are generalized by some characteristic of the habitat.

Categories of microhabitats are certainly informative depending on the questions being asked such as determining whether a species uses different microhabitats to avoid predators (Holomuzki 1986; Kiesecker and Blaustein 1998; Longland and Price 1991;

Main 1987). These types of studies, as mentioned earlier, use microhabitats to explain other phenomena of species ecology. However, the environmental differences between microhabitats can directly impact an organism’s physiology, behavior, and even 27 morphology over evolutionary time, especially when interactions with specific microhabitats are consistent for a species, or populations within a species. Therefore, a downside to these generalizations is that specific characteristics of the microhabitats are lost and that means it is difficult to infer what aspects of microhabitat directly impact the organism.

Microhabitat  Macrohabitat

Microhabitats also can vary based on a multitude of macrohabitat characteristics

such as differences in canopy cover, soil type, climatic conditions, and many others.

Essentially microhabitats are molded by an array of broader-scale environmental

characteristics that filter down from macro- to micro-scale (Fig. 2). The use of micro-

scale variables should include data on the habitat that directly impacts an organism, but

these variables must be used appropriately and in conjunction with other biologically

relevant habitat characteristics. Furthermore, the interactions between the macro- to

micro-scale environmental variables are important to consider during studies of all spatial

scales as these interactions impact the interpretation of the importance of certain

environmental variables. For example, one would assume precipitation is a critical

variable to consider when assessing plethodontid salamander ecology. This group of

salamanders are lungless and rely on cutaneous respiration making temperature and

moisture critically important to their survival. Therefore, it is important to consider precipitation when assessing habitat for this group. However, there are other aspects of

the environment that will influence the available moisture such as slope, vegetative cover,

and soil type. These are variables that will influence the temperature of the soil, drainage, 28 and how much water can be stored and therefore will have a tremendous impact on how much the precipitation is able to impact organisms living at ground level. This could mean a difference between a one-time benefit as precipitation occurs to an extended benefit that last days with moisture remaining available in the soil, leaf litter, etc. This can be circumvented by including these variables in the model, although these data are not as widely available as typical climate data.

One way to improve niche models is to incorporate specific aspects of an organism’s biology (Leibold 1995). This mechanistic approach incorporates macrohabitat data in conjunction with biophysical limitations of the organisms (Kearney and Porter

2009). These biophysical measurements will vary based on the study organism, as an example, a critical biophysical component for amphibians would be evaporative heat/water exchange. These biophysical measurements, based on an organism’s physiology under various conditions, allows models to predict when an organism can be active based on environmental conditions at given points of space and time. These correlative and mechanistic niche models can then be used to assess species ranges

(Guisan and Zimmermann 2000; Phillips et al. 2006), potential for species invasion

(Peterson 2003), and predict impacts of climate change on a species distribution

(Heikkinen et al. 2006; Pearson and Dawson 2003). Both methods have positive and negative attributes, and have advantages when addressing specific questions. Specifically, correlative models perform well when assessing current species distributions and they do not require detailed physiological data to generate. However, mechanistic models outperform correlative models when predicting invasions (Kumar et al. 2014) and the 29 impact of climate change (Morin and Thuiller 2009) as they incorporate physiological tolerances which may not be realized in correlative approaches (e.g. ranges may be limited by biotic factors that will change as climate changes). One major limitation of both is the scale at which the models are built. Both correlative and mechanistic models typically use macro-scale environmental variables and at low resolutions (1 km2 or greater) and are also based on weather stations typically 1-meter above ground level.

Microclim data (Kearney et al. 2014) provides estimates of a variety of environmental variables at a finer scale temporal resolution (hourly estimates for each month) for a variety of microclimatic conditions including different shade levels and substrate types.

However, at a global scale, these data are presented at 15 km2 resolution, which ignores

variation in available microhabitat at the scale of an organism. For this reason, these data

are fantastic for studying a species that exists in homogeneous habitat, or for species that

only use a single habitat when active. However, even for these questions the Microclim

data will still fail to account for fine-scale habitat variation that can be critically

important for a species' persistence in an area, or that allows for increased activity time.

Ultimately, even the best data and models available cannot currently account for

small-scale microhabitat. Species have been shown to utilize this variation in

microhabitat. Therefore, incorporating microhabitat use is of utmost importance for the

continued pursuit of understanding ecological and evolutionary relationships of species

distributions and diversity. Until information on microhabitats can be directly

incorporated within global scale analyses, understanding microhabitat use will be critical

for supplementing these macro-scale analyses. 30

Spatial-scale and Evolutionary History

Phylogenetic scale is clearly relevant for assessing the evolutionary history of ecology, including when testing for patterns of phylogenetic niche conservatism (Losos

2008a; Losos 2008b). However, spatial scale may also play an important role in interpreting the interactions of ecology and phylogeny. Pyron et al. (2015) focus on broad-scale environmental variables in their review of PNC and the role it may play in ecological speciation; but, they mention that PNC also may occur at fine-scale biotic interactions tied to the Eltonian niche. However, to my knowledge there are no studies that explicitly discuss the potential for PNC to vary across spatial scales. Phylogenetic constraints have been proposed as potential catalyst for extinction in the light of climate change (Bernardo and Spotila 2006) with species unable to adapt to increasing temperatures. One potential complication with this scenario is that species may be capable of behaviorally regulating the conditions they experience at a micro-scale, allowing them to buffer themselves from ambient conditions (Bogert 1959; Huey et al.

2003). In this scenario, species may be able to mediate climate change for a period of time before ultimately losing the battle due to a lack of available microhabitats that adequately provide buffering from ambient conditions. Microhabitats offer temporal buffering from broad-scale climatic conditions. Over enough time, microhabitats will increase or decrease in temperature or hydric state coinciding with ambient conditions, but over short periods of time microhabitats can offer refugia (Farallo and Miles 2016;

Fig. 3). Understanding the connection between micro- and macro-scale environmental variables will delineate what questions can be addressed with which scale of data. 31

In plethodontid salamanders PNC of broad-scale climatic niches has been proposed to promote speciation (Kozak and Wiens 2006) and drive elevational patterns of species diversity (Kozak and Wiens 2010). How might these studies differ if micro-scale environmental data were used? Using broad-scale variables, parapatric sister taxa were found in similar climatic niches, separated by inhospitable climates (Kozak and Wiens

2006) and this very well may be cause of speciation in this group, at least in part.

However, divergence in the use of micro-scale environmental conditions could also be a mechanism for speciation, especially through micro-scale allopatry, where micro-scale conditions separate species within otherwise homogeneous climates and this would be lost when using broad-scale climate data. Similarly, as PNC is thought to drive elevational patterns, divergence in microhabitat and microclimate could alter those interpretations. An inability for species to move out of their climatic niche, and therefore elevational band apparently results in high species richness in mid-elevation localities.

However, if species are able to use microhabitats to extend their range past areas suitable to them based on climatic data alone, then low and high elevation species would be found at mid-elevations more frequently. Again, these findings would not be interpretable from typical climatic data.

Case Study: Examining the Habitat of Microendemics

Appropriate scale can be variable even among closely related species, including species that appear to use similar habitats. Salamanders in the family Plethodontidae are a critical component of forest floor ecosystems, especially in the eastern United States where there is relatively high species diversity. The range size of Plethodontid species 32 from this region very greatly, from the widespread Plethodon cinereus which occurs from southeastern Canada, west to Minnesota, USA, and south to North Carolina, USA, spanning approximately 1.8 million km2, to species with extremely limited ranges such as

Plethodon sherando and Plethodon hubrichti, existing respectively within ~36 km2 and

49 km2 of central Virginia, USA. Plethodon cinereus and P. sherando are phenotypically similar and only have slight differences in morphology, and P. hubrichti is of similar size and uses similar habitat as both P. cinereus and P. sherando, yet there is an exponential difference in range size between P. cinereus and these two microendemics.

There are many other Plethodontid species that have small range sizes which are often restricted to specific mountain tops. Being restricted to mountain tops limits dispersal capabilities, as range shifts or expansions often requires moving into often inhospitable lowland habitat, which today is often fragmented, further restricting potential dispersal. Arif et al. (2007) examined parapatry between P. cinereus and P. hubrichti using a variety of methods including climate modeling, 2D morphometrics of head shape, and behavioral trials. The results from their study suggest that P. cinereus is restricted from within the range of P. hubrichti through aggressive behaviors by P. hubrichti. Whereas, P. hubrichti appears to be restricted by abiotic climatic conditions.

However, nothing about the physiology of P. hubrichti indicates that it would be so tightly restricted based solely on climatic conditions compared to P. cinereus as the

CTmax and CTmin are both are separated from the other species by less than 0.5 °C

(Markle 2015). Also, as correlative species distribution models are traditionally used,

they do not indicate where the species are capable of existing, but rather where they are 33 likely to exist based on their current distribution. Therefore, making assumptions regarding the importance of climate may impact the ability to mitigate impacts on populations as climate changes occurs. It is importance to consider other possibilities, especially habitat at a smaller scale as this will be the environment that the salamanders directly experience.

Plethodontid salamanders utilize a variety of microhabitats that buffer them from ambient conditions which can complicate assessments of habitat use. This is where understanding scale becomes critical, especially when attempting to assess how species will be impacted by climate change and other anthropogenic effects. This case study presents analyses of two sets of habitat data for P. cinereus, P. sherando, and P. hubrichti, one composed of microhabitat variables and the other macrohabitat variables.

The macro- and micro-scale analyses are then compared to assess the problems with ignoring either scale in interpretations of these species’ ecology.

Species Distribution Models

I used Maximum Entropy (MaxEnt) models (Phillips et al. 2006; Phillips and

Dudik 2008) to estimate the probability of occurrence for P. cinereus, P. hubrichti, and

P. sherando within central Virginia. MaxEnt models can be generated using presence only data making them particularly popular option for creating species distribution models. However, the ease of creating these models can be misleading, as the models are only as good as the data used to create them and are also impacted by user defined parameters (Yackulic et al. 2013). Therefore, it is important to use data and parameters that are appropriate for the questions the model is trying to address. I used presence data 34 collected from 1980 onward, from three sources including my own personal observations, data from the description of P. sherando (Highton 2004), and data from the Global

Biodiversity Information Facility (gbif.org). Presence data for all species was further trimmed to remove overlapping points. For P. cinereus, only points within the state of

Virginia were used and I used a subset of the data to select random points avoid sampling bias. Only points within Virginia were used to ensure a conservative estimate of the range of P. cinereus as this limits environmental data used in the models which reduces some variability that would be present if I used points from the species entire range.

Furthermore, the reduced dataset limits the impact of local adaptation at more geographically extreme portions of the species range from impacting the models. To be especially conservative I removed presence points of P. cinereus that fell within the range of P. sherando, as P. sherando was recently described and is phenotypically similar to P. cinereus. In all I used 100 presence points for P. cinereus, 37 for P. hubrichti, and 38 for

P. sherando. I created pseudo-absence points by creating a 50-km buffer around each of the species ranges and randomly selecting 100 points within those buffer zones. For my models I used abiotic data that are biologically relevant to plethodontid salamanders at ~

1 km2 resolution including slope, aspect, land cover (USDA Geospatial Gateway), soil water storage (USDA Geospatial Gateway), and 19 BioClim variables (WorldClim).

BioClim variables are often highly correlated as they are calculations based on

temperature and precipitation data, therefore I performed two Principle Component

Analyses (PCA) using the rasterPCA function from the RStoolbox package in R, one

including all BioClim variables related to temperature and the other including variables 35 related to precipitation. Using RStoolbox I created a raster layer for the first three PCA components for each of the analyses, which were used to create the final MaxEnt models.

Finally, I split the data into 5 groups, 4 of which were used to create the initial model, with the final group used to test model.

Microhabitat Data

I collected microhabitat data at a 1 m2 resolution, an order of magnitude smaller

than broad-scale climatic data, for each species from 2013-2015 specifically during

spring and summer months. I also collected microhabitat data from locations not

occupied by salamanders within 10 meters of captured individuals. Data for P. cinereus

includes locations from Ohio south to North Carolina, but I also subsetted the data to

include only presence points from within 10 km of both microendemic ranges.

Microhabitat data include air temperature and relative humidity at ground level and 1-

meter, soil moisture, soil temperature, and ground temperature. These variables were

chosen because they are comparable to the data used in my MaxEnt model; however,

because these microhabitat data are collected at the precise location and time a

salamander is found, they can better encapsulate the types of conditions the species are

experiencing. Climate data approximate conditions these species experience, but

salamanders can mediate the direct effects of climate using different microhabitats (e.g.

cover objects or belowground retreats) or through alterations to activity time. These data

represent an updated version of analyses completed by Farallo and Miles (2016) with

increased data for P. hubrichti and P. sherando. More details on data collection can be found in Chapter 2. 36

The MaxEnt models predict both microendemics to occur primarily within their current known range with minimal areas predicted, with any viable connectivity, outside of their known range Fig. 4). Similar to the results of Arif et al. (2007) I found that P. cinereus is predicted with high probability to occur throughout the range of both P. hubrichti and P. sherando based on these macro-scale abiotic data. In contrast, I found separation in microhabitat use between P. cinereus and both microendemics (Fig. 5).

Importantly, P. cinereus utilized microhabitats that differs from both P. sherando and P. hubrichti. Random points from just outside the range of both micro endemics was were also significantly different from presence points of the corresponding species.

Interpretations

The results from the MaxEnt models indicate that abiotic factors are not restricting P. cinereus from the ranges of either of the microendemics. Given the large range of P. cinereus, and their persistence in a wide range of environmental conditions, these results make understanding the restricted ranges of the microendemic species even more confusing. Arif et al. (2007) concluded that P. hubrichti was restricted by abiotic factors, while aggressive interactions by P. hubrcihti helped keep P. cinereus from overtaking their range. Undoubtedly, these behaviors play a role in this relationship; however, one aspect of behavior that is often overlooked is the microhabitat in which they choose to occur. Furthermore, there are minimal physiological differences between these three species, which indicates that there may be additional factors influencing the range of the microendemics. It is difficult to replicate true field conditions in a lab, and these may play an important role in the behaviors that occur between species. This can 37 include differences in temperature and/or humidity, but also structural differences within the environment that are hard to quantify. Differences in soil type, available vertical space within the soil, availability of cover objects, differences in vegetation, all may play a role in how interactions occur between species. These structural differences can impact thermal and hydric conditions, or also influence performance based on morphological differences. My data suggest that thermal and hydric conditions vary between each of the microendemics and P. cinereus. This may indicate that the species are seeking out specific microhabitat conditions. However, it may also be a byproduct of more refined structural habitat differences within the range of these microendemics. This is similar to what is occurring between P. cinereus and P. shenandoah, another microendemic from

Virginia. P. shenandoah is excluded from soil dominated habitat by P. cinereus, but they can outcompete P. cinereus in rocky talus slopes (Jaeger 1970; Jaeger 1971a; Jaeger

1971b; Jaeger 1972). In fact, one of the reasons that P. shenandoah is able exist in these dry talus slopes is because they can withstand longer periods of dehydration, further supporting that these microhabitats create pockets of differing hydric conditions which can influence interactions between species (Jaeger 1971b). This was discussed in Farallo and Miles (2016; See Chapter 2), as P. sherando exists in an area that has a high proportion of large rocks below ground providing open interstitial areas below ground. P. sherando also has longer limbs than P. cinereus which may provide increased performance in these microhabitats unique to their range. These are reasons why microhabitat use needs to be accounted for when addressing ecological and evolutionary 38 questions, especially when considering anthropogenic impacts. There may be patterns in broad-scale abiotic data that are heavily influenced by microscale factors.

Conclusions

I have highlighted the importance of microscale environmental characteristics and provided examples of why there needs to be a better understanding of the relationship of microhabitat and microclimate within species ecology. Microhabitat use can buffer species from climatic conditions and allow them to persist in locations that would otherwise result in decreased performance or be completely inhospitable. The use of these resulting microclimates may allow species to extend their activity time or even expand their range. These microclimates will obviously be partially dictated by broad- scale climatic conditions and habitat structure; however, microhabitat variation can have a tremendous impact on the environment that organisms experience. As climate change and other anthropogenic disturbances continues to alter ecosystems there needs to be a more complete connection between regional and local-scale processes. Local-scale processes due to species interactions with microhabitats and microclimates may allow species to persist in drastically altered habitat (Hampe and Jump 2011; With and Crist

1995), and result in changes to biotic interactions (Jaeger 1971a; Peñalver‐Alcázar et al.

2016), which will ultimately impact the potential range for species. Furthermore, micro- scale environmental use needs to be brought into an evolutionary context. Species may exhibit strong phylogenetic signal indicating that species all use similar niches when considered at a broad- spatial scale, but species may use very distinct microhabitats within that similar climatic niche. The opposite could also be true, with species being 39 found in very disparate climatic niches, but that use nearly identical microhabitats and microclimates. These data will be critical in helping mediating human caused impacts through forest management, species translocations, and determining where to focus conservation efforts.

Figures

Figure 1: Modified Stommel diagram A modified two dimensional version of figure 1 from Stommel (1963). This diagram represents approximate regions within time and space which would have high levels of variation for thermal and hydric environmental characteristics. These are broken into macro-scale characteristics (air temperature and precipitation) and micro-scale characteristics (ground temperature and soil moisture). 40

Figure 2: Diagram of environmental characteristics and their interactions This diagram highlights how micro-scale characteristics like soil temperature and soil moisture are affected by numerous macro-scale characteristics, which in term results in greater variation at high spatial and temporal resolution. Environmental characteristics near the top are associated with large (regional level) patch sizes, while characteristics at the bottom are associated with small (local level) patch sizes.

41

Figure 3: Temperature fluctuation for five microhabitats over two months Temperature data collected using Thermochron iButtons in Wayne National Forest, Ohio from July to September 2012. Temperature was recorded 50 cm above ground, at ground level, and 10, 20, and 30 cm below ground level. The data show that below ground temperatures, comparable to under cover objects and other microhabitats available at the surface, maintain much more constant temperatures than exposed above ground microhabitats. However, as ambient temperatures began to decrease, below ground temperatures also decrease, but at a slow rate and with less variation.

42

Figure 4: Range maps and Maximum Entropy models Range maps for the widespread Eastern Red-backed Salamander (Plethodon cinerues) and the two microendemics, the Big Levels Salamander (P. sherando) and Peaks of Otter Salamander (P. hubrichti). Below the range maps are probability of occurrence generated from Maximum Entropy Models for each of the species, specifically for the area in central Virginia where both microendemics are found.

43

Figure 5: Results of the NMDS analysis Results from a non-metric multidimensional scaling analysis of microhabitat data updated from Farallo and Miles (2016) with additional sampling for the Peaks of Otter Salamander (P. hubrichti). The 95% confidence ellipses are presented for the capture points for Eastern Red-backed Salamander (Plethodon cinereus), Big Levels salamanders (P. sherando), and Peaks of Otter Salamander (P. hubrichti) (3); random points within the sample plot for each species (3), and random points that are within 10 km of the range of P. sherando and P. hubrichti (2). The first axis predominately describes soil temperature and soil moisture. Points on the positive pole consisted of sites with low soil temperatures and high soil moisture. The second axis describes relative humidity as well as air and soil temperature. Points positioned on the positive end of the second axis have high relative humidity and points towards the negative pole had low values for air temperature (1 m above ground).

44

CHAPTER 2: THE IMPORTANCE OF MICROHABITAT: A COMPARISON OF

TWO MICROENDEMIC PLETHODON SPECIES TO THE WIDESPREAD P.

CINEREUS

Introduction

A major question in ecology concerns the factors that determine the distribution of species (Brown 1984; Rosenzweig 1995; Werner et al. 2014). Abiotic and biotic factors may jointly interact to generate a mosaic of environments that structure the potential distribution of a species. For many species (particularly in ectotherms), climate and abiotic conditions may be the primary drivers of a species distribution (Davis and

Shaw 2001; McCarty 2001; Walther et al. 2002). Characterizing a species distribution has taken on new urgency as ecologists and conservation biologists struggle to predict biotic responses to climate change.

Most analyses of species distributions focus on a niche-theoretical framework to elucidate the factors constraining habitat occupancy of a species (Werner et al. 2014).

However, this approach emphasizes habitat characteristics in coarse detail in order to use modern statistical methods to predict a species range using occurrence data. Species distributional models that use species presence data in conjunction with an ensemble of environmental data have been used to predict species responses to changing climates

(Fitzpatrick et al. 2013). Recent methods have incorporated physiological and energetic data to refine (Kearney and Porter 2009; Kearney and Porter 2004) and predict species ranges (e.g., Buckley et al. 2010). 45

Macro-scale models of species distributions may provide the ability to forecast the distribution of widespread species. However, modeling the niche characteristics of microendemic species requires abiotic and biotic data at a finer resolution. A first step entails quantifying those factors affecting the performance of a species. Moreover, the selection of variables should include environmental aspects that are likely to affect a species population growth rate. In ectothermic organisms, linking habitat variation to population growth rates can be accomplished by modeling how the thermal environment affects physiological performance. Because performance is known to influence key components of fitness (e.g., growth, survivorship, and reproduction), quantifying those environmental attributes that may determine the ability of an individual to attain a physiologically active temperature is a first step. Huey (1991) presented a schema illustrating how the abiotic environment may impinge on an organism’s fitness via Tb. I

have modified the schema to include cutaneous water loss, combined with Tb, as “filters”

which can impact an organism’s fitness. Jointly, these two variables may affect physiological capacities, such as locomotor performance (e.g., Preest and Pough 1989;

Titon et al. 2010). Individual variation in physiological performance has fitness

consequences by affecting growth rates (Sinervo 1990), survivorship (Miles 2004), mating opportunities (Robson and Miles 2000) and fecundity (Stahlschmidt et al. 2013).

Plethodontid Salamanders as a Model System

Plethodontid salamanders are one of the most abundant vertebrates in forests of the eastern United States (Semlitsch et al. 2014). This group of salamanders has a critical role in regulating invertebrate detritivores (Walton 2005; Walton 2013; Walton et al. 46

2006). They also function as energy capacitors through storage of nutrients in forest ecosystems (Hickerson et al. 2012; Semlitsch et al. 2014). Thus plethodontid salamanders serve as a key indicators of forest health because of their sensitivity to habitat disturbances and ease of sampling (Welsh and Droege 2001).

The habitat requirements for plethodontid salamanders are structured by several physiological constraints as illustrated by Huey (1991; Fig. 1). First, a primary physiological feature of Plethodon salamanders is their reliance on cutaneous respiration.

Consequently, individuals require cool and moist habitat to persist (Gatz et al. 1975;

Spotila 1972; Wells 2007). The second attribute of Plethodon salamanders is their nocturnal activity period, which limits their chance of overheating, but also means that any active thermoregulation they utilize must involve exploiting microhabitats which have been heated at different rates during the day. Most plethodontids are imprecise thermoregulators (Feder 1982), but there is evidence that species may select temperatures through habitat selection. Broadly, Plethodon salamanders can move vertically in the soil through the use of burrows and natural crevices created by rocky substrates to find appropriate moisture and temperature (Caceres-Charneco and Ransom 2010;

Diefenbacher 2007; Drake et al. 2012). This vertical movement allows them to persist when surface conditions are not ideal including using below ground refugia during cold winter months (Caldwell 1975; Caldwell and Jones 1973; Grizzell 1949; Hoff 1977), as well as when surface conditions are too dry or warm during the rest of the year (Camp

1988; Houk 1977; Stebbins 1954). 47

The requirement for thermally appropriate microhabitats is especially important in light of the planet’s changing climate. There is predicted to be substantial loss of suitable habitat as a result of climate change (Milanovich et al. 2010). Salamanders are especially vulnerable to habitat changes. Their physiological requirements for cool and moist habitat may drive dispersal to higher elevations until suitable habitat is no longer available.

Therefore, climate change is expected to have detrimental impacts on these high elevation species (Raxworthy et al. 2008; Xu et al. 2009). Species range shifts have already occurred as a result of climate change (Parmesan 2006; Walther et al. 2002).

Some salamander species have shown a decline in body size, presumably in response to

climate change (Caruso et al. 2014) which could impact the habitat they are able to utilize

or cause changes in interspecific competion. Additionally, similar to the potential impacts

of climate change, other disturbances that alter the forest floor habitat have been shown to be detrimental to salamander populations (Gibbs 1998; Hocking et al. 2013).

One advantage that amphibians, as well as other small ectotherms, have is that they are capable of utilizing microhabitat refugia, which may help mitigate the effects of climate change (Seebacher and Alford 2002; Shoo et al. 2011). In addition to temperature amphibians also need to regulate moisture, making microhabitats especially critical for maintaining physiological performance (Spotila 1972). Microhabitat refugia include leaf litter and cover objects such as rocks, woody debris, or other substrates that buffer against warm, dry, or both habitats (Grover 1998; Jaeger 1972; McKenny et al. 2006; Patrick et al. 2006). Time since last rain has been shown to improve the quality of microhabitats

(e.g. increasing moisture on the forest floor) and increase surface activity of P. serratus 48

(O'Donnell et al. 2014). Quantifying how microhabitats are affected by regional scale climate variables as well as the conditions in which salamanders exploit microhabitat refugia will facilitate predictions regarding potential changes to species distribution or ecology.

Plethodontid salamanders are characterized by having small home ranges; however, individuals within a population may utilize a diverse array of microhabitats within a small area, including arboreal substrates (Jaeger 1978; Niemiller 2005; Regester and Samoray 2002; Trauth et al. 2000), various types of cover objects (Jaeger et al. 1982;

Mathis 1990), and burrows below the forest floor (Caceres-Charneco and Ransom 2010;

Diefenbacher 2007; Drake et al. 2012). Therefore, determining detectability is important when attempting to assess the status of populations which will be further enhanced by increasing our knowledge of microhabitat use (Bailey et al. 2004a; Bailey et al. 2004b;

Bailey et al. 2004c; Dodd and Dorazio 2004; MacKenzie et al. 2004). Previous studies have assessed the habitat associations of widespread plethodontids (Gibbs 1998; Hocking et al. 2013; Peterman et al. 2013). However, there is scant information regarding specific ecological requirements of many microendemic salamander species. The absence of key data on habitat requirements complicates the ability to assess population status and predict the consequences of anthropogenic disturbance.

The primary goal of this study is to assess the microhabitat use of two microendemic species whose natural history is largely unknown: the Peaks of Otter

Salamander (P. hubrichti) and the Big Levels Salamander (P. sherando). I use these data to compare with the closely related and widespread congener, the Eastern Red-backed 49

Salamander (P. cinereus). The distribution of P. cinereus spans over 1.8 million km2

(Fig. 7) and completely surrounds the range of both microendemic species, but includes only a narrow area of sympatry with both species (Highton 2004; Kniowski and

Reichenbach 2009). Plethodon hubrichti is hypothesized to be the sister taxa to P. nettingi (Highton et al. 2012), another endemic Plethodon. However, P. hubrichti is also hypothesized to be closer related to P. cinereus than P. sherando, whereas P. sherando is the sister taxa to P. serratus, the Southern Red-backed Salamander (Bayer et al. 2012).

Interestingly, the range of P. sherando is completely surrounded by P. cinereus, while P. serratus is found over 370 km from P. sherando. Both microendemic species have restricted ranges of ~36 km2 and ~49 km2 for P. sherando and P. hubrichti, respectively

(Fig. 7). The restricted ranges of these species and their high elevation habitats make them vulnerable to population declines from climatic warming. Warmer and drier environments at low elevations will force most movement toward even higher elevation habitats. Furthermore, both of their ranges are encompassed by P. cinereus, which means any expansion or change in the geographic distribution of the microendemics as a consequence of climate change may create a challenge for population persistence through competition. For example, P. cinereus is thought to have contributed to the decline of the

Cheat Mountain Salamander (P. nettingi) in the past 30 years as a result of the former species exhibiting an upward shift in elevation and increasing in abundance (Kroschel et al. 2014). Character displacement also occurs in areas where another species, P. hoffmani, is sympatric with P. cinereus indicating that competition occurs between small bodied Plethodon (Jaeger et al. 2002). Despite the potential threats to both species, little 50 is known about either species ecology. Kniowski and Reichenbach (2009) conducted a general assessment of sympatric populations of P. hubrichti and P. cinereus, while

Mitchell et al. (1996) and Reichenbach and Sattler (2007) showed that clear cutting of forests had a negative effect on P. hubrichti populations. To my knowledge no studies have addressed the ecology of P. sherando. Furthermore, no studies have assessed detailed habitat use, such as soil moisture, relative humidity, and temperature of microhabitats for either of the microendemic species. In order to assess the potential impacts of climate change on species with highly restricted ranges we need a better understanding of how species utilize habitat at a biologically realistic scale.

I test two hypotheses in this study. First, do the three species exhibit a preference for habitat characteristics that differs from the available habitat? Second, do the microhabitats used by the microendemic species overlap with the preference exhibited by the widespread species, P. cinereus? I predict that the widespread species, P. cinereus, should exhibit less selectivity than either of the microendemics. This prediction is based on the premise that P. cinereus is expected to be a habitat generalist that exploits multiple types of habitats. Furthermore, the microendemics are predicted to have specialized habitat preferences and exhibit limited overlap with that of P. cinereus. I also comment on the role of microhabitats as potential refugia from climate change. Finally, I suggest additional avenues of research that would enhance our knowledge of these microendemic plethodontids and enhance predictions regarding their vulnerability to climate change.

51

Material and Methods

Field Sites

I measured microhabitat variables at 39 localities throughout the Appalachian

Mountains and foothills within the range of Plethodon cinereus (35), P. sherando (2), and

P. hubrichti (2) between June 8th, 2012 and October 18th, 2014. Each locality consisted of

an area of 1 km2. At each locality I searched sites for salamanders using time constraint

surveys of 1 person hour each and haphazard searching. Surveys were conducted during

the day when salamanders are under cover and in the evening when they are active. I

completed a total of 99 surveys, with each locality being surveyed between 1–5 times.

Plethodon cinereus field sites were located between 39.6° and 36.0° latitude and

represented the southern half of the species range. This latitudinal band is comparable

with the distribution of the two microendemic species from Virginia.

Microhabitat Measurements

I collected microhabitat data at the capture location of each salamander. I also

obtained habitat data at 10 randomly chosen points within the search area. The random

points were selected to provide information on the availability of each microhabitat

category included in my study. All random points were found by using a random number

generator to determine a compass bearing and then walking 10 meters in that direction. In

addition to the systematic collection of random points I also included points 10 meters in

a random direction from presence points and also under cover objects within a 1 m2 areas around presence points where salamanders were found. Each random absence point represents a location that did not have a salamander present during my survey. I do not 52 mean to imply that no salamander ever occurs at the absence points, but rather that a salamander is not present and surface active under the current microhabitat conditions.

These sampling methods ensured that I located salamanders in as many surface active microhabitat locations as possible as well as thoroughly sampling microhabitats available for them to potentially utilize. During surveys I collected ecologically relevant environmental data for plethodontid salamanders that rely on cutaneous respiration. Air temperature (±0.5°C) and relative humidity (±3%) were both recorded 1 m from the

ground using a Kestral® 3500 weather meter and digital psychrometer. I used measurements at 1 m because these are comparable to broad-scale climatic data (e.g.

WorldClim). I recorded soil temperature using either a ThermaPlus thermocouple meter or an Infrared thermometer (IRT) with a high sensitivity probe (±0.5°C; Thermoworks

Inc.). I measured the temperature of the ground and substrate surface with an Infrared thermometer (±0.6°C; Thermoworks Inc.). I measured soil moisture using a HydroSense

II (±3%; Campbell Scientific Inc.). Soil moisture and temperature probes were all inserted at approximately a 45° angle. The microhabitat sample consisted of 152 presence and 839 absence points (991 total).

I also placed iButton temperature loggers (Model: DS1922L, Embeded Data

Systems) at various microhabitats at a site in Perry County, Ohio within P. cinereus

habitat. The loggers were placed in 5 locations, 50 cm above ground, ground level, and

10, 20, and 30 cm below ground level from July 14th though August 24th 2012. I used

these data to quantify the temporal change in the thermal environment. 53

Statistical Analyses

I classified each site into one of eight categories. The first three categories were the capture locations for each of the focal species; another three categories included the random points within the range of each of the focal species; and the final two categories consisted of random points within the range of P. cinereus but located within 10 km of the range of P. sherando or P. hubrichti.

In order to visualize difference in habitat between these categories I utilized non- metric multidimensional scaling and plotted the 95% confidence ellipses. I used a

Generalized Linear Mixed Models (GLMM) from the R package lme4 (Bates et al. 2014) with site treated as random factor and the habitat characteristics as predictor variables to determine if each species utilized the microhabitat variables different from what was available. I also compared the parameter estimates to infer whether each species used similar habitat variables. Prior to running the GLMM, I screened the data for evidence of multicollinearity by examining the pairwise correlations among all variables. None of the correlations exceeded 0.70, suggesting multicollinearity is not likely to affect the results of the GLMM. I used the function “dredge” from the package MuMIn (Barton 2013) to compare all possible subsets models out of the possible models using a saturated model as the initial model. The “dredge” function uses AICc scores to select the best reduced

2 model from the original saturated model. The marginal R GLMM was also calculated for

each model using a modified method of Nakagawa and Schielzeth (2013) [see Johnson

(2014)] using the best supported model for each species. If there are significant

differences in habitat use for specific microhabitat variables at presence sites compared to 54 absence sites the marginal R2 provides an indication of the microhabitat variables selected

by each species. Comparison of significant microhabitat variables for each species assists

in determining the potential mechanisms for the absence of overlap between these

microendemic species and the widespread P. cinereus. All statistical analyses were

conducted using the program R (ver. 2.15; http://www.r-project.org/).

Results

Variation in Microhabitat Traits

Analysis of the microhabitat characteristics by the NMDS revealed substantial variation across all groups. The first axis described a soil temperature and soil moisture gradient. Points on the positive pole consisted of sites with high soil temperatures and low soil moisture (Table 1). Whereas points positioned on the positive end of the second axis have high above ground relative humidity and points towards the negative pole had high values for air temperature (1 m above ground) and below-ground soil temperatures

(Table 1). The first axis separated P. sherando from P. cinereus and P. hubrichti (Fig. 8), which suggests the former species tolerates warmer, and drier soil surfaces. The second axis positioned P. hubrichti and P. sherando at the negative part of the habitat gradient and P. cinereus in the positive zone. Thus, P. cinereus occupied sites with higher relative humidity and lower temperatures. Both P. hubrichti and P. sherando occurred in sites with warmer temperatures. Examination of the 95% confidence ellipses reveals that there is some overlap in the microhabitat variables that characterize the capture sites for the three salamanders. An analysis of distance (function ADONIS) found significant 55 differences between P. cinereus and P sherando (F1,136 = 303.74, p = 0.024) but no

difference between P. cinereus and P. hubrichti (F1,111 = -8.274, p = 0.75).

The widespread P. cinereus showed no overlap between the ellipses for the

capture and random points. In contrast, both microendemic species exhibited variable

amounts of overlap between the capture and random points. The ellipses for P. hubricthi

almost completely overlapped, although the orientation of the random and capture

ellipses differed. The random points for P. sherando overlapped with the capture points,

but not as extensively as P. hubrichti. Note that the size of the confidence ellipses for P. cinereus are smaller than either microendemic species. There is also a striking concordance between the confidence ellipses for P. cinereus and P. sherando. This suggests that although the species occupy environments with similar habitat structure, they are selecting divergent microhabitat features. Finally, the two categories that included random points at sites within the range of P. cinereus but directly adjacent to P. sherando and P. hubrichti random points were positioned away from both the capture points for the three salamander species and their associated random points.

GLMM Analysis

I applied a generalized linear mixed model to determine which habitat variables

predict the presence of each species. Because of the limited number of capture points for

P. hubrichti I did not include the species in the GLMM analysis. However, I present the

patterns for habitat use as a comparison with the remaining species. I detected significant

differences between the habitat characteristics of capture sites and random points for both

2 P. cinereus and P. sherando (Table 2). The best model for P. cinereus (marginal R GLMM 56

= 76%) included two significant variables (ground temperature and relative humidity).

2 The best model for P. sherando (marginal R GLMM = 41%) involved three variables (air temperature, soil temperature, and ground temperature). Plethodon cinereus was found at significantly cooler ground temperature and lower relative humidity than what was available (Fig. 9). Plethodon sherando was found at areas with significantly higher air

temperature, and lower ground and soil temperatures (Fig. 9). Although the GLMM

found no difference in soil moisture used by any of the species compared to what was

available, P. sherando was found in areas on average with 2% higher soil moisture.

Temporal Change of Microhabitats

The thermal characteristics of below ground microhabitats exhibited remarkable

consistency in both mean values throughout the day (Fig. 10A) and over the entire period

of deployment (Fig. 10B). Moreover, the logger positioned 50 cm above ground had a

temperature range of 9.6 °C between the hourly minimum and maximum temperatures

over the course of an average 24 hour period, which was the highest value found among

all sites (Fig. 10A). Notably, the logger 30 cm below ground only spanned 0.2 °C on

average over a 24 hour period (Fig. 10A). When considering the temperatures

experienced throughout the entire deployment, the logger 10 cm below ground only

varied 8.5 °C with the logger at 30 cm below ground staying within 5.0°C throughout the

41 days (Fig 10B). In contrast, the logger at ground level spanned 25.0 °C and the logger

at 50 cm above ground had the greatest temperature range of 29.6 °C (Fig. 10B). 57

Discussion

I found Plethodon cinereus and P. sherando utilize significantly different microhabitats than what is available indicating that, at least when they are at the surface, they are seeking out specific habitats. In contrast, P. hubrichti showed substantial overlap between the habitat characteristics at the site of capture and the random points. All three species exhibited variable amounts of overlap with each other in the microhabitat space.

Furthermore, different microhabitat variables seemed to be important for each species.

Although I could not include P. hubrichti in a GLMM analysis, the species does not appear to exploit habitats that differ from the available habitat characteristics. Plethodon hubrichti also appears to utilize different habitat than P. cinereus based on the NMDS results; however, my statistical analysis did not support this conclusion. Given the low number of capture points for P. hubrichti, these results are tentative and require additional sampling while potentially including variables not included in this study such as vegetation or soil characteristics.

The distribution of P. sherando includes only a small area of sympatry with P. cinereus (Highton 2004). My results suggest that the species are utilizing different microhabitats. One potential explanation for this pattern is that competition is structuring the habitat differences. Many species of Plethodon exhibit territoriality often defending cover objects from other conspecifics (Jaeger and Forester 1993; Mathis et al. 1995).

However, some species exhibit territorial behaviors towards heterospecifics which can result in one species excluding the other from specific habitats (Anthony et al. 1997;

Griffis and Jaeger 1998; Jaeger 1971a). Furthermore, Jaeger et al. (2002) showed 58 character displacement occurs in areas of sympatry between P. cinereus and P. hoffmani.

In addition, the GLMM analyses revealed that the habitat traits predicting the occurrence of both species differed. This heterogeneity in microhabitat selection is also consistent with morphological differences between the species. The microendemic species, P. sherando, is morphologically distinct having a larger head and longer limbs (Highton

2004). These morphological features may enhance the ability of P. sherando to access thermally and hydrically favorable microhabitats. Within the small range of P. sherando there is a high density of large rocky substrates that extend below ground level (Farallo, pers. obv.) I have observed areas where P. cinereus are found in soil dominated habitat, but in an area immediately adjacent where rocky habitat is dominant, P. sherando have been found (Farallo, pers. obs.) The use of rocky substrates has been associated with long limbs in Australian skinks (Goodman et al. 2008). Longer limbs may allow P. sherando better access to preferred microhabitats available in open pockets within rocky crevices below the forest floor. However, limb length may entail a trade-off by limiting access to preferred thermal microhabitats that are dominated by soil and finer substrates.

Davis and Pauly (2011) observed larger heads in a group of Western Slimy Salamanders

(P. albagula) that routinely use subterranean karst habitat when compared to populations that are more likely to utilize terrestrial habitat. Plethodon sherando has a larger head than P. cinereus which may have an unforeseen benefit when utilizing these rocky underground habitats. There may also be morphological traits that are plastic, such as vertebrae number resulting in variable levels of elongation, allowing populations to better exploit different habitat conditions (Jockusch 1997). Additional studies to quantify the 59 performance differences of P. sherando when using subterranean environments could determine the role of limbs in habitat selection. Furthermore, limited data are available for microhabitat temperatures and moisture levels in below ground refugia.

A key question in biogeography and conservation biology is what determines the heterogeneity in species distributions, especially microendemic species. One potential explanation for the limited range of P. sherando and P. hubrichti is that P. cinereus typically out competes other small-bodied Plethodon species (Adams and Rohlf 2000;

Jaeger 1970; Jaeger 1972; Jaeger et al. 2002; Kroschel et al. 2014). The ability of P. cinereus to thrive in a broad array of forest habitats results in other microendemic species being restricted to specific types of habitat where they are able to gain a competitive advantage. Conversely, in lab based behavioral trials, P. hubrichti are more aggressive than P. cinereus (Arif et al. 2007), but it appears that P. hubrichti is restricted by abiotic factors (Arif et al. 2007) which results in their small range despite being competitively superior to P. cinereus. However, these results do not take into account microhabitat use as well as behavioral interactions that may be affected by differences in microhabitat. The laboratory behavioral trials were performed under standardized conditions and they only included climatic data as abiotic factors. These results certainly provide possibilities for the restricted range of P. hubrichti and their interactions with P. cinereus; however, understanding how P. hubrichti utilizes microhabitat and consequently how they interact with P. cinereus when in those specific microhabitat will most likely provide a more complete understanding of their restricted range. 60

My results partially support this pattern of habitat differentiation. The position of the 95% confidence ellipses for both P. sherando and P. hubrichti are significantly shifted away from P. cinereus. However, comparisons of the microhabitat characteristics between capture and random points for P. hubrichti exhibited substantial overlap, suggesting the species is not selective, at least given my limited data. Plethodon sherando shows little overlap between occupied and random points. The species tends to favor warmer habitats that have cooler ground and soil temperatures. Interestingly, the microhabitat features of the random points for P. cinereus and P. sherando occupy similar sections of the microhabitat space defined by the NMDS analysis. In contrast, P. sherando inhabits forested environments that are similar to P. cinereus, but has different thermal requirements. My results emphasize the need to include behavioral and ecological traits to enhance our ability to determine how changes to habitat will affect species persistence.

The thermal data derived from the deployment of dataloggers provide a key pattern. I deployed the dataloggers during summer months when P. cinereus is not engaged in surface activity at low elevations. However, the thermal data demonstrates the consistency of below ground temperatures. Salamanders are able to seek refuge in microhabitats that provide a nearly constant temperature by venturing only 10 cm below ground. It is also very likely that a similar effect would be seen under leaf litter. Although this was not measured by my data loggers, I have seen striking differences in temperatures above and below leaf litter including a 34.2 °C difference at a field site in

West Virginia where salamanders were present under leaf litter (54.2 °C above leaf litter 61 and 20.0 °C below leaf litter; unpubl.). Another factor not measured by the data loggers is relative humidity specifically within underground retreats which may provide pockets of higher humidity levels.

My results pose some critical questions that need to be addressed quickly in order to mitigate potential impacts of climate change on salamander populations. Given the numerous microendemic species whose distributions are limited to high elevation habitats, the potential for dispersal to thermally favorable microclimates is unlikely.

Determination of the exploitation of below ground microhabitats by salamanders enhances our ability to design and implement new habitat studies. If salamanders are able to thrive with minimal or no surface activity, including sufficient food acquisition and mating success, then habitat studies need to shift to include their below ground habitats.

However, if the ability of salamanders to feed and mate requires surface activity then research should focus on the impacts of climate change on these surface habitats even if they are only used for short periods of time.

62

Tables

Table 1: Correlation of environmental variables to the two NMDS axis NMDS 1 NMDS 2

Air temperature 0.328 -0.556

Soil temperature 0.546 0.336

Ground temperature 0.263 -0.571

Soil moisture -0.926 0.045

Relative humidity 0.270 0.752

Table 2: Results of generalized linear mixed models The results of generalized linear mixed models comparing the habitat variables the best predict the presence sites of Plethodon cinereus and P. sherando. Parameter estimates, and standard error are presented. Values in Bold indicates P < 0.05; — indicates that a variable was not included in the best model for that species. Plethodon cinereus Plethodon sherando Parameter Standard Parameter Standard Estimate Error Estimate Error Air temperature 0.067 0.046 1.904 2.255

Soil temperature — — -0.761 0.165

Ground temperature -0.381 0.059 -0.968 0.206

Soil moisture — — — —

Relative humidity -0.047 0.008 — —

63

Figures

Figure 6: Depiction of habitat selection on evolutionary change The distribution of a plethodontid salamander species is affected by the habitat characteristics such as ambient temperature and hydric environments that optimize short- term physiological performance. Variation in these two categories can affect population dynamics via their influence on key fitness traits, such as growth, survivorship, and reproduction. Modified from Huey (1991).

64

Figure 7: Distributions maps of Plethodon cinereus, P. hubrichti, and P. sherando Ranges of Plethodon cinereus, P. sherando, and P. hubrichti taken from IUCN Redlist (www.iucnredlist.org). The inset map in the top left corner contains the range of P. cinereus and also denotes with a black rectangle the extent of the main map within the US. 65

Figure 8: Results of NMDS analysis Results from a non-metric multidimensional scaling analysis of microhabitat data. Symbols refer to the capture points for Plethodon cinereus, P. sherando, and P. hubrichti (3); random points within the sample plot for each species (3), and random points that are within 10 km of the range of P. sherando and P. hubrichti (2). The first axis describes soil temperature and soil moisture. Points on the positive pole consisted of sites with high soil temperatures and low soil moisture. The second axis describes relative humidity as well as air and soil temperature. Points positioned on the positive end of the second axis have high relative humidity and points towards the negative pole had high values for air temperature (1 m above ground) and below-ground soil temperatures. 66

Figure 9: Mean values of microhabitat variables at presence and absence sites Comparison of microhabitat variables for Plethodon cinereus, P. sherando, and P. hubrichti at capture sites (dark grey) and absence points (light grey). Values are means  standard error. 67

Figure 10: Temporal fluctuation in temperature recorded at five different microhabitats over a two month period All measurements are from Perry County, OH. (A) Diurnal variation in mean at each microhabitat. (B) Seasonal variation in temperature between July 14th and August 24th 2012. Approximate thermal preference zone is highlighted by an opaque rectangle. 68

CHAPTER 3: SMALL-SCALE HABITAT USE OF PLETHODONTID

SALAMANDERS ALONG THE APPALACHIAN MOUNTAINS

Introduction

A fundamental goal of ecology is to explain patterns of species diversity which often includes an assessment of species habitat use. Quantifying habitat use and climates where species are found provides information regarding species distributions (e.g. Aarts et al. 2013), species abundance (e.g. Lambert et al. 2006), potential for interspecific interactions (e.g. Munday 2001), and species vulnerability to alterations to habitat or climate (e.g. Mazziotta et al. 2016). Despite most organisms existing at a small spatial scale of less than a m2 (Suggitt et al. 2011), most studies quantify habitat and climate at

much coarser scale, typically of 1 km2 or greater. This discrepancy in scale between the

data collected and what is biologically relevant may alter interpretations of species

habitat and climate use.

Salamanders of the family Plethodontidae are one of the many taxonomic groups

that comprised of individuals that exist at a small spatial scale. These salamanders

represent a diverse group of amphibians being the most species rich salamander family

and the 5th most species rich among all amphibians. These salamanders are found

predominately in North and South America, with isolated species in central Europe and in

South Korea (AmphibiaWeb 2016). However, despite the family being widespread, one

of the most prominent hotspots of species richness is in the southern Appalachian

Mountains with areas having upwards of 24 species of plethodontids (Kozak and Wiens

2012; Fig. 11). This hotspot lies within a temperate region and therefore experiences 69 notable seasonality resulting in both spatial and temporal variation in environmental conditions. The high species richness and abundance of salamanders in this region results in a degree of interspecific interactions that is experienced in only a few places over the families range. Furthermore, salamanders in this region play a critical role in forest floor ecosystems (Hocking and Babbitt 2014; Walton 2005; Walton 2013; Walton and Steckler

2005; Welsh and Droege 2001). This includes functioning as storage of nutrients

(Hickerson et al. 2012; Semlitsch et al. 2014), reduction of leaf litter decomposition (Best and Welsh 2014; Wyman 1998), and the regulation of invertebrates (Walton 2013;

Wyman 1998). Therefore, understanding how these salamanders use habitat, as well as if species differentially use habitat, is important, especially in the light of continued habitat alteration and climate change.

As there is extensive diversity in environmental conditions there is also diversity among the plethodontids. Within the southern Appalachian Mountains plethodontids use a wide variety of habitats ranging from completely aquatic species such as Shovelnose

Salamander (Desmognathus marmoratus) to completely terrestrial species such as the

Eastern Red-backed Salamander (Plethodon cinereus), as well as many species that use a combination of the two, such as Southern Two-lined Salamander (Eurycea cirrigera).

These plethodontids also exhibit tremendous variation in body size ranging from the very small such as the Patch-nosed Salamander (Urspelerpes brucei) and Pygmy Salamander

(Desmognathus wrighti), both with SVLs smaller than 27 mm, as well as larger species such as the Black-bellied Salamander (Desmognathus quadramaculatus), which can have an SVL greater than 120 mm. With a size disparity and habitats that range from aquatic 70 to terrestrial, plethodontid salamanders have a variety of biotic interactions including both competition and predation especially with invertebrates (Anthony et al. 2007; Gall et al. 2003; Hickerson et al. 2004; Walton 2013; Ziemba et al. 2016), and other species of salamanders (Anthony et al. 2008; Anthony et al. 1997; Arif et al. 2007; Jaeger 1970;

Jaeger 1971a; Jaeger and Forester 1993; Jaeger et al. 2002). Furthermore, there are polymorphic species that can contribute to intraspecific competition as a result of differences in the ecology of the morphs (Anthony et al. 2008; Fitzpatrick et al. 2009;

Venesky and Anthony 2007). Plethodontid salamanders also vary in their physiology such as thermal preferences, critical thermal maximum, and water loss rates (Brattstrom

1963; Spotila 1972). Several species of plethodontid salamanders even exhibit intraspecific variation in their resistance to water loss between high and low elevation populations (Riddell and Sears 2015; Winters and Gifford 2013; Chapter 4).

Plethodontids are lungless and must rely on cutaneous respiration and therefore all species typically require cool and moist habitat to persist (Gatz et al. 1975; Spight

1968; Spotila 1972; Wells 2007). This physiological constraint may be why species within this family also typically have very small home ranges with limited dispersal capabilities (Feder 1983). Their small home ranges (< 10 m2) makes using climatic data

at coarse spatial scales (≥ 1 km2) potentially misleading. Wells (2007) summarized 34

studies that assessed home ranges of amphibians, of which 12 were on plethodontid

salamanders. Salamanders have by far the smallest home ranges, spanning from 0.1 – 90

m2, compared to Anurans that span 1-1,900 m2. Furthermore, plethodontids have even

smaller home ranges than other salamanders, with most plethodontid species mean ranges 71 falling under 10 m2. When compared to other small vertebrates (< 20 g), plethodontid

species with the largest home ranges would still fall in the very bottom end of the

spectrum of reptilian, avian, or mammalian species. For example, the median home range

for 34 species of birds under 20 g was 10,000 m2, while for all of salamanders (including

non-plethodontids) the median was 4 m2 (Wells 2007). The small home ranges of

plethodontid salamanders are obviously not just a byproduct of their size, but are instead

a function of dispersal capabilities, which will be impacted by physiology and biotic

interactions. Because plethodontid salamanders rely on cutaneous respiration, as a group,

their dispersal capabilities are limited. They must remain in moist conditions to exchange

oxygen and carbon dioxide, and therefore even small sections of inhospitable habitat can

be an impenetrable barrier to dispersal. Importantly, much of these physiological barriers

to dispersal will be, at least in part, in capsulated by both the coarse spatial scale of

climatic data and the fine spatial scale of microhabitat data. Climatic data may be

relevant for explaining current species richness patterns and species distributions.

However, through the use of microhabitats or altering activity times, species may be able to expand their ranges beyond what would be predicted which may be important for mediating the impacts of climate change. Furthermore, as there are interspecific differences in physiology and morphology between species, there are also differences in habitat use. An extreme example is given by the genus Desmognathus, which has 21

species, many of which are sympatric. Even within this genus there is tremendous

variation in body size including some of the largest and smallest plethodontids. Although,

these species often coexist, they use markedly different habitat types, with larger species 72 using streams most frequently, and smaller species using terrestrial habitat more frequently (Petranka and Smith 2005; Southerland 1986) with Desmognathus wrighti being the smallest and only completely terrestrial Desmognathus salamander. However, in addition to a simple gradient between aquatic and terrestrial habitat, there are many other microhabitats within forest floor ecosystems. These include various degrees of ground cover, a variety of cover object types, leaf litter depth, slope, aspect, soil type, all of which contribute to a heterogeneous mosaic of thermal and hydric conditions at small and temporal spatial scales on the forest floor which can be differentially utilized by plethodontid salamanders (Farallo and Miles 2016; Chapter 4).

Several studies have examined plethodontid habitat use at either a small spatial scale or using coarse approximations for microhabitat and microclimate and have typically demonstrated that salamander microhabitat use is tied to environmental variables such as available moisture (e.g. Farallo and Miles 2016; Jaeger 1980; O'Donnell et al. 2014). In this study I examine both intermediate habitat characteristics such as ground cover leaf litter depth, slope, aspect, canopy cover, referred to as microhabitat and small-scale thermal and hydric conditions of these microhabitats, which will be referred to as microclimate. This study also occurs along a 580-km span of the Appalachian

Mountains providing data from a large geographic area, but with data at a small temporal and spatial scale.

In general, there are three categories of habitats that are important to salamanders.

The first includes retreat sites, typically underground, where salamanders will spend much of their time. These microhabitats will shelter them from inhospitable surface 73 conditions, including heat in the summer, and freezing conditions in the winter. The second are surface conditions when they are active. This includes salamanders, either on leaf litter or climbing vegetation, but this will occur at night or directly after precipitation events when chances for desiccation will be low. Third includes surface microhabitats such as cover objects and leaf litter, which allow salamanders to remain at the surface, but still buffered from unfavorable ambient conditions. The use of the first two categories of microhabitats will most likely be dictated by general plethodontid physiological thresholds such as CTmax, CTmin, and resistance to water loss (Brattstrom 1963; Keen

1984; Spotila 1972), whereas fine scale differences in physiology between species or

clades, and biotic interactions are more likely to impact the use of surface microhabitats.

Specifically, I focus on this third category of microhabitats. The salamanders found near

the surface will only represent a small fraction of the total population because individuals

also rely on underground retreats (Caceres-Charneco and Ransom 2010; Diefenbacher

2007; Drake et al. 2012; O'Donnell et al. 2016). However, the use of cover objects

provides benefits such as increased energy acquisition (Fraser 1976; Jaeger 1980). Being

at the surface during periods of inhospitable conditions, even if salamanders are

predominately inactive, may allow salamanders to move out from under cover to be

active for periods of time that conditions are favorable during the day and quicker access

to the surface at night. Furthermore, as surface microhabitats will provide subtler

variation compared to underground retreats, I expect the detection of differences between

species is facilitated when assessing salamanders that are using these surface retreats.

Farallo and Miles (2016) found that Plethodon cinereus, P. hubrichti, and P. sherando all 74 used microhabitats that differed from the overall available habitat. However, few studies have directly examined microhabitat and microclimate use at a small spatial scale for a wide range of plethodontids, consequently across a broad geographic range, in a field setting.

Here I present a study that addresses two hypotheses focusing on small-scale habitat use of plethodontid salamanders: 1) I hypothesize that plethodontids will use microclimates and microhabitats that differ overall from what is available. If plethodontid salamanders are able to seek out specific conditions at a small spatial scale, then the mechanism exists for differences to occur between species. 2) I also hypothesize that there will be interspecific variation in microhabitat and microclimate use.

This study highlights the importance of microhabitat for plethodontid salamanders including traditional thermal and hydric microclimate characteristics, but also measures of intermediate habitat characteristics that may influence salamander ecology. If salamanders are found in habitat that differs from the available habitat, then salamanders are seeking out specific microhabitats at a fine spatial and temporal scale. These data are especially relevant given the continued threat of climate change. Microhabitats offer buffers from ambient conditions, but long term climate change will inevitably alter suitability of microhabitats. Delineating the core microhabitats that species or clades of plethodontids use and if they are able to select specific microhabitats within their environment, will help us understand how climate change will impact communities.

These data will also provide insight into the role of habitat in the diversification of plethodontids and will serve as a base for studying the evolutionary history of small-scale 75 habitat use. More importantly, the data presented herein highlights the need to understand species ecology at a biologically appropriate scale, especially as making predications regarding species ability to adapt to swift and dramatic environmental changes becomes increasingly necessary.

Materials and Methods

Field Sites

This study was conducted over 59 field sites throughout Ohio, West Virginia,

Virginia, North Carolina, and Tennessee, specifically within Wayne, Monongahela,

Jefferson, George Washington, Pisgah, and Cherokee National Forests, and Blue Ridge

Parkway and Great Smoky Mountains National Parks (Fig. 11, Appendix 1). The field sites were chosen to include both an elevational and latitudinal gradient and include variation in approximate levels of species richness based on IUCN range maps (Fig. 11).

Microhabitat Data Collection

I assessed terrestrial habitat use of plethodontid salamanders. I include a direct comparison of habitat used by salamanders to habitat that was available but unused to determine if salamanders are actively selecting microhabitats. Data collection occurred between May 2012 and April 2015 during the day (0800 – 1700). I searched sites haphazardly for plethodontid salamanders by sifting through leaf litter and turning over cover objects such as coarse woody debris (CWD), fine woody debris (FWD), bark, and rocks. Once a salamander was found, it was placed in a plastic bag with leaf litter while habitat measurements were recorded. To initiate habitat measurements, I placed a 1-m2

frame around the capture point. A photo was taken of the plot and ground cover was 76 estimated by eye. Ground cover included green vegetation, rocks, coarse and fine woody debris (CWD and FWD), bark, leaf litter, and bare ground. The number and variety of cover objects within the plot was recorded, including microclimate data under each cover object. Air temperature (±0.5°C) and relative humidity (±3%) were both recorded using a

Kestral® 3500 weather meter and digital psychrometer at ground level and 50 cm and 1

m above ground level. Thermal and hydric characteristics of the soil were recorded from

five locations within each plot, at the center and each of the four sides. I measured soil

temperature using either a ThermaPlus thermocouple meter or an Infrared thermometer

(IRT) with a high sensitivity probe (±0.5°C; Thermoworks Inc.). I measured soil moisture using a HydroSense II (±3%; Campbell Scientific Inc.). Soil moisture and temperature probes were all inserted at approximately a 45° angle. I also recorded microhabitat data

under each cover object (one measurement per cover object) found in a plot including soil

temperature and soil moisture, as described above, and ground temperature using an

Infrared thermometer (±0.6°C; Thermoworks Inc.). Finally, to ensure no additional salamanders are located within the plot I completely sifted through the leaf litter and searched the remainder of the plot. This process was then repeated at a plot within 10 meters of the presence plot located in a random direction (established using a random number generator and a compass). The same measurements were taken at proposed absence site. I did not immediately search the plot to ensure that I did not alter microhabitat or microclimate conditions. If a salamander was found during the course of taking measurements or during the final search, a new absence plot was found and measurements were repeated again. Furthermore, to supplement my data for habitat 77 comparisons between species and clades I have also included additional data from salamander capture points. These data were collected in a similar manner, but only soil moisture, soil temperature, ground temperature, and relative humidity and air temperature at 1 m and ground level were recorded. For analyses comparing species and testing for phylogenetic signal of microclimate variables, if a salamander was active in the leaf litter

I used the plot data, however if the salamander was under a cover object I used the data for that specific cover object. This allows for a fine scale assessment of microclimate use and ensures the data collected best represents what each salamander was experiencing at the time of capture.

Although soil moisture is relevant to salamanders, it is hard to interpret these data without taking into account the soil texture that will impact the ability of salamanders to use the water in the soil. Therefore, I downloaded soil texture data for all sites including the percentage of sand, clay, silt, and organic matter from the USDA Geospatial Data

Gateway. I then used these data in conjunction with my soil moisture data using the formulas of Saxton and Rawls (2006) to calculate soil moisture tension. Soil moisture tension provides a measure of effort required to extract moisture from the soil, with -1500 kPa indicating that moisture is nearly completely unavailable and 0 kPa is fully saturated soil with moisture being freely available to be used. This is especially important when making comparisons over large geographic areas that may include substantial variation in soil types. Soils high in sand will have much more water available even at lower soil moisture content than soils high in clay. For example, sandy soil with 20% moisture content will have a soil moisture tension of -10 kPa, while a clay heavy soil will require 78 soil moisture content greater than 30% to have a comparable soil moisture tension as the sandy soil. Therefore, I will use soil moisture tension in lieu of soil moisture for all analyses in this chapter.

These data represent habitat use when salamanders are unlikely to be active, but will be utilizing leaf litter or cover objects to maintain appropriate thermal and hydric conditions which allow them to be close to the surface to maximize activity time when ambient conditions become more favorable especially during the evening. Furthermore, plots where salamanders are found will certainly not always contain salamanders, and absence plots may contain salamanders at other points in time. The purpose of these data is to demonstrate that salamanders use microhabitats that provide specific conditions, while avoiding other microhabitats. Specially I am using the definition of microhabitat proposed by Farallo (Chapter 1) that includes a small spatial and temporal scale as well as having a direct impact on the organism. These microhabitats represent areas of 1 m2 or

less and at a temporal scale of 1 hour or less. Therefore, expanding beyond this scale,

either spatially or temporally, may produce a different outcome, but consequently may

also not provide data that are biologically relevant to these salamanders.

Statistical Analyses

In order to assess specific components of the local environment I separated habitat

variables into two categories, 1) Microhabitat, and 2) Microclimate. Microhabitat

variables includes structural components of habitat at the 1 m2 scale such as leaf litter

depth, canopy cover, aspect, slope, and ground cover. Microclimate variables include soil 79 moisture tension, soil temperature, leaf litter moisture, and air temperature and relative humidity recorded 1-m above the ground.

All statistical analyses were conducted using the R statistical environment (ver.

3.3.1 http://www.r-project.org/). To determine if a plethodontid salamander chose specific microhabitats or microclimates I used a conditional logistic regression using the clogit function from the ‘survival’ package. I included a binary dependent variable to indicate whether a plot had a salamander. Each presence plot was paired with an absence plot as described above. For instances when I collected data on multiple presence plots with a single absence plot, the first presence plot was used for the paired comparisons. I used two analyses, one for each category of habitat variables, where the respective variables for the microhabitat and microclimate categories were included as the independent variables for the separate analyses. I removed plots with missing data, resulting in 534 plots and 268 matched pairs. Additionally, I did not include ground cover variables in the microhabitat category for this specific analysis as these would be inherently biased by the haphazard search for salamanders. For variables that differed significantly between presence and absence sites I used odds ratios to determine how each variable impacts salamander presence.

To determine if species are using different microhabitats and microclimates I used the function ‘adonis’ from vegan package to run a Permutational Multivariate Analysis of

Variance (PERMANOVA). For my multivariate response variable, I combined microhabitat and microclimate variables for this analysis. I used the absolute value of soil moisture tension for both the PERMANVOA and the NMDS analyses described below. 80

Therefore, when interpreting those results, larger values indicate drier soil and lower values indicate wetter soil which is the opposite of the raw data. I included species and site as factors as I wanted to determine if environmental variables differed between species, however I was also aware that there would be substantial variation among sites. I included an interaction between species and sites, but it was not significant so I removed it from the analysis. To visualize the differences between species, I include an NMDS analysis using the ‘metaMDS’ function from the vegan package. As NMDS analyses do not allow for missing data I reduced my dataset to only include salamanders that had data for all environment variables, resulting in 367 salamanders, for both the PERMANOVA and NMDS analyses. I performed the analyses with different subsets of the environmental variables that allowed for including more individuals and I found similar results. I used a K = 3 solution which had a stress of 0.17, however I only present data for axes 1 and 2. I plotted ellipses of the 95% confidence intervals of standard error for select species. Specifically, I include all species from the Plethodon glutinosus and Plethodon cinereus groups that had a minimum of 10 observations with measurements for all included microhabitat and microclimate variables. I also include ellipses for elevation and latitude categories to demonstrate spatial differences in species use of microhabitat and microclimate.

After finding that environmental variables differed significantly among species, I tested each habitat variable for phylogenetic signal. I used the phylogenetic tree from

Pyron and Wiens (2011). I pruned their tree to only include the species in this study and reduced all Eurycea species and Desmognathus species, besides D. wrighti, to individual 81 branches. I also grafted Plethodon sherando onto the tree based on the study by Bayer et al. (2012) which indicated this species was most closely related to P. serratus despite being geographically isolated by over 200 km. I made the tree ultrametric using the

‘chronos’ function in the APE package (Paradis et al. 2004). The mean value and standard error was calculated for each of the variables for each species/clade (Table 3). I used the phytools package (Revell 2012) to examine habitat values under a phylogenetic framework. I used the function ‘phylosig’ to compute Pagel’s λ including a log likelihood test for each calculation. Subsets of the full salamander dataset were used including the removal of juvenile salamanders and species with low sample sizes that may have influenced the results. If a variable had a λ > 0.5 and P < 0.05 for the log likelihood test I concluded that there is a significant phylogenetic signal for that variable within plethodontids. I also mapped the traits that displayed phylogenetic signal onto the phylogeny using the function ‘contMap’ to examine phylogenetic patterns that may be occurring.

Results

In total, 631 1 m2 plots were sampled, 306 absence plots and 325 presence plots. I encountered 518 salamanders from the family Plethodontidae including 18 completely terrestrial species, 17 of which were from the genus Plethodon as well as Desmognathus wrighti, the smallest and only completely terrestrial species from the genus

Desmognathus. I also encountered species that are partially terrestrial including

Psuedotriton ruber, Gyrinophilus porphyriticus, three species of Eurycea, and approximately eight other species of Desmognathus. However, species of the genus 82

Desmognathus are difficult to identify, especially in many of my field sites in North

Carolina and Tennessee where several closely related and phenotypically similar species overlap. Any potential for misidentification makes species level habitat assessment challenging for this specific study. Therefore, for my analyses that focus on comparing species all Eurycea and Desmonathus species other than D. wrighti were lumped into their respective genus. Although I am confident in my identification for most individuals

I did this because these genera are primarily aquatic and therefore will be less dependent on terrestrial habitat, which is the focus of this study (Table 3). Finally, for several analyses I will focus on species from specific phylogenetic groups including the

Plethodon cinereus group, Plethodon glutinosus group, and Plethodon wehrli-welleri groups (Table 3) classified from Fisher-Reid and Wiens (2011).

I found significant differences between presence and absence sites for both microclimatic and microhabitat variables (Table 4). Of the five microclimate variables, four of them significantly differed between presence and absence sites. For every increase in soil moisture tension of 100 kPa and 1 °C decrease in soil temperature I found an expected 7% and 54% increase in salamander occurrence, respectively. An increase in leaf litter moisture of 10% water content resulted in a 30% expected increase in salamander occurrence. For every 10% decrease in relative humidity there was an expected 52% increase in salamander occurrence. This difference is likely a consequence of other habitat variables and relative humidity positively correlates (r = 0.62) with higher temperatures, which is expected. There was no difference in air temperature between presence and absence plots. Of the four microhabitat variables only one significantly 83 differed between presence and absence plots. For every 1 cm increase in leaf litter depth I found an expected 37% increase in salamander occurrence. There was no difference in slope, aspect, or canopy cover between presence and absence plots.

I found that plethodontid species utilize different microhabitat and microclimatic

2 conditions (F22,290 = 6.87, P < 0.001, R = 0.20). The differences in environmental

variables between species accounted for approximately 20% of the variation. There was

also a significant difference in microhabitat and microclimatic variables between sites, as

would be expected, which accounted for approximately 40% of the variation (F54, 290 =

5.50, P < 0.001, R2 = 0.40). I plotted the species with 10 or more observations with

values for all included environmental variables for the Plethodon cinereus and P.

glutinosus species groups. There is distinct separation between three species of the P.

cinereus group (Fig. 12a). However, there is substantial overlap, between five members

of the P. glutinosus group (Fig. 12b). The P. cinereus group also appears to use more of a

moisture gradient as the three species presented range from negative to positive values

along the NMDS1 axis, which is heavily correlated with soil moisture tension, with

positive values on NMDS1 axis equating to drier conditions (Table 5).

Only two microclimate variables exhibited significant phylogenetic signal, soil

temperature (λ = 0.90, P < 0.01) and relative humidity (λ = 0.95, P < 0.01). I removed all

species with fewer than 10 individuals (13 tips) and re-calculated phylogenetic signal

using this reduced dataset, to ensure that species with low sample sizes were not biasing

my results. I found the same pattern of phylogenetic signal as I did in my complete

dataset and therefore I only include the results for the full dataset. Small-bodied species 84 from the P. cinereus group and P. wehrlei-welleri group use microclimates that are cooler, but also having lower relative humidity (Fig. 13). In contrast the larger bodied

Plethodon glutinous group were found in warmer microclimates with higher relative humidity (Fig. 13).

Discussion

This study has shown that plethodontid salamanders seek out specific microclimates and microhabitats. Salamanders are actively selecting the environment they experience and are able to refine these conditions through behavioral regulation even at a small spatial and temporal scale. These results are important because they confirm that salamanders are able to buffer variation in broad-scale environmental conditions through the use of small-scale habitat use, which will be relevant for forest management and for predicting the impacts that climate change will have on this taxonomic group.

Specifically, plots that had low soil moisture tension and temperature as well as high leaf litter moisture and increased depth were more likely to contain salamanders. Importantly, because these data were paired, I remove the potential bias of latitude, elevation, time, date, or weather as presence and absence plots were assessed under the same conditions for each comparison. This study echoes the findings of Farallo and Miles (2016) that showed Plethodon cinereus, P. sherando, and P. hubrichti use distinct microclimates, and that these species use microclimates that differ from random locations within central

Virginia. However, in this study, I have greatly expanded on these previous findings, comparing absence plots to paired presence plots for 17 Plethodon, 9 Desmognathus, 3 85

Eurycea, 1 Pseudotriton, and 1 Gyrinophilus species spanning a 580-km latitudinal gradient along the Appalachian Mountains (Appendix 2).

Using a combination of microclimate and microhabitat data I also showed that species differentially use the environment. Interestingly, species from the Plethodon cinereus group are distinctly separated in the NMDS space (Fig. 12a) whereas there little to no separation between species in the Plethodon glutinosus group (Fig. 12b). The P. glutinosus group is more clustered around the center, with P. jordani and P. glutinosus found in areas with lower soil moisture tension values. There are a few reasons why this might be the case. Species in the Plethodon glutinosus group are typically larger, including P. yonhalossee, the largest of the Plethodon species. Of the species included in my NMDS plot, P. glutinosus and P. cylindraceus are both larger than any member of the

P. cinereus group with a mean SVL, from this study, of 52 mm and 59 mm, respectively

(Appendix 2). This larger size may make them more susceptible to desiccation when exposed to extended period of desiccating conditions primarily due to the time amount of moisture required to rehydrate. Members of the P. glutinosus group have been shown to seek out higher moisture content soils in laboratory tests, while members of the P. cinereus group did not (Vernberg 1955). These results are further supported by Grover

(1998) that showed adult P. glutinosus were more prevalent in plots that included supplemental water, while only juvenile P. cinereus, but not adults, were more prevalent in plots with supplemental water. In this study, the mean soil moisture tension for individuals from the Plethodon glutinosus group was 415 ± 67 kPa while the mean values for individuals from the Plethodon cinereus group was 530 ± 40 kPa. These values are 86 above the level that any plethodontid salamander would be able to actively use moisture in the soil based on the work by Spotila (1972), which put the upper limit of water availability for plethodontids, based on 14 species, at approximately 283 kPa. From captures, 40% of the individuals from the Plethodon cinereus group and 70% of individuals from the Plethodon glutinosus group were found at soil moisture tension below this value, indicating they would have likely been able to use the moisture in the soil. Heatwole and Lim (1961) focused on P. cinereus and found they had a lower absorption threshold of 101 kPa. Using this cut off point, 32% of individuals from the

Plethodon cinereus group and 61% of individuals from the Plethodon glutinosus group were found at lower soil moisture tension conditions. These data suggest that species from the Plethodon glutinosus group may be more reliant on optimal moisture conditions to remain active at the surface, while species of the Plethodon cinereus group may be more tolerant to drier conditions. In terms of physiology within the P. glutinosus group, there is both inter- and intraspecific variation. Several studies have shown lower evaporative water loss (EWL) rates or higher resistance to water loss for low elevation populations of P. montanus (Winters and Gifford 2013) and P. metcalfi (Riddell and

Sears 2015; Chapter 4). Plethodon teyahalee found at both low and high elevations sites also have lower rates of EWL than its high elevation P. metcalfi congeners. In short, hydric focused physiology is variable in this group, and appears that this is partially due to maintaining appropriate hydric states, which may be an indication, in conjunction with the additional data presented, that moisture is especially important for this group of plethodontids. 87

Salamanders from high elevation/high latitude, low elevation/low latitude, and low elevation/high latitude fall at the extremes of the environmental spectrum (Fig. 12c).

The other geographic categories overlap substantially, indicating that salamanders may seek out specific environmental conditions, which is supported by my paired analyses of salamander presence and absence plots.

Only two of the microclimate variables exhibited strong phylogenetic signal, soil temperature and relative humidity. Specifically, the small bodied salamanders of the

Plethodon cinereus and Plethodon wehleri-welleri groups use cooler microclimates with lower relative humidity, while the species from the Plethodon glutinosus group use warmer microclimates with higher relative humidity (Fig. 13). This may be a further indication that species from the P. glutinosus group require conditions with high moisture content and maybe less dependent on temperature. Species from the P. cinereus group may be more dependent on cool temperatures which precludes them from being on the surface at times of the highest humidity during the day. However, these only include surface activity during the day, this is not to imply that the P. cinereus group would not be active on the surface at night during times of high humidity. Rather, it may be an indication of day time conditions they are unable to tolerate, which will obviously impact when and how long they are able to be active. If they are forced to retreat underground during the day this will also impact the time available to be active when ambient conditions become more favorable.

If species are phylogenetically constrained to specific habitat characteristics, then they are unlikely to be able to adapt to changing environmental conditions. In contrast, if 88 species are utilizing different microhabitats, then niche divergence at the micro-scale may play a role in speciation at local levels as well as providing a mechanism to mediate impacts of climate change. The data suggest that plethodontids can select specific environments and that there are species and potentially clade level differences. I have also shown that there is phylogenetic signal for soil temperature and relative humidity.

These data are important because they indicate that phylogenetic niche conservatism

(PNC) may be occurring at a much smaller spatial scale then previous research would suggest, at least for some aspects of habitat use. PNC of climatic niches has been proposed as a mechanism for the increased species richness at the mid-elevation as it prevents species from dispersing to lower or higher elevations (Kozak and Wiens 2010).

My data partially supports this at the micro-scale as a generalized pattern, many of the salamander’s captures included substantial overlap in environmental space even between disparate elevation and latitudes indicating that they may be seeking specific niches (Fig.

12c). However, in contrast, there is distinct separation between even closely related species. More analyses need to be done using micro-scale data to determine if PNC is observed in these salamanders or, unlike at the macro-scale climatic level, if species are locally adapting.

This study sheds light on habitat use of plethodontid salamanders at a biologically relevant spatial scale that is assessed across a large geographic area incorporating both inter- and intra-specific variation. These data tell us that plethodontid salamanders are able to choose environments which may contribute to differences in biotic interactions and possibly play a role in speciation that has occurred within this species rich family. 89

Fully understanding the ability to behaviorally select environments will be a critical component in understanding how these salamanders will be impacted by climate change.

My data has shown that plethodontids as a whole are clearly capable of selecting the environment that they experience, however, I have also shown that microhabitat and microclimate use varies between species. For future research, it will be important to know if these micro-scale niches are phylogenetically conserved or if species are able to adapt to novel environments at this spatial scale. If species can adapt then species may be able to evolutionarily mediate the impacts of climate change, at least partially. However, if species are phylogenetically restricted by the microhabitats and microclimates that they will be unable to persist in their current state as change occurs. One potential mechanism that would allow salamanders to mediate climate change despite evolutionary stasis would be behavioral modification, aka the Bogert effect (Huey et al. 2003). These data suggest that this is possible, as they are able to select specific environments, however; it is unclear if salamanders would be able to mediate long term climate change through simple habitat selection.

90

Tables

Table 3: Species captured during this study Included are the total number of individuals from each species captured, number of study sites in which they were captured, and the mean elevation the species was captured during this study. It should be noted that juvenile salamanders were removed for most analyses. Summery data with juveniles removed can be found in Appendix 2. Species Species Group N Sample Elevation Sites Plethodon cinereus 117 33 816 Plethodon hoffmani 18 7 646 Plethodon hubrichti 39 3 947 Plethodon cinereus Plethodon nettingi 9 1 1445 Group Plethodon richmondi 8 3 988 Plethodon serratus 26 9 761 Plethodon sherando 42 3 1044 Plethodon cylindraceus 27 16 780 Plethodon glutinosus 17 10 880 Plethodon jordani 30 4 1402 Plethodon glutinosus Plethodon metcalfi 11 2 1289 Group Plethodon montanus 22 7 1169 Plethodon teyahalee 3 2 941 Plethodon yonahlossee 2 2 1115 Plethodon wehrlei 10 3 1217 Plethodon Plethodon welleri 8 1 674 wehrlei- welleri Group Plethodon dorsalis 12 1 390 Desmognathus sp. 68 30 1147 Desmognathus wrighti 17 7 1380 Eurycea sp. 23 16 884 Pseudotriton ruber 6 3 559 Gyrinophilus porphyriticus 3 3 1275

91

Table 4: Results of the conditional logistic regression for microclimates and microhabitats Variable Odds Lower 95% Upper 95% Z-statistic P Ratio CI CI Microclimate Soil Moisture 1.001 0.99 1.00 -2.28 0.02 Tension Soil Temperature 0.582 0.39 0.88 -2.58 0.01 Leaf Litter Moisture 1.032 1.01 1.05 3.11 <0.01 Air Temperature 0.987 0.83 1.17 -0.15 0.88 Relative humidity 0.949 0.91 0.99 -2.68 <0.01 Microhabitat Leaf Litter Depth 1.448 1.21 1.74 3.96 <0.01 Slope 1.007 0.94 1.08 0.20 0.84 Canopy Cover 1.044 0.98 1.11 1.35 0.18 Aspect 1.002 1.00 1.01 1.20 0.23

Table 5: Correlation of the environmental variables to the first two NMDS axes. Higher R2 values indicate a greater relationship between the environmental variable and the NMDS axes. P < 0.05 are in bold and indicate that the data perform better than the randomly permutated data. Environmental Variable NMDS1 NMDS2 R2 P % Green Vegetation 0.64 0.77 0.20 <0.01 % Coarse Woody Debris -0.25 -0.97 0.54 <0.01 % Fine Woody Debris 0.12 0.99 0.03 <0.01 % Rock 0.70 0.72 0.31 <0.01 % Bare Ground 0.91 -0.42 0.08 <0.01 % Bark -0.01 1.00 0.13 <0.01 Canopy 0.65 0.76 0.13 <0.01 Aspect 0.92 -0.39 0.06 <0.01 Slope 0.92 -0.38 0.04 <0.01 Soil Moisture Tension -0.41 -0.91 0.10 <0.01 Leaf Litter Depth 0.91 -0.42 0.68 <0.01 Soil Temperature -0.99 -0.13 0.00 0.43 Ground Temperature -1.00 -0.08 0.00 0.49 Air Temperature -0.66 -0.78 0.06 <0.01 Relative Humidity 0.01 1.00 0.05 <0.01

92

Figures

Figure 11: Location of field sites. Sites (N = 59) are overlaid on an elevation map (above) from WorldClim, and a map of plethodontid species richness (below), calculated by summing the number of overlapping IUCN range maps for plethodontid salamanders.

93

94

Figure 12: Results of the NMDS analyses. Results from a non-metric multidimensional scaling analysis of microhabitat and microclimate data. I have plotted the 95% confidence ellipses of standard error in three separate panels, a) Plethodon cinereus group, b) Plethodon glutinosus group, c) data separated by elevation and latitude. Elevation: high > 1150 m, low < 700 m, and mid is between 1150 and 700 m. Latitude: high > 38°, low< 36°, and mid was between 38° and 36°. I only include ellipses for species with greater than 10 observations as they provide the easiest to interpret visualization. The elevation and latitude categories are included to provide an overview of the environment used by salamanders in each of these geographic spaces. The correlation of axes with environmental variables can be found in Table 5.

Figure 13: Microhabitat variables mapped on the phylogenetic tree. Mean values of relative humidity and soil temperature, the two variables that had significant phylogenetic signal, mapped onto the phylogenetic tree of plethodontids from this study.

95

CHAPTER 4: THE BOGERT EFFECT REVISITED: WILL BEHAVIORAL

COMPENSATION SAVE SPECIES IN A CHANGING CLIMATE?

Introduction

Determining the ability of a species to persist in fluctuating environments or expand their range in response to changing environmental conditions is a critical goal for evolutionary ecologists. The expected response of a species to fluctuating environments depends, in part, on specific niche requirements and tolerance to variation in key abiotic and biotic factors. For example, many species are ecological generalists and have broad tolerances for variation along abiotic or biotic niche axes, such as temperature, humidity or prey type. In contrast, other species exhibit restricted or specialized niche requirements. Moreover, specialization on one niche axis may be compensated by greater breadth on other niche axes.

Species with broad distributions, such as along a latitudinal or altitudinal gradients have populations that occupy environments characterized by substantial variation for numerous factors. Comparisons among populations may reveal either adaptation to local conditions, i.e., geographic variation in phenotypic traits such that trait values coincide with local optima, or plasticity in trait variation. Individuals exhibiting plasticity and inhabiting divergent habitat types along a gradient may select microhabitats that have similar values for key niche attributes. In contrast, species with limited distributions may have limited variation in phenotypic traits. Thus, a species may respond to environmental changes using alterations in behavior or physiological adaptions.

Reponses to changing environments: The Bogert effect and Behavioral Compensation 96

The term Bogert effect was coined by Huey et al. (2003), based on the pioneering research on behavioral thermoregulation by Charles Bogert (1949; 1959). Under this process, regulatory behaviors, such as thermoregulation or substrate selection, may buffer and ameliorate environmental variation while maintaining similar levels of physiological performance. This form of behavioral compensation allows individuals to exploit specific microhabitats that favor maintaining body temperatures without a diminishment in physiological performance across their range. Subsequent studies extended the Bogert effect to suggest that plasticity of thermal tolerance should be favored if the organism has limited capacity for regulatory behavior (e.g., Mitchell et al. 2013). Alternatively, other studies predicted that species using behavioral flexibility to minimize the magnitude of environmental variation could exhibit diminished capacities to use phenotypic plasticity to persist in changing environments (Marais and Chown 2008). In this study, I focus on the definition proposed by Huey et al. (2003). If the species use behavioral compensation to buffer changes in environmental conditions, then I expect that individuals in different parts of a species range should exploit a shift in microhabitat selection to attain optimal levels of performance. These shifts in microhabitat use may entail changes in spatial patterns of habitat exploitation, altering the timing of activity patterns, or a combination of the two. Thus, testing whether a behavioral compensation is occurring requires a study design that can detect spatial and temporal changes in microhabitat use.

The Bogert effect has the potential to allow species to persist in an environment challenged by a changing climate. The Bogert effect assumes an individual will alter their behavior to select microhabitats to buffer against suboptimal habitat. A critical 97 assumption of the Bogert effect is that behavioral compensation inhibits potential evolutionary responses to environmental change. Thus, rather than behavior driving evolution in trait values, the Bogert effect would predict evolutionary stasis in trait evolution given there will be little to no selection on physiological traits. Therefore, determining the potential for behavioral compensation to buffer species responses to climate change is a critical goal in evolutionary ecology. In addition, estimating how trait variation is affected by behavioral modification in habitat use will provides a test of behavior as a driver or inhibitor of evolutionary change. In addition, the Bogert effect assumes no cost to exploiting divergent habitats to maintain a physiological optimum state. Yet, morphological or behavioral modifications may be required to exploit new habitats. Rather than inhibiting trait evolution, there may be indirect (correlated) changes in traits that facilitate exploitation of habitats to maintain high levels of physiological performance.

Since the publication of Huey et al. (2003), several studies have attempted to demonstrate the Bogert effect occurring in nature. Drosophila subobsurca have been shown to use behavioral thermal regulation to mitigate thermal variation resulting in shifts of chromosomal inversion clines (Castañeda et al. 2013). Cybotoids, a clade of tropical Anolis lizards, use behavioral thermoregulation to maintain relatively constant diurnal body temperatures (Munoz et al. 2014). However, these lizards are exposed to cooler nocturnal temperatures that they are unable to buffer against, which results in selection for cold tolerance, but evolutionary stasis for heat tolerance (Munoz et al.

2014). In the light of climate change, which in most environments is resulting in 98 increasingly warmer temperatures, the inability to physiologically adapt to warmer temperatures will mean that behavioral compensation could prevent dramatic changes to species ranges through shifts or extirpations. Furthermore, acclimation appears to have minimal effects on thermal tolerances, which highlights the importance behavioral compensation in buffering against climate change (Gunderson and Stillman 2015).

Although accumulating evidence demonstrates that rising environmental temperatures is affecting most terrestrial environments, an underappreciated aspect of climate change is the concomitant modification of hydric environments. Not only is there evidence for a reorganization of precipitation regimes (Felzer and Heard 1999), but there are increases in evaporation as well. Thus, some regions are experiencing wetter conditions, but there are localities that are becoming hotter and drier. Amphibians, although often considered to be thermal conformers (Feder 1983; Feder and Lynch 1982;

Moore and Sievert 2001), have restricted thermal and hydric tolerances. Previous work has suggested that, at least some species, may actively regulate their temperature through microhabitat use (Camp et al. 2013; Farallo and Miles 2016). Therefore, the ability to use behavioral compensation, i.e. the Bogert effect, to locate optimal thermal patches, has the potential to help buffer amphibians from the multitude of environmental alterations that they will face as the climate changes. For amphibian, and especially salamanders, hydroregulation may also play a key role in how species are impacted by climate change.

Salamanders as a Model System

Plethodontidae is the largest family of salamanders with approximately 450 species. These salamanders are lungless and use cutaneous respiration, which requires 99 moist skin for gas exchange. Thus, hotter and drier conditions will affect their ability to remain surface active without succumbing to thermal stress or lose the capacity for gas exchange through evaporation. Thus, these plethodontid salamanders are especially sensitive to rising thermal conditions and drier hydric conditions. Habitat use by plethodontids varies tremendously and includes a broad diversity of microhabitat use including arboreal, subterranean, cave, and a full spectrum of aquatic to terrestrial microhabitats. This family primarily exists in the new world extending from southern

Canada south to Bolivia with the exception of six species found in Europe and a single newly discovered species in South Korea. The Appalachian Mountains in the Eastern

United States is home to approximately 55 species of Plethodontid salamanders, which accounts for over 12% of all plethodontid salamanders and 8% salamander species worldwide (AmphibiaWeb 2016). This hotspot also includes numerous endemic species that are restricted to a small number of mountains tops. The Big Levels Salamander

(Plethodon sherando), a microendemic from central Virginia, have been shown to use microhabitats that differ from closely related Eastern Red-backed Salamander (Plethodon cinereus) that completely surround the range of P. sherando and are extremely widespread (Farallo and Miles 2016). There are even differences in microhabitat use between P. sherando and P. cinereus from within 10 km of each other indicating there are refined uses of microhabitat occurring in Plethodontids (Farallo and Miles 2016).

The Interaction Between Thermal and Hydric Ecology

Some amphibians, such as many anurans, are large bodied and will routinely bask in the sun, especially species that shuffle between aquatic and terrestrial habitats. 100

However, both field studies and computer simulations have shown the core temperature of Northern Leopard Frog (Lithobates pipiens) does not differ from ambient temperatures despite these basking behaviors indicating that they are poor thermoregulators (Tracy

1976). However, amphibians have thermal limitations just as reptiles do, but most amphibian species also have strict hydric requirements. Although all species have some hydric requirements, amphibians are more constrained by water than ectothermic squamate counterparts. For instance, the Green Anole (Anolis carolinensis) has a cutaneous resistance to evaporative water loss of 196 s/cm while a comparably sized

Mountain Dusky Salamander (Desmognathus ochrophaeus) has a skin resistance of 0.9 s/cm (Lillywhite 2006; Spotila and Berman 1976). Therefore, amphibians must meet the challenge to maintain appropriate thermal and hydric conditions that allow them to maximize performance while also dealing with the constant interaction of the thermal and hydric environments. Furthermore, the process of thermoregulation in amphibians is more complex because it is influenced not only by behavioral modification through selection of specific microhabitats but also by physiological temperature regulation through evaporative water loss (Brattstrom 1979).

Salamanders and the Bogert Effect

Although the Bogert Effect originally emphasized behavioral compensation in thermoregulation, I extend the phenomenon to include any abiotic factor whose influence on physiological performance may be limited by shifts in habitat use. I posit that the interaction between thermal and hydric environments determines performance in plethodontid salamanders. Hence, behavioral accommodation entails selecting 101 microhabitats with favorable thermal and hydric conditions throughout a species range.

Species in the family Plethodontidae use a variety of terrestrial microhabitats including cover objects on the forest floor, such as rocks and woody debris (Petranka 1998), within the leaf litter (Petranka 1998), and recently it has become clear that that many species also routinely use arboreal habitats (Jaeger 1978; McEntire 2016; Niemiller 2005;

Regester and Samoray 2002; Trauth et al. 2000). These salamanders have both strict thermal and hydric requirements and are also capable of using microhabitats to buffer themselves from deleterious conditions including vertical movement within the soil

(Caldwell 1975; Caldwell and Jones 1973; Grizzell 1949; Hoff 1977; Seebacher and

Alford 2002). Many species also exist along elevational gradients exposing them to varying degrees of thermal and hydric extremes.

This study used the null model approach of Huey et al. (2003) to assess the ability of behavioral compensation to buffer variation in environmental conditions across an elevational gradient in plethodontid salamanders. I determined whether species select environments to attain similar thermal and hydric conditions at both low and high elevation sites. I address four hypotheses regarding the thermal and hydric conditions of microhabitats as well as microhabitat use by salamanders: 1) thermal and hydric conditions will vary between microhabitat types as well as differ spatially and temporally, 2) microhabitat use by salamanders should diverge to maintain thermal and hydric conditions by elevation and time of day, 3) Salamanders will use microhabitats that differ in thermal and hydric conditions compared to other available microhabitats, 102 and 4) Salamanders maintain similar thermal and hydric conditions at low and high elevations.

I selected field sites that ensured I would encounter the Red-cheeked Salamander

(P. jordani), Southern Gray-cheeked Salamander (P. metcalfi), and the Southern

Appalachian Salamander (P. teyahalee). All three of these species have small ranges compared to truly widespread Plethodon species such as P. cinereus and P. glutinosus, with P. jordani and P. metcalfi primarily restricted to high elevations and P. teyahalee restricted to lower elevations. However, I included all plethodontid salamander species encountered in my study. By assessing a variety of plethodontids I am able to make a more comprehensive assessment of the role behavioral compensation in maintaining thermal and hydric conditions. Finally, I use the thermal and hydric microhabitat data I collected to make predictions regarding how climate change will impact microhabitat environments including if optimal thermal and hydric will continue to exist in any capacity.

Materials and Methods

Field Sites

I set up sample plots at seven sites to assess microhabitat variation in thermal and hydric conditions available to salamanders. I established plots at four high elevation localities (1332 m ± 28 m) and three at low elevation localities (757 m ± 20 m). All sites were situated within the Great Smoky Mountains National Park (GSMNP) in North

Carolina and Tennessee. Habitats within the park support one of the highest salamander species richness in the world. My study focused on two salamander species restricted to 103 high elevation sites, the Red-cheeked Salamander (Plethodon jordani) and Southern

Gray-cheeked Salamander (P. meycalfi). My sample of low elevation species included the Southern Appalachian Salamander (P. teyahalee). However, I collected data on all plethodontid salamanders (Table 6). The large salamander population sizes and high diversity found within the park offered the best location to assess how climate change will impact plethodontid salamanders. I present data on the full set of salamanders, of which the three focal species accounted for 57% of my survey captures, as there was minimal difference when comparing species separately.

Agar Model Salamanders

I assessed water loss at different microhabitats by using agar models of salamanders (Fig. 14). Numerous studies have shown inter and intraspecific variation in water loss rates (Peterman et al. 2013; Riddell and Sears 2015; Winters and Gifford

2013). Therefore, my models are not intended to perfectly model water loss for salamanders, as there is certainly variation. However, my models represent an approximation of the hydric condition for each microhabitat, with models losing less water representing microhabitats that would result in less hydric stress. I present the hydric condition of microhabitats as mg of water loss per minute. The agar model salamanders were constructed from a solution of 40 g of pure Agarose (VWR Life

Science AMRESCO Agarose) per 1 L of water. Black food coloring was added to mimic the color of the three focal species. The solution was heated to boiling, allowed to cool and then poured into a salamander mold made of latex rubber. The molds were constructed using a anatomically correct rubber Red-cheeked Salamander as a model. 104

The agar models solidified in approximately 20 minutes at which point they were removed from the molds using latex gloves to avoid transferring human odor to the models. Although this does not impact the water loss rate, it does help reduce disturbance from animals that may be attracted to human odors.

Microhabitat Plots

At each site I located microhabitats (17-24) routinely utilized by plethodontid salamanders. I divided these microhabitats into six different categories, 1) above ground, which includes microhabitats at ground level exposed to the air (e.g. above leaf litter, coarse woody debris), 2) tree/shrub which includes animals captured at elevated microhabitats (where one would expect an increase in airflow) 3) under woody cover, 4) under rock, 5) under moss, and 6) under leaf litter. I placed a Thermochron iButton

(DS1922L) and Agar model salamander at each microhabitat within the site. The types of microhabitats used at each site depended on what was available, but always included leaf litter and some sort of woody cover object. I weighed agar models before they were deployed and then reweighed them in the morning, afternoon, and evening for three days.

If at any point during their deployment a model lost more than 20% of its initial mass, I replaced it with a new model to ensure that the water loss rate was consistent. Models were weighed to the nearest 0.0001 grams (Veritas - S123 - Precision Balance).

Furthermore, because all weighing took place in the field, I used a portable table, level, and wind/rain guard to ensure accurate weights even in inclement weather. In addition to assessing the thermal and hydric properties of microhabitats potentially used by salamanders, I also placed data loggers to assess ambient conditions at each site and 105 deployment. I placed an iButton in each of the four cardinal directions 1-m from above the ground attached to a tree in order to monitor air temp comparable to data used to make climate predictions. I also placed temperature and relative humidity data loggers

(HOBO Pro v2) at ground level above leaf litter at all sites and below leaf litter at select sites. Finally, I placed rain gauges at each site which were checked and emptied every time I weighed the agar models.

Salamander Surveys

I complemented the sampling of microhabitat thermal and hydric data with a concurrent assessment of salamander microhabitat use. Immediately before or after I weighed the agar model salamanders I completed an exhaustive survey for salamanders within 5 m2 plot near the site. The survey plot was chosen by walking a random distance

(1-20 m) in a random direction (1-360°). A minimum of two people completed each

survey by starting on opposite sides of the plot working towards the center. All cover

objects were flipped and all leaf litter was searched. Once a salamander was detected I

identified the species and held them until the end of the survey before being releasing it at

their point of capture. As surveys were completed in conjunction with the weighing of

agar models they occurred throughout the day and evening with the earliest morning

survey being completed at 7:15 AM and the latest night survey at 12:44 AM. This

provided us with a comprehensive assessment of microhabitat use by salamanders both

when they are active, typically at night, and when they are inactive under cover. I used

the microhabitats used by salamanders to partition the microhabitat plots into used and

unused for each survey period. Therefore, if salamanders were only captured under 106 woody debris during a survey, I used the temperature and water loss data from under woody debris as the used habitat data, and data from all other microhabitats to represent unused microhabitat temperature and water loss rates. In total, I completed 126 surveys and found 389 plethodontid salamanders.

EWL of Salamanders

To demonstrate potential differences in water loss rates between species or between low and high elevation populations at my study sites I include an assessment of evaporative water loss (EWL) for a low and high elevation population of Plethodon metcalfi and Plethodon teyahalee. I collected 10-12 individuals from each of these populations. The elevations differed for the two species, with Plethodon metcalfi being collected from 860 m and 1425 m, while P. teyahalee was collected from 635 m and

1140 m.; however, there was still an approximate 600 meter differences between low and high elevation for each species. I followed the methods of Winters and Gifford (2013) for my dehydration and rehydration experiments. I used experimental chambers constructed per the description in Winters and Gifford (2013). To begin each dehydration trial, the bladder of each salamander was expressed by gently pressing on their abdomen.

Salamanders were then weighed using an analytical balance to the nearest 0.0001 g. The salamanders were then placed in individual mesh pouches and weighed again.

Salamanders, while still in their pouch, were then placed into individual dehydration chambers. Every hour salamanders were weighed again until they had lost approximately

10% of their body mass, at which point they were removed from the pouch and immediately placed in a shallow dish with distilled water to rehydrate. Salamanders were 107 weighed every hour until they returned to their initial mass. All salamanders were returned to their point of capture at the end of the study.

Statistical Analyses

All analyses were conducted using the R computing environment (R Core Team

2016). To test whether thermal and hydric conditions vary between microhabitat types as well as differ spatially and temporally I calculated water loss as a mean change in mass of the agar models between successive measurement periods. I also noted the change in temperature between each weighing period. I categorized the sites as high (>1100 m; mean = 1332 ± 28) or low (<1000 m; mean = 757 ± 20). In addition, I noted the time each agar model was measured, with 3 levels: morning (4:00 am – 11:59 am), afternoon

(12:00 pm – 7:59 pm) and evening (8:00 pm – 3:59 am). I used linear mixed effect models using the ‘lme4’ package to determine differences among microhabitats. I created three models each assessing one of three response variables, mean temperature, water loss rate, and temperature amplitude. Each initial model included microhabitat type, elevation, and time of day as fixed effects including all possible interactions. I also included site and the date of sampling as random effects. I evaluated contribution of each factor to the overall model using the ANOVA function in the ‘lmerTest’ package and the

Satterthwaite approximation of degrees of freedom. To make comparisons between the different levels of significant effects I plotted the least square means and their standard error.

I used the census data to assess salamander microhabitat use including potential interactions based on elevation and time of day. I had few salamanders captured in some 108 categories, thus I retained only Surface (e.g. any salamander on the surface of any substrate), Under Leaf Litter, and Under Cover Object for the purpose of these analyses. I used generalized-linear mixed effect models using the ‘lme4’ package fitted with presence/absence of a salamander as a response variable and microhabitat type, elevation, and time of day as the fixed effects including the interaction of microhabitat and time of day and microhabitat and elevation. I also included site and date of the census as random effects. To determine if salamanders used microhabitats with different frequencies I repeated the analysis using count data from the surveys and completed the analysis under the Poisson distribution. As it is not possible to get a reliable estimate of the residual degrees of freedom for this analysis I made comparisons between microhabitats at different elevations and times of day solely by visually examining least square means and their standard errors generated from my model.

To determine if microhabitats used by salamanders differ in thermal and hydric conditions compared to other available microhabitats I used a conditional logistic regression using the clogit function in the ‘survival’ package to determine if salamanders were using microhabitats that differed in temperatures or water loss rates compared to other available microhabitats. The condition logistic regression examines matched pairs which allows us to compare used habitat to available but unused habitat (Breslow and

Day 1980; Compton et al. 2002; Hosmer and Lemeshow 1989). For each survey I calculated the mean temperature and mean water loss for the agar models within each microhabitat used by salamanders. For temperature at each microhabitat I used an iButton reading from within 20 mins of the survey. For water loss rates, I used the water loss rate 109 of agar models from the time of the survey until the following period the agar models were weighed. For example, water loss rates for the evening surveys were based on the change in mass of agar models from the evening survey to the following morning. I use this as an approximation for water loss as the surveys take place during those times that salamanders are active. For example, if I used the water loss rates from the time of the survey then I would be using water loss that includes time in the hot midafternoon to approximate water loss for the evening surveys which is unlikely to be biologically relevant for salamander microhabitat use at that time.

Finally, I used linear mixed effect models with temperature and water loss rates of used microhabitats generated from the survey data to determine whether: a) microhabitats used by salamanders (presence plots) differed in temperature or water loss rates at high and low elevation, b) presence microhabitats differed from absence microhabitats at each elevational belt, and c) absence microhabitats differed between high and low elevation. I used two models, with the response variable of temperature for one and water loss for the other. In both models elevation (High and Low) and occurrence (Presence or Absence) were treated as predictor variables. I included site, date, time of day, and survey ID as random effects.

I used an ANCOVA to compare dehydration and rehydration rates of P. metcalfi and P. teyahalee from low and high elevation populations. The dehydration or rehydration rate, expressed as EWL (mg cm-2 hr-1) was the response variable with

species, elevation, and their interaction as dependent variables. I included a reciprocal

transformation of initial mass, to normalize the data, as a covariate. 110

Results

I found temperatures are higher at low elevation and evening temperatures are lower than afternoon temperatures, as one would expect (Fig. 15; Table 7). The three-way interaction term for my model of mean temperature was significant which indicates that temperatures of microhabitats differ but this is dependent on spatial and temporal factors.

Specifically, there appears to be very little difference in temperatures between microhabitats with significant differences only apparent between above ground microhabitats compared to below rock microhabitats during the morning hours. The interaction between microhabitat type and elevation and microhabitat type and time was significant for the temperature amplitude and water models (Table 7). Temperature amplitude is higher at low elevation for tree/shrub habitats, above ground, below leaf litter, and woody debris, with amplitude being similar at high and low elevation for below moss and below rock (Fig. 16a; Table 7). There is lower temperature amplitude for all microhabitats during the morning. The temperature amplitude for tree/shrub and above ground microhabitats is greater in the afternoon than in the morning or evening (Fig. 16b;

Table 7). However, all the below ground microhabitats have similar temperature amplitudes in the afternoon and evening, and significantly lower amplitudes in the morning.

There was no difference in water loss rates between elevations among microhabitat types (Fig. 17a; Table 7). Water loss rates were similar at all times of the day in subsurface microhabitats (Fig. 17b; Table 7). However, rates of water loss for above ground microhabitats significantly exceeded subsurface values in the morning and 111 afternoon. Arboreal substrates manifested higher rates of water loss at high versus low elevation.

My analysis of salamander occurrence and relative abundance data produced similar results. Based on occurrence data, salamanders were found under leaf litter most frequently at low elevation sites, while at high elevation sites salamanders were more likely to be found under cover objects or leaf litter (Fig. 18a). However, more salamanders were captured under cover objects or leaf litter regardless of elevation (Fig.

18a). In both cases, salamanders were captured more frequently and in higher numbers above ground in the evening (Fig. 18b). Based on occurrence data salamanders were less likely to be found under cover objects in the evening, but I found no difference among habitats when using the frequency of use.

Based on my paired comparison I found that salamanders used microhabitats with significantly lower air temperature (odds ratio = 0.267, p < 0.001) and lower water loss rates (odds ratio = 0.394, p < 0.001). Both analyses suggest that salamanders are found in much higher frequency in cooler and wetter microhabitats. This contrasts with the unpaired analysis which found no difference in temperature between presence and absence microhabitats, but did show that temperature for both presence and absence microhabitats were lower at high elevation sites (Fig. 19a; Table 8). However, salamanders chose microhabitats that had significantly lower water loss rates compared to unused microhabitats at high elevation sites (Fig. 19b; Table 8). Furthermore, water loss rates between high and low elevation were statistically the same for presence sites

(Fig. 19b; Table 8). 112

Evaporative water loss rates varied by species and initial body mass. There was no statistical difference between elevations with both species maintaining statistically similar EWL rates for the low and high elevation populations (Fig. 20a; Table 9). Species significantly differed in EWL rates, but there was also a significant effect of initial body mass which most likely is the cause of the species level difference (Fig. 20a; Table 9). P. teyahalee had a significantly higher rehydration rate (Fig. 20b; Table 9). Rehydration rates did not differ between elevation (Fig. 20b; Table 9).

Discussion

My results suggest that salamanders selected microhabitats providing lower rates of water loss and cooler temperatures. However, despite using microhabitats with lower temperatures, salamanders were unable to maintain similar temperatures across elevation.

In contrast, salamanders were able to maintain a relatively constant water loss rate while they were active at both high and low elevations, which is consistent with the Bogert effect including maintenance of significantly lower water loss rates compared to unused microhabitats at high elevation. In Puerto Rico, Anolis cristatellus maintained cooler body temperatures at low elevation relative to null temperatures, and body temperatures below a null lizard at high elevations. Lizards at both elevations exhibited Tb's were

closer to their preferred temperature compared to null model temps (Huey et al. 2003).

When comparing my study to Huey et al. (2003) I expected a slightly different outcome

as my measures of performance differ in several regards. In my study, I use water loss rate as a proxy for physiological performance; however, my data estimates potential rates of water loss salamanders would experience among different microhabitats. Therefore, I 113 used a combination of data on habitat use, temperature, and approximate water loss in order to provide a similar assessment to that of Huey et al. (2003).

Salamanders are also more limited in their ability to thermoregulate. In contrast, a lizard can bask in the sun while a salamander is restricted to moving to a warmer, yet still shaded microhabitat. Therefore, I view my finding that salamanders selected cooler temperatures as a similar result to that of Huey et al. (2003) in that they appear to be actively selecting specific temperatures. Huey et al. (2003) also have a more traditional measure of performance; however, for salamanders maintaining moist conditions that allow them to breathe is critical for survival. In my study, I found that salamanders used microhabitats that would provide significantly lower water loss rates at high elevation.

Although, there was no difference at low elevation, the water loss rate of both used and unused microhabitats was comparable to the used microhabitats at high elevation.

Furthermore, salamanders consistently chose microhabitats that provided lower water loss rates based on the paired analysis from all of my surveys. My results are similar to the findings of Huey et al. (2003), where lizards maintained higher performance at both high and low elevation compared to null models. Therefore, it appears that the Bogert effect is occurring in plethodontid salamanders. Salamanders are selecting cooler microhabitats and more importantly microhabitats that offer lower rates of water loss.

I also found that dehydration and rehydration rates differed by species (Fig. 20). I did not find a significant effect of elevation; however, I had a low effect size for the comparison of elevation for both dehydration (-0.25) and rehydration (0.12) indicating that the sample size was not appropriate for comparing between elevations. Furthermore, 114 when I examine the data I see a clear trend for P. metcalfi to have lower dehydration rates at low elevation compared to high elevation. I feel this is in line with the work by Winters and Gifford (2013) and Riddell and Sears (2015), who found significant differences between low and high elevation populations for P. montanus and P. metcalfi, respectively. I found that P. teyahalee also has a lower dehydration rate and higher rehydration rate then P. metcalfi. These differences may be an indication that P. teyahalee, which is primarily found in low elevations, is adapted to the warmer and often drier conditions associated with low elevations. However, this partially contrasts with my results, as unused microhabitats had higher water loss rates than low elevation microhabitats. This may be a function of the sampling period, as I did not sample in the heart of summer or in fall. Conditions may vary by season, which would not be captured by my data. The lower water loss rates by P. teyahalee may be an adaptation to deal with extreme conditions they face even when not active, which also would not be picked up by my data. It may also be a phenotypically plastic trait that is influenced by some environmental cue. Furthermore, salamanders used microhabitats that maintained lower water loss rates in both high and low elevation indicating that salamanders still seek out optimal hydric microhabitats regardless of differences in the EWL they may experience.

The use of consistently wetter microhabitats implies that either not all species have either adapted or have a plastic physiology that allows for lower water loss rates in less optimal hydric conditions, or that all species have adapted in some capacity or have some level of plasticity, but this is not sufficient to allow them to completely negate the drier microhabitats present in the habitat. 115

My data also demonstrate that salamanders may not be strict thermal conformers as is traditionally thought. There are benefits for exploiting microhabitats at both warmer and cooler temperatures. Warmer temperatures are known to increase growth rates in salamanders, through an enhancement of foraging capabilities and higher digestive assimilation (Brattstrom 1963; Rome et al. 1992). However, warmer temperatures also increase the risk of desiccation, which has deleterious consequences for cutaneous respiration. Therefore, my observations suggest salamanders, and other terrestrial amphibians, engage in selecting habitats with temperatures that enhances physiological processes, but reduce the risk of desiccation. However, this does not mean that given the opportunity (e.g. when unconstrained by desiccation risk) that they will not seek out specific temperatures. Given desiccation results in death, salamanders can only maintain ideal temperatures when hydric conditions allow. An interesting aspect of this study is that it shows that, at least in the Great Smoky Mountains National Park, thermal properties of the environment are not a tremendous limiting factor. The values for presence and absence microhabitats both fall within the range of preferred temperatures for many species of plethodontid salamanders and none of the mean temperatures for any microhabitat at either elevation or time of day was close to critical thermal maximums for any species (Spotila 1972). This also explains why I do not see a more extreme difference in temperature between unused microhabitats and those used by salamanders. If this study was conducted in a location with less optimal habitat there may have been more variation in microhabitat temperatures especially in incidences where large deeply rooted cover objects offer more substantial protection from intense above ground thermal 116 environments. Farallo and Miles (2016) found salamanders in West Virginia below leaf litter at 20 °C while above the leaf litter temperature was over 50 °C. The high temperature, exacerbated by solar radiation, was mitigated by the thick layer of leaf litter.

Throughout my study I recorded 106,893 temperatures at microhabitats, of those recorded temperatures there were only 14 instances of the temperature exceeding 30 °C and 379 instances exceeding 25 °C which account for 0.01% and 0.35% of all total measurements respectively. Based on critical thermal maximum and thermal preference trials completed during this study (unpublished data) with Plethodon teyahalee and P. metcalfi as well as data from Spotila,(1972) and Brattstrom (1963),both of which assessed several species of plethdontid salamanders from the eastern United States, the thermal preference for this group is typically 12.0 to 25.6 °C. There is typically much less variation in critical thermal maximum which ranges from 31.5 and 34.8 °C. These data indicate that the temperature virtually never exceeded critical thresholds for salamanders at any microhabitat or point in time during the study and most microhabitats maintained a temperature preferred by many species of plethodontid. In fact, 89,204 (83.45%) of the recorded temperatures were under 20 °C yet salamanders still consistently selected slightly cooler microhabitats. Furthermore, salamanders are present in the Great Smoky

Mountains National Park almost year-round unlike many other areas in the eastern United

States where salamanders are more temporally restricted by warm and dry conditions, which means any microhabitat selection is most likely less important for salamanders at my study sites than in other locations that experience more extreme conditions making my results conservative estimates of the ability of salamanders to select microhabitats. 117

My data also unexpectedly highlights the importance of scale when addressing evolutionary and ecological questions. I found minimal differences in the thermal and hydric properties of microhabitats and in the use of microhabitats by salamanders, yet I see clear differences in the temperature and water loss rates of microhabitats used by salamanders at the time of my surveys. The differences in temperature are most likely a result of using temperatures from the exact time of capture, data which is lost when using the average temperature even over only hours. The difference in water loss rates between used and unused microhabitat is more complicated as I were using estimated water loss rates over 6-8 hours. The water loss at specific microhabitats are most likely effected by several factors such as temperature, rain fall, and permeability that interact to create water rates that varies temporally and at small spatial scales. Short periods of rain may wet above ground microhabitats, but below ground microhabitats may be kept dry by cover objects. Soaking rains may temporarily eliminate water loss from all microhabitats, however microhabitats under cover may experience longer term benefits, as moisture is maintained for increased periods of time after heavy rains. Here I see that fine scale differences in the timing of data collection can produce different, biologically relevant, results.

I conclude that the Bogert effect is occurring in regards to water loss in plethodontid salamanders. However, based on the data, it seems unlikely that microhabitat variation is sufficient to help them buffer themselves from the impacts of climate change in areas where habitat is not as optimal as it is in the Great Smoky

Mountains National Park. Although I stated earlier that temperatures recorded during this 118 study were low and nowhere near the critical thermal maximum for most plethodontids, increases in temperature will ultimately result in increased evaporation and water loss rates. Even in the Great Smoky Mountain National Park soil water storage is expected to decrease and evaporative deficit is predicted to increase despite relatively constant precipitation levels (Alder and Hostetler 2013; McCabe and Wolock 2011; Thrasher et al.

2013). In other areas of the eastern United States with lower species richness and habitat that is less suitable or simply less suitable for longer periods of time I might expect climate change posing a significantly larger problem for salamanders. Salamanders are clearly able to choose microhabitats, and this ability may help them maintain activity levels longer than would be possible otherwise, but ultimately microhabitats only offer a buffer for a short period of time. As temperatures continue to increase microhabitats will continue to warm and desiccate, decreasing the time and space available for salamander activity. Given salamanders are selecting cooler and wetter microhabitats it is less likely they will be able to adapt to these changes as there will not be selection for warmer and drier conditions if the salamanders are able to choose microhabitats that include reduced temperature and water loss rates. The Bogert effect offers salamanders a brief respite, but in the end, it will not shelter them from the effects of long term climate change.

119

Tables

Table 6: List of species captured during surveys Mean elevation ± standard errors (SE) is included. The majority of Desmognathus species were not identified to species as much of the study was conducted in area were identification is difficult. I also found at least one hybrid individual which exhibited traits of both Plethodon jordnai and P. teyahalee. Species N Elevation ± SE Desmognathus imitator 3 810 ± 1 Desmognathus monticola 2 846 ± 0 Desmognathus sp. 15 1150 ± 82 Desmognathus wrighti 45 1115 ± 42 Eurycea wilderae 64 926 ± 24 Plethodon jordani 130 1244 ± 26 Plethodon jordani x teyahalee 1 801 Plethodon metcalfi 35 1256 ± 36 Plethodon serratus 37 1058 ± 34 Plethodon teyahalee 56 1071 ± 20 Pseudotriton ruber 1 789 120

Table 7: Results of the three linear mixed effects models The models compare mean temperature, amplitude of temperature, and water loss rate. These models test for differences between microhabitats, elevation, time of day, and interactions when relevant. I included site and date that each plot was initiated as random factors. Degrees of freedom (df) were calculated using a Satterthwaite approximation. Comparison Sum of df F P Squares Mean Temperature Microhabitat 0.77114 5, 2245.85 55.74 <0.001 Elevation 0.03486 1, 5.66 12.6 0.013 Time of Day 2.61825 2, 2245.1 473.13 <0.001 Microhabitat x Elevation 0.03012 5, 2245.74 2.18 0.054 Microhabitat x Time of Day 0.56423 10, 2245.01 20.39 <0.001 Elevation x Time of Day 0.04839 2, 2245.32 8.74 <0.001 Microhabitat x Elevation x Time of Day 0.0795 10, 2245.01 2.87 0.001 Amplitude of Temperature Microhabitat 182.366 5, 2227.58 115.577 <0.001 Elevation 3.608 1, 1.89 11.432 0.083834 Time of Day 39.023 2, 2226.28 61.828 <0.001 Microhabitat x Elevation 4.933 5, 2227.07 3.127 0.008 Microhabitat x Time of Day 13.406 10, 2226.03 4.248 <0.001 Water Loss Rate Microhabitat 865.96 6, 2226.8 75.365 <0.001 Elevation 7.88 1, 2.7 3.43 0.1711 Time of Day 155.27 2, 2226.3 33.782 <0.001 Microhabitat x Elevation 120.5 5, 2227.4 10.487 <0.001 Microhabitat x Time of Day 241.54 10, 2226 10.511 <0.001

121

Table 8: Results of the two generalized linear mixed effects models The models compare mean temperature and water loss rate. These models test for differences between used (Presence) and unused (Absence) microhabitats, high and low elevation, and interactions between the two. I included site, date, and time of day as random factors. Degrees of freedom (df) were calculated using a Satterthwaite approximation. Comparison Sum of df F P Squares Mean Temperature Presence/Absence 0.016 1, 141.18 6.318 0.013 Elevation 0.035 1, 4.317 13.675 0.018 Presence/Absence X Elevation <0.001 1, 141.18 0.0003 0.985 Water Loss Presence/Absence 9.758 1,128.30 10.139 0.002 Elevation 2.180 1, 150.39 2.265 0.134 Presence/Absence X Elevation 3.817 1, 128.30 3.966 0.049

122

Table 9: Results of the ANCOVA comparing dehydration and rehydration rates Each ANCOVA compares of Plethodon metcalfi and Plethodon teyahalee from low and high elevation populations. I normalized mass data using a reciprocal transformation. Significant P values are in bold. Comparison df F P Dehydration Species 1,36 7.011 0.012 Elevation 1,36 0.740 0.402 Species X Elevation 1,36 0.432 0.522 Mass 1,36 13.010 0.001 Elevation X Mass 1,36 2.230 0.144 Rehydration Species 1,36 16.056 <0.001 Elevation 1,36 0.090 0.767 Species X Elevation 1,36 0.006 0.941 Mass 1,36 0.835 0.367 Elevation X Mass 1,36 0.501 0.484

123

Figures

Figure 14: A photo of an agar model salamander The agar model was used to assess water loss rates of microhabitats.

124

Figure 15: Least square means from a linear mixed effects model with mean air temperature as a response variable The analysis revealed a significant interaction between microhabitat type, time of day, and elevation. Standard error bars are included.

125

Figure 16: Least-squared means from a linear mixed effects model using temperature amplitude as a response variable There was a significant effect of the interaction between elevation and microhabitat type, and time of day and microhabitat type, but not for the full interaction term. A) Least square means of temperature amplitude versus the interaction between microhabitat and elevation. B) Least square means of temperature amplitude versus the interaction between time of day and microhabitat interactions. Standard error bars are included. 126

Figure 17: Least squared means from a linear mixed effects model using water loss rate as the response variable I observed significant interactions between elevation  microhabitat type and time of day  microhabitat type. I found no statistical support for the full interaction term. A) Least squared means for the microhabitat  elevation interaction. B) Least squared means for the time of day  microhabitat interactions. Standard error bars are included. 127

Figure 18: Least square means from the generalized linear mixed effect model of salamander occurrence There was a significant effect of the interaction between elevation and microhabitat type, and time of day and microhabitat type, but not for the full interaction term. A) The least square means for the microhabitat and elevation interaction. B) The least square means for the time of day and microhabitat interactions. Standard error bars are included. 128

Figure 19: Least square means from the linear mixed effect model of temperature and water loss rates The least square means from the linear mixed effect model of temperature (A) and water loss rates (B) separated by used (Presence) and unused (Absence) microhabitats at both high and low elevation. Standard error bars are included. 129

Figure 20: Mean dehydration and rehydration rates Dehydration (above) and rehydration (below) rates (mg cm-2 hr-1) for Plethodon metcalfi and Plethodon teyahalee collected from low and high elevation sites. Standard error bars are included. 130

CHAPTER 5: PHYLOGENETIC NICHE CONSERVATISM IN PLETHODONTID

SALAMANDERS: TAKING A SMALL-SCALE APPROACH

Introduction

The connection between eco- evolutionary dynamics is at the forefront of biological research (Hendry 2016). Ecological interactions can result in trait evolution, and conversely evolutionary changes or even stasis can be crucial in generating and maintaining ecological interactions. The evolutionary change is typically addressed in the literature more frequently because it provides an explanation for evolutionary patterns, which is often a goal in ecology (e.g. why are two species different). However, evolutionary stasis is not as readily apparent and often involves a complex understanding of the eco-evolutionary dynamics as is demonstrated by the recent discrepancies on identifying and quantifying these patterns (e.g. Losos 2008a; Losos 2008b; Wiens 2008).

Species traits may be phylogenetically constrained (Grafen 1989; Harvey and Pagel

1991) resulting in species that exhibit similar traits to their relatives or they may be free of constraints which allows species to adapt to novel conditions and substantially diverge from their close relatives (Losos and Miles 1994). Ultimately, phylogeny needs to be incorporated to make comparisons between species as evolutionary history will impact the interpretations of analyses.

Furthermore, as humans continue to alter the environment through habitat destruction and climate change, understanding what habitats species use, what climates they can tolerate, and how they interact with other species is crucial for predicting how species distributions and communities will be impacted. However, as has become 131 apparent in recent years, phylogenetic history can play a tremendous role in all of these facets of a species existence. Therefore, a unified approach that combines an ecological and the historical component of trait evolution, will be the most appropriate method for gauging species response to climate change.

A major goal in ecology is to understand how the interaction between a species and the biotic and abiotic components of the environment structures the range of habitats it occupies. One facet of ecology focuses on estimating key niche axes that govern the types of environments a species may exploit, the physiological limits within an environment, and the biotic interactions the constrain habitat occupancy (Chase and

Leibold 2003; Elton 1927; Grinnell 1917; Hutchinson 1957). There are several hypotheses that explain similarity of niches between closely related taxa including phylogenetic niche conservatism (PNC), phylogenetic time lags, and different adaptive responses (Harvey and Pagel 1991, pp. 38-48). One of the most prevalent in the literature is PNC. The outcome of PNC is similar to that of a phylogenetic constraint as closely related species would be expected to have similar niches in both instances. However, the mechanism differs between the two. A phylogenetic constraint often arises due to genetic road blocks that prevent traits from evolving, however a phylogenetic constraint can be any evolutionary outcome that occurs as a specific result of phylogenetic history and requires no adaptation (McKitrick 1993). On the other hand, PNC exists when closely related species have continually adapted to a similar niche over evolutionary time to the point that sufficient variation may no longer exist to allow decedents to move into novel niches (Lord et al. 1995). PNC can best be defined as the tendency of species to retain 132 niche characteristics through speciation events, resulting in closely related species with similar niches (Ackerly 2003; Cooper et al. 2010; Crisp and Cook 2012). In this regard clades are expected to exhibit PNC to be more similar than would be expected through the presence of phylogenetic signal alone (Losos 2008a). The pattern of PNC can emerge as a result of numerous mechanisms. The pattern of PNC can be dependent on co-existing species throughout evolutionary time (Lord et al. 1995): if more distantly related species are present and better suited to exist in adjacent niches, then the closely related species may be forced to continue to occupy the same niche, and further adapt to that niche because other niches were not available. This is a feedback loop that would reinforce

PNC over time. Similarly, physiological constraints could also force species to use similar niches, even though speciation events, resulting in patterns of PNC. If species are unable to expand into adjacent habitats, or disperse to novel habitats away from their current range, due to physiological limitations (e.g. CTmax, CTmin, hydric requirements)

then species may retain similar niches.

Despite the rising interest of PNC in the literature, there remain two substantial controversies regarding the concept. The first concerns the best approach for

identification of the occurrence of PNC (Losos 2008a; Losos 2008b; Wiens 2008).

Second, is the biological interpretation of phylogenetic niche conservatism. Moreover,

there is confusion regarding whether PNC is a pattern or a process. One of the key points

of confusion stems from the scale that PNC is tested. Losos (2008b) clarifies the degree

to which phylogenetic scale is important when considering PNC. Wiens (2008) uses an

example of Anolis lizards from Cuba (Knouft et al. 2006) which exhibit no phylogenetic 133 conservatism for climatic niches. However, there is clearly some phylogenetic constraint for climatic niches as all Anolis are restricted to tropical regions, highlighting the importance of phylogenetic scale when addressing PNC. As is true with all science, the question being asked will untimely guide what data are required, including the appropriate scale.

Climate, Habitat, and Spatial Scale of Phylogenetic Niche Conservatism

Climate and habitat are both topics of utmost importance as the planet experiences tremendous and continuous anthropomorphic changes. To assess the impacts of these changes, as well as continue to make ecological assessments of species, phylogenetic history needs to be included in analyses. PNC is especially relevant, as it indicates the occurrence of evolutionary stasis. If species have limited abilities to adapt to changing conditions, then they will not persist. Furthermore, when testing for PNC data need to be included at the appropriate ecological scale. Phylogenetic scale is clearly important

(Cavender-Bares et al. 2009; Losos 2008a), but what about spatial scale? PNC of climatic niches has been proposed for numerous taxa including invertebrates, vertebrates, and plants. PNC of temperate climatic conditions has been used to explain the increased diversity in freshwater arthropods (Moriniere et al. 2016). PNC has also been proposed as the primary role in promoting high species richness within the eastern United States

(Kozak and Wiens 2006) and also for explaining the prevalence of the mid-elevation peak in species richness (Kozak and Wiens 2010) for plethodontid salamanders. There is also evidence for PNC in higher plants, including pteridophytes, gymnosperms, and 134 angiosperms, where niches were retained by closely related species by higher degrees than would be expected by chance alone (Prinzing 2001).

However, most analyses of climatic niches are accomplished using data at a coarse spatial scale, that is, the data on temperature, precipitation and other variables are estimated at scales larger than the organism's home range, e.g., as invertebrates and salamanders. Downscaling regional climate conditions to a local scale is complicated by habitat structure that induces substantial microclimate heterogeneity (Farallo and Miles

2016; Suggitt et al. 2011). However, this variation is primarily temporal, and when examining microclimate conditions over a long period of time it is expected that even through the use of microhabitat conditions, that two extreme regional climates will still produce different microclimates despite substantial buffering from ambient conditions.

Therefore, to understand the importance of spatial scale and PNC requires knowing how organisms use microhabitat conditions and to what degree they are able to choose the conditions they experience.

The selection of specific microhabitats by a species will determine their exposure to specific microclimates (e.g., warm, dry or cool, mesic). Closely related species that share similar microhabitats may provide support for phylogenetic niche conservatism.

Alternatively, the selection of similar microhabitats may not result in exposure to a common microclimate. Similar to broad-scale climatic variables, phylogenetic conservatism of small-scale niches may inhibit species dispersal capabilities. For organisms that function at small spatial scales, inhospitable microhabitats may reduce or eliminate dispersal, even in locations that would appear to be optimal based on broad- 135 scale climatic data. In contrast, if there is no phylogenetic signal for these small-scale niche characteristics, this may be evidence of niche divergence which could result in local adaptation and interspecific diversification. If niche divergence is occurring this would mean that species may partition microhabitats spatially or temporally, either to maximize the microclimatic properties of microhabitats that is physiologically most beneficial or to avoid negative species interactions. Another scenario exists where there is a mix of PNC and recent divergence among microhabitat and microclimatic variables then species may be partially geographically constrained based on more broad climatic conditions, while diverging within local environments based on micro-scale habitat choices. If this is occurring species may be able to expand their ranges outside what would be expected based on climatic data alone. In short, micro-scale habitat and climate may play a bigger role in the evolutionary ecology of species that exist at these small spatial scales than is currently hypothesized.

Plethodontid Salamanders as a Model System

Plethodontids are a diverse group of lungless salamanders, existing primarily in

North and South America. There is tremendous variation in the range of species distribution, from wide spread species such as Plethodon cinereues, which occurs over

1.8 million km2 throughout northeastern United States and into Canada to the recently

described Urspelerpes brucei which is only known to exist over less than 21 km2 near the border of Georgia and South Carolina (Todd Pierson pers comm). There are also numerous other species that exist on restricted mountain tops with ranges less than 100 km2. All plethodontids also have extremely small home ranges making micro-scale 136 habitat and climate of primary importance to their existence. Moreover, these lungless species require cool and moist habitat to allow cutaneous respiration (Gatz et al. 1975;

Spight 1968; Spotila 1972; Wells 2007). The combination of restricted habitat use via small home ranges and narrow physiological requirements suggest that broad-scale environmental data have limited relevance for ecological questions involving these salamanders. Broad scale variables will, most likely, be relevant when making predictions about species approximate distributions. However, it is unlikely that broad- scale variables will adequately predict the edges of species ranges, as species may be able to use microhabitats to buffer themselves from ambient conditions. Broad scale variables are also unlikely to adequately distinguish between habitat use of overlapping species as there may be significant variation in microhabitat use that allows species to experience very difference microclimates even when found within the same geographic range. For plethodontid salamanders it is clear that they are able to actively select microhabitats, and that there are differences between species that are closely related (Chapter 3), and between species that have overlapping ranges (Farallo and Miles 2016).

Phylogenetic Niche Conservatism and Plethodontid Salamanders

As Kozak and Wiens (2006) detected significant phylogenetic signal of broad- scale climatic variables, my study examines phylogenetic signal of niches, using microhabitat and -climate variables to test for patterns of PNC. As these microhabitats and microclimates will undoubtedly interact, out study also include tests of phylogenetic signal for combined habitat and climate data. If PNC is occurring at small spatial scales, similar to PNC of broad-scale climatic variables, then species will be geographically 137 restricted, as they will be experiencing evolutionary stasis that limits adaptation to new conditions. Therefore, I also tested for phylogenetic clustering and phylogenetic overdispersion based on altitudinal and latitudinal geographic categories.

However, if there is no evidence of PNC at small spatial scales then species may be diverging at the local level promoting local species richness and providing the opportunity for local adaptation that may help mitigate climate change. Another possibility, considering individual components of habitat use as separate traits, they may diversify independently if there are different degrees of selection. In this instance, species would be expected to be partially restricted by the phylogenetically conserved habitat characteristics, while species are able to diverge habitat use within this restricted range, based on the phylogenetically unconstrained habitat characteristics.

Materials and Methods

Microhabitat Data Collection

Microhabitat data was collected from 59 field sites throughout Ohio, West

Virginia, Virginia, North Carolina, and Tennessee, specifically within Wayne,

Monongahela, Jefferson, George Washington, Pisgah, and Cherokee National Forests, and Blue Ridge Parkway and Great Smoky Mountains National Parks. I recorded microhabitat and microclimate data for 518 plethodontid salamanders (See Chapter 3); however, the total number used will vary by analysis as some lack specific data for some individuals.

Data collection occurred between May 2012 and April 2015 during the day (0800

– 1700). I searched sites haphazardly for plethodontid salamanders by sifting through leaf 138 litter and turning over cover objects such as coarse woody debris (CWD), fine woody debris (FWD), bark, and rocks. Once a salamander was found, it was placed in a plastic bag with leaf litter while habitat measurements were recorded. To initiate habitat measurements, I placed a 1 m2 frame around the capture point. A photo was taken of the

plot and ground cover was estimated by eye. Ground cover included green vegetation,

rocks, coarse and fine woody debris (CWD and FWD), bark, leaf litter, and bare ground.

The number and variety of cover objects within the plot was recorded, including

microclimate data under each cover object. Air temperature (±0.5°C) and relative

humidity (±3%) were both recorded using a Kestral® 3500 weather meter and digital

psychrometer at ground level and 50 cm and 1 m above ground level. Thermal and hydric

characteristics of the soil were recorded from five locations within each plot, at the center

and each of the four sides. I measured soil temperature using either a ThermaPlus thermocouple meter or an Infrared thermometer (IRT) with a high sensitivity probe

(±0.5°C; Thermoworks Inc.) and soil moisture using a HydroSense II (±3%; Campbell

Scientific Inc.). The values for soil moisture were converted to soil moisture tension because soil texture influences the ability of salamanders (and other organisms) to use the

moisture contained in the soil (see Chapter 3 for further explanation). Soil moisture and

temperature probes were all inserted at approximately a 45° angle. I also recorded

microclimate data under each cover object (one measurement per cover object) found in a

plot including soil temperature and soil moisture, as described above, and ground

temperature using an Infrared thermometer (±0.6°C; Thermoworks Inc.). Finally, to

ensure no additional salamanders are located within the plot I completely sifted through 139 the leaf litter and searched the remained of the plot. For all analyses of microclimate variables, if a salamander was active in the leaf litter I used the plot data, however if the salamander was under a cover object I used the data for that specific cover object. This allows for a fine scale assessment of microclimate use and ensures the data collected best represents what each salamander was experiencing at the time of capture.

Phylogenetic Tree and Phylogenetic Signal of Microhabitats and Microclimates

In order to test for phylogenetic signal I used the phylogenetic tree from Pyron and Wiens (2011). I pruned their tree to only include the species in this study and reduced all Eurycea species and Desmognathus species, besides D. wrighti, to individual branches. I also grafted Plethodon sherando onto the tree based on the study by Bayer et al. (2012) which indicated this species was most closely related to P. serratus despite being geographically isolated by over 200 km. I made the tree ultrametric using the

‘chronos’ function in the ape package (Paradis et al. 2004). The mean value and standard error was calculated for each of the variables for each species/clade (Table 10). To assess habitat and climate, as a whole, I completed three nonmetric multidimensional scaling

(NMDS) analyses, 1) only including microhabitat variables, 2) only including microclimate variables, and 3) combining all microhabitat and microclimate variables. I used ‘metaMDS’ function from the vegan package to complete these analyses. The dataset used for each NMDS analysis was reduced to include only individual salamanders that had data for all included variables resulting in 450 salamanders for microhabitat analysis, 487 salamanders for the microclimate analysis, and 367 salamanders for the combined analysis. I also further pruned the phylogenetic tree for the microclimate and 140 combined analysis to remove one species, Plethodon welleri, that had only a single observation after removing incomplete individuals, which prevent the inclusion of standard error. I then tested for phylogenetic signal of the NMDS axes 1-3 for both analyses.

I used the phytools package (Revell 2012) to examine the phylogenetic relationship of microhabitat and microclimate use. I used the function ‘phylosig’ to compute Pagel’s λ and Blomberg’s K including a log likelihood test and randomization test, respectively, for each calculation. Subsets of the full salamander dataset were used including the removal of juvenile salamanders, removing species with low sample sizes, and removing morphologically distinct species that may have influenced the results. If λ

> 0.5 and P < 0.05 for a respective habitat variables I considered it phylogenetically conserved. I mapped the traits that displayed phylogenetic signal onto the phylogeny using the function ‘contMap’ to examine phylogenetic patterns.

Phylogenetic Clumping or Overdispersion

I used the picante package to calculate nearest relative index (NRI) and nearest taxon index (NTI) for the captured species grouped into elevational and latitudinal categories. NRI values are associated with total relatedness, with positive values associated with phylogenetic clumping while negative values are associated with phylogenetic overdispersion (Webb et al. 2002). NTI values are interpreted the same way, however NTI is a measure of phylogenetic distance between each taxon in the sample to the closest related taxon in the sample. Therefore, NTI provides less information about deep phylogenetic relationships (Webb et al. 2002). Species were 141 grouped into all possible combinations of low, mid, and high elevation and latitude. For elevation, high > 1200 m, low < 700 m, and mid is between 1200 and 700 m. For latitude, high > 38°, low< 36°, and mid was between 38° and 36°. By creating these categories species were group by approximate geographic position and these groups should have similar general climates. Therefore, if phylogenetic clumping is detected within these groups, then this suggests that closely related species are more likely to exist within these within these climatically similar geographic locations than would be expected by chance alone. If there is phylogenetic overdispersion then these groups contain species that are more distantly related indicating that closely related species are occupying different climatic zones.

Results

For the independent microclimate variables, there is significant phylogenetic signal for soil temperature and relative humidity (Table 10; Fig. 21). For the independent microhabitat variables there was a significant phylogenetic signal for green vegetation based on the likelihood ratio test of lambda; however, this is very likely due to high levels of variation within and among species for this variable. The lambda value for green vegetation was also greater than 1 indicating rapid evolution near the tips, which is also likely due to the high levels of variation (Table 10). Blomberg’s K value for canopy cover and moss are also unusually high despite the randomization test indicating there is not a significant signal (Table 11). These also have very high variation between and among species, with select tips having much higher standard errors than other tips.

Furthermore, when lambda and K are calculated without incorporating standard error of 142 the measurements, there is not any phylogenetic signal for green vegetation and the lambda and K values are reduced for these variables with high levels of variation, while there is no change to the results for any of the other variables. Standard error was included in the calculations as this is likely important given the low sample size for some species and provide the most biologically relevant results. For the NMDS axes of microclimate, microhabitat, and all variables combined, there is no evidence of phylogenetic signal (Table 1). The first NMDS axis for microhabitat, similar to the independent microhabitat variables above, had a high K value that appears to be due solely to high levels of variation. There was also no evidence of phylogenetic clumping or overdispersion for any of the elevational and latitudinal groupings (Table 11).

Discussion

My study has shown that that microhabitat and microclimates appear to exhibit minimal levels of phylogenetic niche conservatism. There is no phylogenetic signal when examining these environmental variables as groups, and only two individual microclimates demonstrate any indication of phylogenetic signal (Fig. 21). These findings suggest that spatial scale is critically important when making ecological assessments rooted in evolutionary history. However, this is not to say that only small spatial scale is important, but rather that different spatial scales can provide us with different layers of information to help us understand the evolutionary ecology of a taxonomic group.

Within plethodontids there has been some controversy regarding the degree that

PNC occurs and how it impacts the evolutionary ecology of these organisms. Kozak and 143

Wiens (2006) proposed that PNC was primarily responsible for speciation in plethodontids. Specifically, climatic niches for allopatric sister taxa are comparable but they are separated by unsuitable climates within lowland habitats. Their results suggest that closely related species are restricted to similar niches, but have become isolated as climatic conditions have shifted over time resulting in climatic barriers to dispersal. PNC of climatic niches has also been proposed as the cause of the mid-elevation peak in species richness (Kozak and Wiens 2010). The peak in species at mid-elevation is ultimately caused by species inhabiting mid-elevation for a longer period of time, resulting in more species, and evolutionary stasis preventing species dispersal into low and high elevation environments. However, using different statistical techniques Wooten et al. (2013) found there was more evidence for phylogenetic divergence, specifically in the P. glutinosus group of plethodonditds. There are several possible reasons why these studies provide contradicting information. One reason is that Kozak and Wiens (2006) used BioClim climatic variables while removing ones that are highly correlated, while

Wooten et al. (2013) used all BioClim variables in addition to vegetation data all of which were reduced to principle component axes. Neither study include soil data in any capacity which may have importance for salamanders. Soil texture can have a tremendous impact on water availability to organisms (Saxton and Rawls 2006; Chapter

1). This is also true of vegetation, which will impact the conditions that organisms, such as salamanders, experience (Chapter 1 and 3). Another possible reason for the discrepancy between the two studies is phylogenetic scale. Kozak and Wiens (2006) explicitly examined parapatric sister taxa within several plethodontid groups, while 144

Wooten et al. (2013) only used the Plethodon glutinosus group. This difference may indicate different patterns within all plethodontidae.

This study has demonstrated a lack of phylogenetic signal for the majority of microhabitat and microclimatic variables and has also shown no phylogenetic clumping based on elevation and latitude, which lends support to a lack of PNC in this system.

However, a lack of phylogenetic signal does not always mean that there is no pattern of

PNC in the system. Evolutionary stasis can mask the connection between phylogeny and ecology (Hansen 1997; Martins and Hansen 1997; Revell et al. 2008). There are two sets of analyses that would help confirm the presence or absence of PNC of micro-scale environmental variables. One analysis would follow the method of Butler and King

(2004) which would allow me to compare micro-scale environmental data to several evolutionary models including white noise (WN), Brownian motion (BM), and three separate Ornstein-Uhlenbeck (OU) model. This approach was used to test for PNC of climatic niches at different elevations by Kozak and Wiens (2010). The four OU models would include one with a single global optimum and two that are based on previous analyses. As there is apparent phylogenetic signal for relative humidity and soil temperature that appears to be distinguished by differences in body size the best approach would be to create two models, one where small bodied salamanders are the ancestral state and one where large bodied salamanders are the ancestral state. Mean SVL from my study would be used to categorize the tips. Models would then be compared with the microhabitat and microclimate variables using sample size correct Akaike Information

Criteria (AICc), specifically relative humidity, soil temperature, and the three NMDS 145 axes for the combination of microhabitat and microclimate to determine which model provides the best fit. Another approach, which is arguably more appropriate, would be to explicitly compare matrices of species occurrence, phylogenetic distance, and micro- scale environmental characteristics (Ulrich et al. 2012). This method would allow us to test for PNC, as well as phylogenetic assortment, and habitat filtering. This method would provide a more detailed representation of the evolutionary ecology of plethodontid salamander; however, the data collected for this study, would not be appropriate for this analysis as it requires a species co-occurrence matrix which would be better developed from standardized survey methods.

Although my study cannot definitively say that there are no patterns of PNC present for micro-scale environmental variables, my analyses have demonstrated a lack of phylogenetic signal and provided evidence that no phylogenetic clumping is occurring based on elevation or latitude. These data highlight the uncertainty in the patterns of PNC proposed within plethodontids. Most importantly, my study has shown that spatial scale may play a critical role in these ecological and evolutionary patterns. The next step will be to make a simultaneous comparison of small-scale and broad-scale environmental variables which will help solidify the role scale plays in these important patterns.

146

Tables

Table 10: Values of Lambda and Blomberg’s K for all environmental variables Based on log-likelihood tests (Lambda) and randomization tests (Blomberg’s K), P- values less than 0.05 indicate significant phylogenetic signal. Variables marked with asterisk (*) indicate the values are not reliable due to the degree of variation for that specific variable. Variable Lambda P Blomberg’s K P Geographic Variables Elevation 0 1 0.53 0.19 Latitude 0 1 0.42 0.40 Microclimate Variables Air Temperature 0.50 0.30 0.64 0.08 Relative Humidity 0.95 <0.01 1.15 <0.01 Soil Temperature 0.90 <0.01 1.14 <0.01 Soil Moisture Tension 0 1 0.53 0.69 Leaf Litter Moisture 0.39 0.33 0.61 0.36 NMDS 1 0 0.99 0.55 0.61 NMDS 2 0.27 1 0.58 0.27 NMDS 3 0.20 1 0.79 0.54 Microhabitat Variables Canopy Cover* 0 1 3.50 0.23 Slope 0 1 0.46 0.44 Aspect 0 1 0.37 0.79 Leaf Litter Depth 0 1 0.46 0.91 Green Vegetation* 1.12 0.04 1.33 0.09 Moss* 0 1 4.02 0.74 Coarse Woody Debris 0 1 0.46 0.20 Fine Woody Debris 0 1 0.56 0.82 Rock 0.57 0.21 0.71 0.13 Bare Ground 0 1 0.66 0.87 NMDS 1* 0.63 0.49 1.31 0.08 NMDS 2 0.30 1 0.75 0.63 NMDS 3 0.07 1 0.40 0.98 Combined Microclimate and Microhabitat Variables NMDS 1 0.49 0.63 0.71 0.17 NMDS 2 0.84 1 0.58 0.46 NMDS 3 0.41 1 0.70 0.22

147

Table 11: Values of NRI and NTI calculated from elevation and latitude categories NRI refers to the Nearest Relatedness Index and NTI refers to the Nearest Taxon Index, both of which provide a measure of phylogenetic clumping or overdispersion. However, no combination of latitude and elevation was significant, indicating that neither phylogenetic clumping or dispersion is occurring. Elevation # of taxa NRI P NTI P High Latitude High 4 0.717 0.292 0.599 0.274 Mid 6 -0.123 0.502 0.391 0.337 Low 4 -0.672 0.687 -0.449 0.632 Mid Latitude High 7 0.559 0.281 0.426 0.334 Mid 13 -0.441 0.663 0.358 0.343 Low 6 -1.045 0.838 -0.596 0.719 Low Latitude High 7 -1.361 0.934 -0.823 0.788 Mid 7 -0.298 0.554 0.274 0.386 Low 5 -0.744 0.741 -1.096 0.864

148

Figures

Figure 21: Microhabitat variables mapped on the phylogenetic tree Mean values of relative humidity and soil temperature, the two variables that had significant phylogenetic signal, mapped onto the phylogenetic tree of plethodontids from this study.

149

CHAPTER 6: CONCLUSIONS

Since the term ecology was coined by Haeckel (1866), the field has increasing expanded as researchers have explored new ways to study biotic and abiotic interactions.

Ecology became a prominent field of study in the early 20th century. At this time Joseph

Grinnell (1917) introduced ecology to the concept of the niche and began thinking about

species habitat requirements while Charles Elton (1927) focused on biotic interactions.

George Evelyn Hutchinson (1957) expanded on the niche concept, combining both biotic

and abiotic factors into an n-dimensional hyper volume. These older definitions of the

ecological niche all have a degree of vagueness that makes them hard to quantify and

may contribute to why they have been used less frequently in the literature. More recently

Chase and Leibold (2003) have attempted to modify the concept of the ecological niche

to include more tangible and measurable data that can be incorporated into mechanistic

models. The niche concept is important because it allows researchers to discuss

quantifiable aspects of species habitat use which provides a means to assess causes of

species distributions, community interactions, and species diversity.

One consistent issue within the field of ecology, especially when attempting to

quantify a species niche, is the concept of scale. The scale at which data is collected can

influence productivity-biodiversity relationships (Chase and Leibold 2002),

interpretations of species distributions (Chapter 1), distribution-abundance relationships

(Werner et al. 2014), species richness patterns (Rahbek 2005), and many others. The

importance of scale in ecology has been discussed at length (Levin 1992; Levin 1993;

Morris 1987; Wiens et al. 1986). However, including small-scale environmental data is 150 difficult, especially when making assessments over broad geographic areas. Therefore, despite knowing that spatial and even temporal scale can influence ecological processes small-scale data is often ignored. Just as important, most organisms exist at a small spatial scale (1 m2), and therefore broad-scale environmental data (≥ 1 km2) is less likely

to be biologically relevant for these organisms (Suggitt et al. 2011).

This dissertation has focused on plethodontid salamanders, which exist primarily

at a small spatial scale, and examines their use of microhabitats and microclimates, while

also making comparisons to studies that have used broad-scale climatic data to make

interpretations about the groups evolutionary ecology. The dissertation as whole completed three general tasks, 1) it provided detailed information on plethodontid salamander use of microhabitat and microclimates including interspecific comparisons at

a broad geographic scale, 2) it highlighted the importance of micro-scale environmental

data for addressing evolutionary and ecological questions, and 3) it examined the role of

microhabitats in mitigating the effects of climate change.

The data presented in this dissertation have demonstrated the ability of species to

behaviorally regulate the microhabitats they use and the microclimates they experience,

specifically selecting for select thermal and hydric conditions. Specially I have shown

that, plethodontids as a whole are able to choose cooler and wetter microhabitats (Chapter

3), specific species, particularly microendemics, differ from other plethodontid species

(Chapter 2 and 3), and that species are able to use microhabitats that maintain water loss rates across an elevational gradient (Chapter 4). These results build upon previous research and helps unite field and laboratory studies. Many studies have shown 151 differences in physiological performance and constraints in the lab, such as resistance to water loss, CTmax, CTmin, and thermal preference (Brattstrom 1963; Brattstrom 1979;

Hutchison 1961; Riddell and Sears 2014; Riddell and Sears 2015; Spight 1968; Spotila

1972; Winters and Gifford 2013). Studies have also examined microhabitat use through

field surveys over relatively small geographic areas (Farallo and Miles 2016; Grover

1998; Jaeger 1980; Keen 1984; O'Donnell et al. 2014; Petranka and Smith 2005).

However, to my knowledge, no studies, besides this dissertation, have connected habitat

and climate at a small spatial scale combined with comparisons across broad geographic

area. This study demonstrates that salamanders are able to use specific microhabitats in

order to find the most suitable microclimate. Given previous studies have shown

differences in physiology both between species (e.g. Spotila 1972) and within species

(e.g. Riddell and Sears 2014), the data in this dissertation that demonstrates these

salamanders are capable of choosing the environment they experience in the field

provides a mechanism for physiology to structure communities, potential for sympatric

speciation, and buffering against the impacts of climate change.

Even more importantly, this dissertation is a stepping stone for the increased

inclusion of small-scale environmental data into large scale questions of evolutionary

ecology. Much of previous studies have used broad-scale climatic data to make

inferences regarding ecological patterns across phylogenies (e.g. Hawkins et al. 2007;

Kozak and Wiens 2006; Wiens and Graham 2005). For some organism, broad-scale are

likely to be biologically relevant, as their size or dispersal capabilities make habitat at a

small spatial scale less important. However, as has been discussed at length in this 152 dissertation, most organisms exist at small spatial scale which will impacted by habitat and climate heterogeneity only perceivable at small-scales (Suggitt et al. 2011). I have shown that there is minimal phylogenetic signal for microhabitat and microclimates used by plethodontid salamanders (Chapter 5) despite studies pointing to patterns of phylogenetic niche conservatism as causes of peaks of diversity at midelevations (Kozak and Wiens 2010) and speciation (Kozak and Wiens 2006). A lack of phylogenetic signal may indicate that diversification, based on variation in use of small-scale environmental characteristics, plays a role in this groups phylogenetic history. More recent studies have suggested diversification of climatic variables may have promoted specification in the

Plethodon glutinosus group (Wooten et al. 2013). Although more research is currently needed, the data within this dissertation suggest that both regional and local scale habitat and climate variables interact to create different patterns that will only be detectable at certain scales. For example, niche conservatism of climatic characteristics may promote speciation, but diversification of microclimatic variables may bolster these speciation events. It is possible that speciation only occurs because of this interaction between local and regional processes. However, to truly test these relationships, they need to be simultaneously assessed, ideally with both local and regional species richness assessments combined with microclimates assessed over a broad geographic area that can be compared to traditional climatic data. Understanding the connection between these ecological scales in the context of phylogenetic history will be crucial for assessing species abilities to cope with changes to climate and habitat, both of which are occurring at an alarming rate. 153

In addition to expanding on our knowledge of plethodontid salamander evolutionary ecology, this dissertation has shed light on the role small-scale habitats and climates play in the maintenance of species distributions, community structure, and in helping species buffer themselves from climate change and other anthropogenic habitat alterations. The data herein also suggest that divergence of microhabitat and microclimate use may exist for plethodontid salamanders, despite previous studies indicating patterns of phylogenetic niche conservatism of climatic niches. This dissertation offers a substantial contribution to the literature on plethodontid salamander ecology and the importance of scale for assessing the impacts of climate change rooted both in ecology and evolutionary history. Overall, this work will hopefully promote further research that incorporates small-scale environmental data when addressing questions across broad geographic areas and lead to more unified studies that include a variety of spatial scales.

154

REFERENCES

Aarts, G., J. Fieberg, S. Brasseur, and J. Matthiopoulos. 2013. Quantifying the effect of habitat availability on species distributions. Journal of Animal Ecology 82:1135- 1145.

Ackerly, D. D. 2003. Community assembly, niche conservatism, and adaptive evolution in changing environments. International Journal of Plant Sciences 164:S165- S184.

Adams, D. C., and F. J. Rohlf. 2000. Ecological character displacement in Plethodon: Biomechanical differences found from a geometric morphometric study. Proceedings of the National Academy of Sciences of the United States of America 97:4106-4111.

Adolph, S. C. 1990. Influence of behavioral thermoregulation on microhabitat use by two Sceloporus lizards. Ecology 71:315-327.

Alder, J. R., and S. W. Hostetler. 2013. USGS National Climate Change Viewer. US Geological Survey.

Allee, W. C. 1931. Co-operation among animals. American Journal of Sociology 37:386- 398.

AmphibiaWeb. 2016. http://amphibiaweb.org. University of California, Berkeley, CA, USA.

Anthony, C. D., C. A. M. Hickerson, and M. D. Venesky. 2007. Responses of juvenile terrestrial salamanders to introduced (Lithobius forficatus) and native centipedes (Scolopocryptops sexspinosus). Journal of Zoology 271:54-62.

Anthony, C. D., M. D. Venesky, and C.-A. M. Hickerson. 2008. Ecological separation in a polymorphic terrestrial salamander. Journal of Animal Ecology 77:646-653.

Anthony, C. D., J. A. Wicknick, and R. G. Jaeger. 1997. Social interactions in two sympatric salamanders: Effectiveness of a highly aggressive strategy. Behaviour 134:71-88.

Arif, S., D. C. Adams, and J. A. Wicknick. 2007. Bioclimatic modelling, morphology, and behaviour reveal alternative mechanisms regulating the distributions of two parapatric salamander species. Evolutionary Ecology Research 9:843-854.

Ash, A. N., and R. C. Bruce. 1994. Impacts of timber harvesting on salamanders. Conservation Biology 8:300-301. 155

Bailey, L. L., T. R. Simons, and K. H. Pollock. 2004a. Estimating detection probability parameters for plethodon salamanders using the robust capture-recapture design. Journal of Wildlife Management 68:1-13.

—. 2004b. Estimating site occupancy and species detection probability parameters for terrestrial salamanders. Ecological Applications 14:692-702.

—. 2004c. Spatial and temporal variation in detection probability of plethodon salamanders using the robust capture-recapture design. Journal of Wildlife Management 68:14-24.

Barton, K. 2013.Package 'MuMIn'. http://cran.rproject.org/web/packages/MuMIn/index.html

Bates, D., M. Maechler, B. Bolker, and S. Walker. 2014.lme4: Linear mixed-effects models using Eigen and S4. R package version 1.1-7, http://CRAN.R- project.org/package=lme4

Baust, J. G. 1976. Temperature buffering in an arctic microhabitat. Annals of the Entomological Society of America 69:117-119.

Bayer, C. S. O., A. M. Sackman, K. Bezold, P. R. Cabe, and D. M. Marsh. 2012. Conservation genetics of an endemic mountaintop salamander with an extremely limited range. Conservation Genetics 13:443-454.

Bell, D. M., and D. R. Schlaepfer. 2016. On the dangers of model complexity without ecological justification in species distribution modeling. Ecological Modelling 330:50-59.

Belmaker, J., and W. Jetz. 2012. Regional Pools and Environmental Controls of Vertebrate Richness. The American Naturalist 179:512-523.

Bernardo, J., and J. R. Spotila. 2006. Physiological constraints on organismal response to global warming: mechanistic insights from clinally varying populations and implications for assessing endangerment. Biology Letters 2:135-139.

Best, M. L., and J. H. H. Welsh. 2014. The trophic role of a forest salamander: impacts on invertebrates, leaf litter retention, and the humification process. Ecosphere 5:1- 19.

Blanquart, F., O. Kaltz, S. L. Nuismer, and S. Gandon. 2013. A practical guide to measuring local adaptation. Ecology Letters 16:1195-1205.

Bogert, C. M. 1949. Thermoregulation in reptiles, a factor in evolution. Evolution 3:195- 211. 156

—. 1959. How reptiles regulate their body temperture. Scientific American 200:105-120.

Brattstrom, B. H. 1963. A preliminary review of the thermal requirements of amphibians. Ecology 44:238-255.

—. 1979. Amphibian temperature regulation studies in the field and laboratory. American Zoologist 19:345-356.

Braun-Blanquet, J. 1932, Plant sociology (Transl. G. D. Fuller and H. S. Conrad). New York, McGraw-Hill.

Breslow, N. E., and N. E. Day. 1980, Statistical methods in cancer research. Volume I: The analysis of case-control studies. Lyon, France, International Agency for Research on Cancer.

Brown, J. H. 1984. On the relationship between abundance and distribution of species. The American Naturalist 124:255-279.

Buckley, L. B., M. C. Urban, M. J. Angilletta, L. G. Crozier, L. J. Rissler, and M. W. Sears. 2010. Can mechanism inform species' distribution models? Ecology Letters 13:1041-1054.

Bury, R. B. 1983. Differences in amphibian populations in logged and old growth redwood forest. Northwest Science 57:167-178.

Butler, Marguerite A., and Aaron A. King. 2004. Phylogenetic comparative analysis: A modeling approach for adaptive evolution. The American Naturalist 164:683-695.

Caceres-Charneco, R. I., and T. S. Ransom. 2010. The influence of habitat provisioning: use of earthworm burrows by the terrestrial salamander, Plethodon cinereus. Population Ecology 52:517-526.

Caldwell, R. S. 1975. Observations on the winter activity of the Red-backed Salamander, Plethodon cinereus, in Indiana. Herpetologica 31:21-22.

Caldwell, R. S., and G. S. Jones. 1973. Winter congregations of Plethodon cinereus in ant mounds, with notes on their food habits. The American Midland Naturalist 90:482-485.

Camp, C. D. 1988. Aspects of the life history of the Southern Red-back Salamander (Plethodon serratus Grobman) in the southeastern United States. The American Midland Naturalist 119:93-100.

Camp, C. D., J. A. Wooten, J. B. Jensen, and D. F. Bartek. 2013. Role of temperature in determining relative abundance in cave twilight zones by two species of lungless salamander (family Plethodontidae). Canadian Journal of Zoology 92:119-127. 157

Caruso, N. M., M. W. Sears, D. C. Adams, and K. R. Lips. 2014. Widespread rapid reductions in body size of adult salamanders in response to climate change. Global Change Biology 20:1751-1759.

Castañeda, L. E., J. Balanyà, E. L. Rezende, and M. Santos. 2013. Vanishing chromosomal inversion clines in Drosophila subobscura from Chile: is behavioral thermoregulation to blame? The American Naturalist 182:249-259.

Cavender-Bares, J., K. H. Kozak, P. V. A. Fine, and S. W. Kembel. 2009. The merging of community ecology and phylogenetic biology. Ecology Letters 12:693-715.

Chase, J. M., and M. A. Leibold. 2002. Spatial scale dictates the productivity-biodiversity relationship. Nature 416:427-430.

Chase, J. M., and M. A. Leibold. 2003, Ecological niches: Linking classical and contemporary approches, University of Chicago Press.

Chen, P. F., E. O. Wiley, and K. M. McNyset. 2007. Ecological niche modeling as a predictive tool: silver and bighead carps in North America. Biological Invasions 9:43-51.

Clements, F. E. 1916, Plant succession: an analysis of the development of vegetation. Washington, Carnegie Institution of Washington.

Compton, B. W., J. M. Rhymer, and M. McCollough. 2002. Habitat selection by wood turtles (Clemmys insculpta): An application of paired logistic regression. Ecology 83:833-843.

Cooper, N., W. Jetz, and R. P. Freckleton. 2010. Phylogenetic comparative approaches for studying niche conservatism. Journal of Evolutionary Biology 23:2529-2539.

Corn, P. S., and R. B. Bury. 1989. Logging in western oregon: Responses of headwater habitats and stream amphibians. Forest Ecology and Management 29:39-57.

Crisp, M. D., and L. G. Cook. 2012. Phylogenetic niche conservatism: what are the underlying evolutionary and ecological causes? New Phytologist 196:681-694.

Daubenmire, R. 1959. A canopy-coverage method of vegetational analysis. Northwest Science 33:43-64.

Davis, D. R., and G. B. Pauly. 2011. Morphological variation among populations of the Western Slimy Salamander on the Edwards Plateau of Central Texas. Copeia 2011:103-112.

Davis, M. B., and R. G. Shaw. 2001. Range shifts and adaptive responses to quaternary climate change. Science 292:673-679. 158

De'Ath, G. 2002. Multivariate regression trees: a new technique for modeling species- environment relationships. Ecology 83:1105-1117.

Diefenbacher, E. H. 2007. Plethodon glutinosus (Slimy Salamander). Burrowing behavior. Herpetological Review 38:67-68.

Dobrowski, S. Z. 2010. A climatic basis for microrefugia: the influence of terrain on climate. Global Change Biology 17:1022-1035.

Dodd, C. K., and R. M. Dorazio. 2004. Using counts to simultaneously estimate abundance and detection probabilities in a salamander community. Herpetologica 60:468-478.

Douglass, R. J. 1976. Spatial interactions and microhabitat selections of two locally sympatric voles, Microtus montanus and Microtus pennsylvanicus. Ecology 57:346-352.

Downes, S. 2001. Trading heat and food for safety: Costs of predator avoidance in a lizard. Ecology 82:2870-2881.

Drake, D. L., K. O'Donnell, and B. Ousterhout. 2012. Plethodon serratus (Southern Red- backed Salamander). Cicada burrow use. Herpetological Review 43:318-319.

Elton, C. 1927. Animal Ecology. London, Sidgwick and Jackson.

Farallo, V. R., and D. B. Miles. 2016. The importance of microhabitat: a comparison of two microendemic species of Plethdoon to the widespread P. cinereus. Copeia 104:67-77.

Feder, M. E. 1982. Thermal ecology of neotropical lungless salamanders amphibia plethodontidae environmental temperatures and behavioral responses. Ecology 63:1665-1674.

Feder, M. E. 1983. Integrating the ecology and physiology of plethodontid salamanders. Herpetologica 39:291-310.

Feder, M. E., and J. F. Lynch. 1982. Effects of latitude, season, elevation, and microhabitat on field body temperatures of neotropical and temperate zone salamanders. Ecology 63:1657-1664.

Felzer, B., and P. Heard. 1999. Precipitation differences amongst GCMs used for the U.S. National assessment. Journal of the American Water Resources Association 35:1327-1339.

Fisher-Reid, M. C., and J. J. Wiens. 2011. What are the consequences of combining nuclear and mitochondrial data for phylogenetic analysis? Lessons from 159

Plethodon salamanders and 13 other vertebrate clades. BMC Evolutionary Biology 11:300.

Fitzpatrick, B. M., K. Shook, and R. Izally. 2009. Frequency-dependent selection by wild birds promotes polymorphism in model salamanders. BMC Ecology 9:Article No.: 12.

Fitzpatrick, M. C., N. J. Gotelli, and A. M. Ellison. 2013. MaxEnt versus MaxLike: empirical comparisons with ant species distributions. Ecosphere 4.

Fraser, D. F. 1976. Empirical evaluation of the hypothesis of food competition in salamanders of the genus Plethodon. Ecology 57:459-471.

Gall, S. B., C. D. Anthony, and J. A. Wicknick. 2003. Do behavioral interactions between salamanders and beetles indicate a guild relationship? The American Midland Naturalist 149:363-374.

Gatz, R. N., E. C. Crawford, and J. Piiper. 1975. Kinetics of inert-gas equilibration in an exclusively skin-breathing salamander, Desmognathus fuscus. Respiration Physiology 24:15-29.

Gibbs, J. P. 1998. Distribution of woodland amphibians along a forest fragmentation gradient. Landscape Ecology 13:263-268.

Gleason, H. A. 1926. The Individualistic Concept of the Plant Association. Bulletin of the Torrey Botanical Club 53:7-26.

Goodman, B. A., D. B. Miles, and L. Schwarzkopf. 2008. Life on the rocks: Habitat use drives morphological and performance evolution in lizards. Ecology 89:3462- 3471.

Grafen, A. 1989. The phylogenetic regression. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 326:119-157.

Griffis, M. R., and R. G. Jaeger. 1998. Competition leads to an extinction-prone species of salamander: Interspecific territoriality in a metapopulation. Ecology 79:2494- 2502.

Grinnell, J. 1917. The niche-relationships of the California Thrasher. The Auk 34:427- 433.

Grizzell, R. A. 1949. The hibernation site of three snakes and a salamander. Copeia 1949:231–232. 160

Grover, M. C. 1998. Influence of cover and moisture on abundances of the terrestrial salamanders Plethodon cinereus and Plethodon glutinosus. Journal of Herpetology 32:489-497.

Guisan, A., and N. E. Zimmermann. 2000. Predictive habitat distribution models in ecology. Ecological Modelling 135:147-186.

Gunderson, A. R., and J. H. Stillman. 2015. Plasticity in thermal tolerance has limited potential to buffer ectotherms from global warming. Proceedings of the Royal Society B-Biological Sciences 282:8.

Haeckel, E. 1866, Generelle morphologie der organismen : Allgemeine grundzüge der organischen formen-wissenschaft, mechanisch begründet durch die von Charles Darwin reformirte descendenz-theorie, Berlin, G. Reimer.

Hampe, A., and A. S. Jump. 2011. Climate relicts: Past, present, future, Pages 313-333 in D. J. Futuyma, H. B. Shaffer, and D. Simberloff, eds. Annual Review of Ecology, Evolution, and Systematics, Vol 42. Annual Review of Ecology Evolution and Systematics. Palo Alto, Annual Reviews.

Hansen, T. F. 1997. Stabilizing selection and the comparative analysis of adaptation. Evolution 51:1341-1351.

Harvey, P. H., and M. D. Pagel. 1991, The comparative method in evolutionary biology. Oxford, Oxford University Press.

Hawkins, B., D.-F. JA, C. Jaramillo, and S. Soeller. 2007. Climate, niche conservatism, and the global bird diversity gradient. The American Naturalist 170:S16-S27.

Heatwole, H., and K. Lim. 1961. Relation of substrate moisture to absorption and loss of water by the salamander, Plethodon cinereus. Ecology 42:814-819.

Heikkinen, R. K., M. Luoto, M. B. Araujo, R. Virkkala, W. Thuiller, and M. T. Sykes. 2006. Methods and uncertainties in bioclimatic envelope modelling under climate change. Progress in Physical Geography 30:751-777.

Hendry, A. P. 2016, Eco-evolutionary dynamics. Princeton, New Jersey, Princeton University Press.

Hickerson, C. M., C. D. Anthony, and M. B. Walton. 2012. Interactions among forest- floor guild members in structurally simple microhabitats. The American Midland Naturalist 168:30-42.

Hickerson, C. M., C. D. Anthony, and J. A. Wicknick. 2004. Behavioral interactions between salamanders and centipedes: competition in divergent taxa. Behavioral Ecology 15:679-686. 161

Highton, R. 2004. A new species of woodland salamander of the Plethodon cinereus group from the Blue Ridge Mountains of Virginia. Jefersoniana 14:1-22.

Highton, R., A. P. Hastings, C. Palmer, R. Watts, C. A. Hass, M. Culver, and S. J. Arnold. 2012. Concurrent speciation in the eastern woodland salamanders (Genus Plethodon): DNA sequences of the complete albumin nuclear and partial mitochondrial 12s genes. Molecular phylogenetics and evolution 63:278-290.

Hocking, D. J., and K. J. Babbitt. 2014. Effects of Red-backed Salamanders on ecosystem functions. PLoS One 9:e86854.

Hocking, D. J., G. M. Connette, C. A. Conner, B. R. Scheffers, S. E. Pittman, W. E. Peterman, and R. D. Semlitsch. 2013. Effects of experimental forest management on a terrestrial, woodland salamander in Missouri. Forest Ecology and Management 287:32-39.

Hoff, J. G. 1977. A Massachusetts hibernation site of the Red-backed Salamander, Plethodon cinereus. Herpetological Review 8:33.

Holomuzki, J. R. 1986. Predator avoidance and diel patterns of microhabitat use by larval Tiger Salamanders (Ambystoma tigrinum). Ecology 67:737-748.

Holt, R. D., and M. S. Gaines. 1992. Analysis of adaptation in heterogeneous landscapes: implications for the evolution of fundamental niches. Evolutionary Ecology 6:433-447.

Hooper, D. U., F. S. Chapin, J. J. Ewel, A. Hector, P. Inchausti, S. Lavorel, J. H. Lawton et al. 2005. Effects of biodiversity on ecosystem functioning: A consensus of current knowledge. Ecological Monographs 75:3-35.

Hosmer, D. W., and S. Lemeshow. 1989, Applied logistic regression. New York, New York, USA, Wiley Interscience.

Houk, L. D. 1977. Life history patterns and reproductive biology of neotropical salamanders in D. H. Taylora, and S. I. Guttman, eds. The Reproductive Biology of Amphibians. New York, Plenum Press.

Huey, R. B. 1991. Physiological consequences of habitat selection. The American Naturalist 137:S91-S115.

Huey, R. B., P. E. Hertz, and B. Sinervo. 2003. Behavioral drive versus behavioral inertia in evolution: A null model approach. The American Naturalist 161:357-366.

Hughes, D. A. 1966. Role of responses to light in selection and maintenance of microhabitat by nymphs of two species of mayfly. Animal Behaviour 14:17-33. 162

Hutchinson, G. E. 1957. Concluding remarks. Cold Spring Harbor Symposia on Quantitative Biology 22:415-427.

Hutchison, V. H. 1961. Critical thermal maxima in salamanders. Physiological Zoology 34:92-125.

Ide, F. P. 1935, The effect of temperature on the distribution of the mayfly fauna of a stream: Publications of the ontario fisheries research laboratory. The University of Toronto Press.

Jaeger, R. G. 1970. Potential extinction through competition between two species of terrestrial salamanders. Evolution 24:632-642.

—. 1971a. Competitive exclusion as a factor influencing distributions of two species of terrestrial salamanders. Ecology 52:632-637.

—. 1971b. Moisture as a factor influencing the distribution of two species of terrestrial salamanders. Oecologia 6:191-207.

Jaeger, R. G. 1972. Food as a limited resource in competition between two species of terrestrial salamanders. Ecology 53:535-546.

Jaeger, R. G. 1978. Plant climbing by salamanders: Periodic availability of plant- dwelling prey. Copeia 1978:686-691.

—. 1980. Microhabitats of a terrestrial forest salamander. Copeia 1980:265-268.

Jaeger, R. G., and D. C. Forester. 1993. Social behavior of plethodontid salamanders. Herpetologica 49:163-175.

Jaeger, R. G., D. Kalvarsky, and N. Shimizu. 1982. Territorial behavior of the Red- backed Salamander: expulsion of intruders. Animal Behaviour 30:490-496.

Jaeger, R. G., E. D. Prosen, and D. C. Adams. 2002. Character displacement and aggression in two species of terrestrial salamanders. Copeia 2002:391-401.

Jockusch, E. L. 1997. Geographic variation and phenotypic plasticity of number of trunk vertebrae in slender salamanders, Batrachoseps (Caudata : Plethodontidae). Evolution 51:1966-1982.

Johnson, P. C. D. 2014. Extension of Nakagawa & Schielzeth's R2 GLMM to random slopes models. Methods in Ecology and Evolution 5:944-946.

Kawecki, T. J., and D. Ebert. 2004. Conceptual issues in local adaptation. Ecology Letters 7:1225-1241. 163

Kearney, M., and W. Porter. 2009. Mechanistic niche modelling: combining physiological and spatial data to predict species' ranges. Ecology Letters 12:334- 350.

Kearney, M., and W. P. Porter. 2004. Mapping the fundamental niche: Physiology, climate, and the distribution of a nocturnal lizard. Ecology 85:3119-3131.

Kearney, M. R., A. P. Isaac, and W. P. Porter. 2014. microclim: Global estimates of hourly microclimate based on long-term monthly climate averages. Scientific Data 1:140006.

Keen, W. H. 1984. Influence of moisture on the activity of a plethodontid salamander. Copeia 1984:684-688.

Kiesecker, J. M., and A. R. Blaustein. 1998. Effects of introduced bullfrogs and smallmouth bass on microhabitat use, growth, and survival of native red-legged frogs (Rana aurora). Conservation Biology 12:776-787.

Kniowski, A., and N. Reichenbach. 2009. The ecology of the peaks of otter salamander (Plethodon hubrichti) in sympatry with the eastern red-backed salamander (Plethodon cinereus). Herpetological Conservation and Biology 4:285-294.

Knouft, J. H., J. B. Losos, R. E. Glor, and J. J. Kolbe. 2006. Phylogenetic analysis of the evolution of the niche in lizards of the Anolis sagrei group. Ecology 87:S29-S38.

Kozak, K. H., and J. J. Wiens. 2006. Does niche conservatism promote speciation? A case study in North American salamanders. Evolution 60:2604-2621.

—. 2010. Niche conservatism drives elevational diversity patterns in Appalachian salamanders. The American Naturalist 176:40-54.

—. 2012. Phylogeny, ecology, and the origins of climate–richness relationships. Ecology 93:S167-S181.

Kroschel, W. A., W. B. Sutton, C. J. W. McClure, and T. K. Pauley. 2014. Decline of the Cheat Mountain Salamander over a 32-year period and the potential influence of competition from a sympatric species. Journal of Herpetology 48:415-422.

Kumar, S., L. G. Neven, and W. L. Yee. 2014. Evaluating correlative and mechanistic niche models for assessing the risk of pest establishment. Ecosphere 5:1-23.

Lambert, T. D., J. R. Malcolm, and B. L. Zimmerman. 2006. Amazonian small mammal abundances in relation to habitat structure and resource abundance. Journal of Mammalogy 87:766-776. 164

Leibold, M. A. 1995. The niche concept revisited: Mechanistic models and community context. Ecology 76:1371-1382.

Lemen, C. A., and M. L. Rosenzweig. 1978. Microhabitat selection in two species of Heteromyid rodents. Oecologia 33:127-135.

Levin, S. A. 1992. The problem of pattern and scale in ecology. Ecology 73:1943-1967.

Levin, S. A. 1993. Concepts of scale at the local level, Pages 7-19 in J. R. Ehleringer, and C. B. Field, eds. Scaling physiological process: Leaf to globe. San Diego, California, Academic Press.

Lillywhite, H. B. 2006. Water relations of tetrapod integument. Journal of Experimental Biology 209:202-226.

Loman, J. 1978. Macrohabitat and microhabitat distribution in Rana arvalis and Rana temporaria (Amphibia, Anura, Ranidae) during summer. Journal of Herpetology 12:29-33.

Longland, W. S., and M. V. Price. 1991. Direct observations of owls and heteromyid rodents can predation risk explain microhabitat use? Ecology 72:2261-2273.

Lord, J., M. Westoby, and M. Leishman. 1995. Seed size and phylogeny in 6 temperate floras: constraints, niche conservatism, and adaptation. The American Naturalist 146:349-364.

Losos, J. B. 2008a. Phylogenetic niche conservatism, phylogenetic signal and the relationship between phylogenetic relatedness and ecological similarity among species. Ecology Letters 11:995-1003.

—. 2008b. Rejoinder to Wiens (2008): Phylogenetic niche conservatism, its occurrence and importance. Ecology Letters 11:1005-1007.

Losos, J. B., and D. B. Miles. 1994. Adaptation, constraint, and the comparative method: Phylogenetic issues and methods, Pages 60-98 in P. C. Wainwright, and S. M. Reilly, eds. Ecological morphology: Integrative organismal biology. Chicago, University of Chicago Press.

Lotka, A. J. 1925, Elements of Physical Biology, Williams and Wilkins Company.

MacKenzie, D. I., L. L. Bailey, and J. D. Nichols. 2004. Investigating species co- occurrence patterns when species are detected imperfectly. Journal of Animal Ecology 73:546-555.

Main, K. L. 1987. Predator avoidance in seagrass meadows: Prey behavior, microhabitat selection, and cryptic coloration. Ecology 68:170-180. 165

Marais, E., and S. L. Chown. 2008. Beneficial acclimation and the Bogert effect. Ecology Letters 11:1027-1036.

Markle, T. M. 2015. Ecology and evoltion of geographic range size variation in North America plethodontid salamanders: Perspectives from thermal physiology, University of Minnesota.

Martins, E. P., and T. F. Hansen. 1997. Phylogenies and the comparative method: A general approach to incorporating phylogenetic information into the analysis of interspecific data. The American Naturalist 149:646-667.

Mathis, A. 1990. Territoriality in a terrestrial salamander: the influence of resource quality and body size. Behaviour 112:162-175.

Mathis, A., R. G. Jaeger, W. H. Keen, P. K. Ducey, S. C. Walls, and B. W. Buchanan. 1995. Aggression and territoriality by salamanders and a comparison with the territorial behavior of frogs, Pages 633-676 in H. Heatwole, and B. K. Sullivan, eds. Amphibian biology, Vol. 2: Social behaviour. Chipping Norton, New South Wales, Australia, Surrey Beatty and Sons.

Mazziotta, A., M. Triviño, O.-P. Tikkanen, J. Kouki, H. Strandman, and M. Mönkkönen. 2016. Habitat associations drive species vulnerability to climate change in boreal forests. Climatic Change 135:585-595.

McCabe, G. J., and D. M. Wolock. 2011. Independent effects of temperature and precipitation on modeled runoff in the conterminous United States. Water Resources Research 47.

McCarty, J. P. 2001. Ecological consequences of recent climate change. Conservation Biology 15:320-331.

McCreadie, J. W., and P. H. Adler. 1998. Scale, time, space, and predictability: Species distributions of preimaginal black flies (Diptera : Simuliidae). Oecologia 114:79- 92.

McEntire, K. D. 2016. Arboreal ecology of Plethodontidae: a review. Copeia 104:124- 131.

McKenny, H. C., W. S. Keeton, and T. M. Donovan. 2006. Effects of structural complexity enhancement on Eastern Red-backed Salamander (Plethodon cinereus) populations in northern hardwood forests. Forest Ecology and Management 230:186-196.

McKitrick, M. C. 1993. Phylogenetic constraint in evolutionary-theory: has it any explanatory power? Annual Review of Ecology and Systematics 24:307-330. 166

McShea, W. J., and E. N. Francq. 1984. Microhabitat selection by Peromyscus leucopus. Journal of Mammalogy 65:675-678.

Milanovich, J. R., W. E. Peterman, N. P. Nibbelink, and J. C. Maerz. 2010. Projected loss of a salamander diversity hotspot as a consequence of projected global climate change. PLoS One 5:e12189.

Miles, D. B. 2004. The race goes to the swift: fitness consequences of variation in sprint performance in juvenile lizards. Evolutionary Ecology Research 6:63-75.

Mitchell, J. C., J. A. Wicknick, and C. D. Anthony. 1996. Effects of timber harvesting practices on peaks of otter salamander (Plethodon hubrichti) populations. Amphibian and Reptile Conservation 1:15-19.

Moore, C. M., and L. M. Sievert. 2001. Temperature-mediated characteristics of the dusky salamander (Desmognathus fuscus) of southern Appalachia. Journal of Thermal Biology 26:547-554.

Morin, X., and W. Thuiller. 2009. Comparing niche- and process-based models to reduce prediction uncertainty in species range shifts under climate change. Ecology 90:1301-1313.

Moriniere, J., M. H. Van Dam, O. Hawlitschek, J. Bergsten, M. C. Michat, L. Hendrich, I. Ribera et al. 2016. Phylogenetic niche conservatism explains an inverse latitudinal diversity gradient in freshwater arthropods. Scientific Reports 6.

Morris, D. W. 1987. Ecological scale and habitat use. Ecology 68:362-369.

Munday, P. L. 2001. Fitness consequences of habitat use and competition among coral- dwelling fishes. Oecologia 128:585-593.

Munoz, M. M., M. A. Stimola, A. C. Algar, A. Conover, A. J. Rodriguez, M. A. Landestoy, G. S. Bakken et al. 2014. Evolutionary stasis and lability in thermal physiology in a group of tropical lizards. Proceedings of the Royal Society B- Biological Sciences 281:9.

Murua, R., and L. A. Gonzalez. 1982. Microhabitat selection in two chilean cricetid rodents. Oecologia 52:12-15.

Nakagawa, S., and H. Schielzeth. 2013. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods in Ecology and Evolution 4:133-142.

Niemiller, M. L. 2005. Pseudotriton ruber ruber (Northern Red Salamander). Tree climbing. Herpetological Review 36:52-52. 167

Niewiarowski, P. H., and W. Roosenburg. 1993. Reciprocal transplant reveals sources of variation in growth rates of the lizard Sceloporus undulatus. Ecology 74:1992- 2002.

O'Donnell, K. M., F. R. Thompson, III, and R. D. Semlitsch. 2014. Predicting variation in microhabitat utilization of terrestrial salamanders. Herpetologica 70:259-265.

O'Donnell, K. M., F. R. Thompson, and R. D. Semlitsch. 2016. Prescribed fire alters surface activity and movement behavior of a terrestrial salamander. Journal of Zoology 298:303-309.

O'Neil, R. V., and A. W. King. 1998. Homeage to St. Michael; or, why are there so many books on scale? in D. L. Peterson, and V. T. Parker, eds. Ecological scale. New York, Columbia University Press.

Paradis, E., J. Claude, and K. Strimmer. 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20:289-290.

Parmesan, C. 2006. Ecological and evolutionary responses to recent climate change. Annual Review of Ecology Evolution and Systematics 37:637-669.

Patrick, D. A., M. L. Hunter, Jr., and A. J. K. Calhoun. 2006. Effects of experimental forestry treatments on a Maine amphibian community. Forest Ecology and Management 234:323-332.

Patterson, P. M. 1940. Corticolous bryophyte societies at Mountain Lake, Virginia. The American Midland Naturalist 23:421-441.

Pearson, R. G., and T. P. Dawson. 2003. Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful? Global Ecology and Biogeography 12:361-371.

Peñalver‐Alcázar, M., P. Aragón, M. C. Breedveld, and P. S. Fitze. 2016. Microhabitat selection in the common lizard: implications of biotic interactions, age, sex, local processes, and model transferability among populations. Ecology and Evolution:10.1002/ece1003.2138.

Peterman, W. E., J. L. Locke, and R. D. Semlitsch. 2013. Spatial and temporal patterns of water loss in heterogeneous landscapes: using plaster models as amphibian analogues. Canadian Journal of Zoology 91:135-140.

Peterson, A. T. 2003. Predicting the geography of species' invasions via ecological niche modeling. The Quarterly Review of Biology 78:419-433.

Peterson, A. T., J. Soberón, and V. Sánchez-Cordero. 1999. Conservatism of ecological niches in evolutionary time. Science 285:1265-1267. 168

Petranka, J. W. 1998, Salamanders of the United States and Canada. Washington, DC, Smithsonian Books.

Petranka, J. W., M. P. Brannon, M. E. Hopey, and C. K. Smith. 1994. Effects of timber harvesting on low elevation populations of southern Appalachian salamanders. Forest Ecology and Management 67:135-147.

Petranka, J. W., and C. K. Smith. 2005. A functional analysis of streamside habitat use by southern Appalachian salamanders: Implications for riparian forest management. Forest Ecology and Management 210:443-454.

Phillips, S. J., R. P. Anderson, and R. E. Schapire. 2006. Maximum entropy modeling of species geographic distributions. Ecological Modelling 190:231-259.

Phillips, S. J., and M. Dudik. 2008. Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecography 31:161-175.

Preest, M. R., and F. H. Pough. 1989. Interaction of temperature and hydration on locomotion of toads. Functional Ecology 3:693-700.

Prinzing, A. 2001. The niche of higher plants: Evidence for phylogenetic conservatism. Proceedings of the Royal Society of London. Series B: Biological Sciences 268:2383-2389.

Pyron, R. A., G. C. Costa, M. A. Patten, and F. T. Burbrink. 2015. Phylogenetic niche conservatism and the evolutionary basis of ecological speciation. Biological Reviews 90:1248-1262.

Pyron, R. A., and J. J. Wiens. 2011. A large-scale phylogeny of Amphibia including over 2800 species, and a revised classification of extant frogs, salamanders, and caecilians. Molecular Phylogenetics and Evolution 61:543-583.

R Core Team. 2016. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

Rahbek, C. 2005. The role of spatial scale and the perception of large-scale species- richness patterns. Ecology Letters 8:224-239.

Raxworthy, C. J., R. G. Pearson, N. Rabibisoa, A. M. Rakotondrazafy, J.-B. Ramanamanjato, A. P. Raselimanana, S. Wu et al. 2008. Extinction vulnerability of tropical montane endemism from warming and upslope displacement: a preliminary appraisal for the highest massif in Madagascar. Global Change Biology 14:1703-1720.

Regester, K. J., and S. T. Samoray. 2002. Plethodon glutinosus (Slimy Salamander). Tree climbing. Herpetological Review 33:45-45. 169

Reichenbach, N., and P. Sattler. 2007. Effects of timbering on Plethodon hubrichti over twelve years. Journal of Herpetology 41:622-629.

Revell, L. J. 2012. phytools: an R package for phylogenetic comparative biology (and other things). Methods in Ecology and Evolution 3:217-223.

Revell, L. J., L. J. Harmon, and D. C. Collar. 2008. Phylogenetic signal, evolutionary process, and rate. Systematic Biology 57:591-601.

Ricklefs, R. E. 1989, Ecology. New York, New York, Freeman & Company, W. H.

Riddell, E. A., and M. W. Sears. 2014. High and dry: Responses of evaporative water loss along an elevational gradient for two species of lungless salamanders. Integrative and Comparative Biology 54:E176.

Riddell, E. A., and M. W. Sears. 2015. Geographic variation of resistance to water loss within two species of lungless salamanders: Implications for activity. Ecosphere 6:1-16.

Robson, M. A., and D. B. Miles. 2000. Locomotor performance and dominance in male Tree Lizards, Urosaurus ornatus. Functional Ecology 14:338-344.

Rodda, G. H., C. S. Jarnevich, and R. N. Reed. 2009. What parts of the US mainland are climatically suitable for invasive alien pythons spreading from Everglades National Park? Biological Invasions 11:241-252.

Rome, L. C., E. D. Stevens, and H. B. John-Alder. 1992. The influence of temperature and thermal acclimation of physiological function, Pages 183-205 in M. E. Feder, and W. W. Burggren, eds. Environmental physiology of the amphibians. Chicago, IL, The University of Chicago Press.

Rosenzweig, M. L. 1995, Species Diversity in Space and Time. Cambridge, MA, Cambridge University Press.

Saxton, K. E., and W. J. Rawls. 2006. Soil water characteristic estimates by texture and organic matter for hydrologic solutions. Soil Science Society of America Journal 70:1569.

Scherrer, D., and C. Körner. 2010. Infra-red thermometry of alpine landscapes challenges climatic warming projections. Global Change Biology 16:2602-2613.

Schumm, S. A., and R. W. Lichty. 1965. Time, space, and causality in geomorphology. American Journal of Science 263:110-119.

Seastedt, T. R., and D. A. Crossley. 1981. Microarthropod response following cable logging and clear-cutting in the southern Appalachians. Ecology 62:126-135. 170

Seebacher, F., and R. A. Alford. 2002. Shelter microhabitats determine body temperature and dehydration rates of a terrestrial amphibian (Bufo marinus). Journal of Herpetology 36:69-75.

Semlitsch, R. D., K. M. O’Donnell, and F. R. Thompson. 2014. Abundance, biomass production, nutrient content, and the possible role of terrestrial salamanders in Missouri Ozark forest ecosystems. Canadian Journal of Zoology 92:997-1004.

Shoo, L. P., D. H. Olson, S. K. McMenamin, K. A. Murray, M. Van Sluys, M. A. Donnelly, D. Stratford et al. 2011. Engineering a future for amphibians under climate change. Journal of Applied Ecology 48:487-492.

Shrader, A. M., C. Bell, L. Bertolli, and D. Ward. 2012. Forest or the trees: At what scale do elephants make foraging decisions? Acta Oecologica-International Journal of Ecology 42:3-10.

Simon, C. A., and G. A. Middendorf. 1976. Resource partitioning by an iguanid lizard: Temporal and microhabitat aspects. Ecology 57:1317-1320.

Sinervo, B. 1990. Evolution of thermal physiology and growth rate between populations of the Western Fence Lizard Sceloporus occidentalis. Oecologia 83:228-237.

Slack, N. G., and E. O. G. Wilson. 2010, Evelyn Hutchinson and the Invention of Modern Ecology, Yale University Press.

Southerland, M. T. 1986. Coexistence of three congeneric salamanders: The importance of habitat and body size. Ecology 67:721-728.

Spight, T. M. 1968. The water economy of salamanders: Evaporative water loss. Physiological Zoology 41:195-203.

Spotila, J. R. 1972. Role of temperature and water in the ecology of lungless salamanders. Ecological Monographs 42:95-125.

Spotila, J. R., and E. N. Berman. 1976. Determination of skin resistance and the role of the skin in controlling water loss in amphibians and reptiles. Comparative Biochemistry and Physiology Part A: Physiology 55:407-411.

Stahlschmidt, Z. R., O. Lourdais, S. Lorioux, M. W. Butler, J. R. Davis, K. Salin, Y. Voituron et al. 2013. Morphological and physiological changes during reproduction and their relationships to reproductive performance in a capital breeder. Physiological Biochemistry and Zoology 86:398-409.

Stebbins, R. C. 1954. Natural history of the salamanders of the plethodontid genus Ensatina. University of California Publications in Zoology 54:47-124. 171

Stommel, H. 1963. Varieties of oceanographic experience. Science 139:572-576.

Suggitt, A. J., P. K. Gillingham, J. K. Hill, B. Huntley, W. E. Kunin, D. B. Roy, and C. D. Thomas. 2011. Habitat microclimates drive fine-scale variation in extreme temperatures. Oikos 120:1-8.

Suggitt, A. J., C. Stefanescu, F. Paramo, T. Oliver, B. J. Anderson, J. K. Hill, D. B. Roy et al. 2012. Habitat associations of species show consistent but weak responses to climate. Biology Letters 8:590-593.

Thrasher, B., J. Xiong, W. Wang, F. Melton, A. Michaelis, and R. Nemani. 2013. Downscaled climate projections suitable for resource management. Eos, Transactions American Geophysical Union 94:321-323.

Titon, B., Jr., C. A. Navas, J. Jim, and F. R. Gomes. 2010. Water balance and locomotor performance in three species of neotropical toads that differ in geographical distribution. Comparative Biochemistry and Physiology Part A Molecular & Integrative Physiology 156:129-135.

Tracy, C. R. 1976. Model of dynamic exchanges of water and energy between a terrestrial amphibian and its environment. Ecological Monographs 46:293-326.

Trauth, S. E., M. L. McCallum, B. J. Ball, and V. E. Hoffman. 2000. Caudata: Plethodon caddoensis (Caddo Mountain Salamander) and Plethodon serratus (Southern Redback Salamander). Nocturnal climbing activity. Herpetological Review 31:232-233.

Ulrich, W., M. Piwczyński, F. T. Maestre, and N. J. Gotelli. 2012. Null model tests for niche conservatism, phylogenetic assortment and habitat filtering. Methods in Ecology and Evolution 3:930-939.

Venesky, M. D., and C. D. Anthony. 2007. Antipredator adaptations and predator avoidance by two color morphs of the Eastern Red-backed Salamander, Plethodon cinereus. Herpetologica 63:450-458.

Vernberg, F. J. 1955. Correlation of Physiological and Behavior Indexes of Activity in the Study of Plethodon cinereus (Green) and Plethodon glutinosus (Green). The American Midland Naturalist 54:382-393.

Volterra, V. 1926. Variazioni e fluttuazioni del numero d’individui in specie animali conviventi. Memoria della Reale Accademia Nazionale dei Lincei 2:31-113.

Waldorf, E. S. 1976. Spider size, microhabitat selection, and use of food. The American Midland Naturalist 96:76-87. 172

Walther, G. R., E. Post, P. Convey, A. Menzel, C. Parmesan, T. J. C. Beebee, J. M. Fromentin et al. 2002. Ecological responses to recent climate change. Nature 416:389-395.

Walton, B. M. 2005. Salamanders in forest-floor food webs: Environmental heterogeneity affects the strength of top-down effects. Pedobiologia 49:381-393.

—. 2013. Top-down regulation of litter invertebrates by a terrestrial salamander. Herpetologica 69:127-146.

Walton, B. M., and S. Steckler. 2005. Contrasting effects of salamanders on forest-floor macro- and mesofauna in laboratory microcosms. Pedobiologia 49:51-60.

Walton, B. M., D. Tsatiris, and M. Rivera-Sostre. 2006. Salamanders in forest-floor food webs: Invertebrate species composition influences top-down effects. Pedobiologia 50:313-321.

Webb, C. O., D. D. Ackerly, M. A. McPeek, and M. J. Donoghue. 2002. Phylogenies and community ecology. Annual Review of Ecology and Systematics 33:475-505.

Wells, K. D. 2007, The ecology and behavior of amphibians. Chicago, University of Chicago Press.

Welsh, H. H., and S. Droege. 2001. A case for using plethodontid salamanders for monitoring biodiversity and ecosystem integrity of North American forests. Conservation Biology 15:558-569.

Werner, E. E., C. J. Davis, D. K. Skelly, R. A. Relyea, M. F. Benard, and S. J. McCauley. 2014. Cross-Scale Interactions and the Distribution-Abundance Relationship. PLoS One 9:e97387.

Wiens, J. A. 1989. Spatial scaling in ecology. Functional Ecology 3:385-397.

Wiens, J. A., J. F. Addicott, T. J. Case, and J. Diamond. 1986. Overview the importance of spatial and temporal scale in ecological investigations, Pages 145-153 in J. A. T. J. C. Diamond, ed. Community Ecology.

Wiens, J. J. 2008. Commentary on Losos (2008): Niche conservatism déjà vu. Ecology Letters 11:1004-1005.

Wiens, J. J., D. D. Ackerly, A. P. Allen, B. L. Anacker, L. B. Buckley, H. V. Cornell, E. I. Damschen et al. 2010. Niche conservatism as an emerging principle in ecology and conservation biology. Ecology Letters 13:1310-1324. 173

Wiens, J. J., and C. H. Graham. 2005. Niche conservatism: Integrating evolution, ecology, and conservation biology. Annual Review of Ecology Evolution and Systematics 36:519-539.

Wilbanks, T. J., and R. W. Kates. 1999. Global change in local places: How scale matters. Climatic Change 43:601-628.

Winters, A., and M. E. Gifford. 2013. Geographic variation in the water economy of a lungless salamander. Herpetological Conservation and Biology 8:741-747.

With, K. A., and T. O. Crist. 1995. Critical thresholds in species responses to landscape structure. Ecology 76:2446-2459.

Wooten, J. A., C. D. Camp, J. R. Combs, E. Dulka, A. Reist, and D. M. Walker. 2013. Re-evaluating niche conservatism versus divergence in the Woodland Salamander genus Plethodon: a case study of the parapatric members of the Plethodon glutinosus species complex. Canadian Journal of Zoology 91:883-892.

Wyman, R. L. 1998. Experimental assessment of salamanders as predators of detrital food webs: effects on invertebrates, decomposition and the carbon cycle. Biodiversity and Conservation 7:641-650.

Xu, J., R. E. Grumbine, A. Shrestha, M. Eriksson, X. Yang, Y. Wang, and A. Wilkes. 2009. The Melting Himalayas: Cascading effects of climate change on water, biodiversity, and livelihoods. Conservation Biology 23:520-530.

Yackulic, C. B., R. Chandler, E. F. Zipkin, J. A. Royle, J. D. Nichols, E. H. Campbell Grant, and S. Veran. 2013. Presence-only modelling using MAXENT: when can we trust the inferences? Methods in Ecology and Evolution 4:236-243.

Ziemba, J. L., C.-A. M. Hickerson, and C. D. Anthony. 2016. Invasive asian earthworms negatively impact keystone terrestrial salamanders. PLoS One 11:e0151591.

174

APPENDIX 1: CHAPTER 3 FIELD SITE LIST

A list of all field sites used during this study with the number of plots sampled from each site, latitude, longitude, elevation, and the national park or forest in which the site was located. Site Name Plots Latitude Longitude Elevation Park/Forest LHL3 4 39.41 -82.22 127 Wayne National Forest LHL1 9 39.45 -82.16 234 Wayne National Forest George Washington and LMM3 8 37.52 -79.63 327 Jefferson National Forest LHL2 11 39.62 -82.05 334 Wayne National Forest LLM1 4 35.89 -82.81 380 Pisgah National Forest LLM3 10 35.94 -82.92 391 Cherokee National Forest George Washington and LML2 10 37.56 -79.56 422 Jefferson National Forest George Washington and LML1 2 37.57 -79.46 447 Jefferson National Forest Great Smoky Mountains National LLH3 10 35.67 -83.66 450 Park George Washington and LHL4 10 38.79 -78.52 473 Jefferson National Forest George Washington and LMM4 11 37.93 -79.00 489 Jefferson National Forest LMH2 10 36.10 -82.45 490 Cherokee National Forest LHM2 12 39.12 -79.77 520 Monongahela National Forest LHM5 4 39.12 -79.73 531 Monongahela National Forest George Washington and LMM5 12 37.91 -79.02 547 Jefferson National Forest LMH3 4 36.07 -82.46 633 Cherokee National Forest Great Smoky Mountains National LLH2 10 35.65 -83.58 650 Park Great Smoky Mountains National LLH1 15 35.76 -83.21 670 Park George Washington and MML3 10 37.89 -79.05 707 Jefferson National Forest George Washington and MHL3 10 38.67 -78.59 708 Jefferson National Forest MMH4 12 36.14 -82.35 713 Cherokee National Forest Blue Ridge Parkway National MML1 15 37.45 -79.61 722 Park George Washington and MMM4 11 37.90 -79.04 731 Jefferson National Forest George Washington and MHL4 10 38.61 -78.63 732 Jefferson National Forest MMH1 2 36.59 -81.82 749 Cherokee National Forest George Washington and MHL2 10 38.74 -78.57 756 Jefferson National Forest Great Smoky Mountains National MLH4 10 35.55 -83.31 767 Park 175

Appendix 1: continued Site Name Plots Latitude Longitude Elevation Park/Forest MMH2 16 36.55 -81.74 823 Cherokee National Forest MHM1 8 38.61 -79.80 835 Monongahela National Forest Great Smoky Mountains MLH3 10 35.59 -83.36 838 National Park George Washington and MMM2 2 37.49 -79.55 860 Jefferson National Forest MLM1 4 35.81 -82.21 887 Pisgah National Forest Blue Ridge Parkway National MML2 10 37.87 -79.15 905 Park Blue Ridge Parkway National MML4 1 36.82 -80.35 908 Park MHM4 13 38.77 -79.57 931 Monongahela National Forest MMH3 18 36.17 -82.01 938 Cherokee National Forest Great Smoky Mountains MLH2 17 35.63 -83.21 939 National Park MHM2 10 38.58 -79.78 953 Monongahela National Forest George Washington and MMM1 19 37.51 -79.52 955 Jefferson National Forest MMM6 14 36.03 -82.42 979 Pisgah National Forest George Washington and MMM5 14 37.92 -79.06 1019 Jefferson National Forest MHM3 10 39.11 -79.60 1074 Monongahela National Forest MLM2 11 35.78 -82.96 1135 Pisgah National Forest Great Smoky Mountains HLH3 5 35.63 -83.17 1191 National Park George Washington and HMM4 12 37.37 -80.53 1199 Jefferson National Forest George Washington and HMM2 10 37.25 -80.86 1211 Jefferson National Forest George Washington and HMM3 10 37.35 -80.54 1219 Jefferson National Forest HMH2 22 36.50 -81.89 1237 Cherokee National Forest HHM1 10 38.62 -79.93 1244 Monongahela National Forest HHM2 15 38.59 -79.76 1251 Monongahela National Forest HMH4 12 36.12 -82.05 1310 Pisgah National Forest Great Smoky Mountains HLH1 19 35.58 -83.40 1368 National Park Great Smoky Mountains HLH2 12 35.59 -83.07 1376 National Park HHM4 16 38.70 -79.53 1446 Monongahela National Forest Great Smoky Mountains HLH5 19 35.61 -83.45 1465 National Park HLM2 10 35.71 -82.27 1600 Pisgah National Forest HLM3 12 35.75 -82.33 1618 Pisgah National Forest 176

Appendix 1: continued Site Name Plots Latitude Longitude Elevation Park/Forest HLM1 10 35.73 -82.28 1633 Pisgah National Forest Great Smoky Mountains HLH4 12 35.59 -83.47 1719 National Park

177

APPENDIX 2: CHAPTER 3 SPECIES LIST

Summary table for species included in this study. I present counts for the number of salamanders (N) and number of field sites (Sites) each species was found. All other variable I provide the mean value for adults of each species. N Sites SVL Mass Lat Long Elevation Soil Ground Air Relative Species Temperature Temperature Temperature Humidity Plethodon cinereus 112 31 3.31 0.75 38.01 -80.10 809.85 13.04 14.90 20.74 69.47 Plethodon hoffmani 14 7 4.15 1.24 38.19 -78.87 665.71 12.77 14.49 19.72 53.03 Plethodon hubrichti 39 3 4.47 1.60 37.50 -79.53 947.33 11.25 13.59 17.18 52.13 Plethodon nettingi 9 1 4.16 1.21 38.70 -79.53 1445.33 8.79 10.87 16.97 69.42 Plethodon richmondi 8 3 4.03 1.14 36.43 -81.88 987.63 10.15 12.14 20.26 42.20 Plethodon serratus 25 9 3.14 0.71 35.67 -83.33 765.72 11.06 13.34 20.08 70.48 Plethodon sherando 40 3 4.07 1.17 37.92 -79.06 1041.45 9.97 11.80 15.92 66.79 Plethodon cylindraceus 19 13 5.88 5.63 37.31 -80.32 777.00 14.61 17.17 22.69 78.31 Plethodon glutinosus 12 10 5.24 3.94 38.17 -80.31 809.25 13.88 15.24 22.71 75.77 Plethodon jordani 25 4 4.70 2.43 35.59 -83.42 1395.40 14.57 15.99 19.54 93.88 Plethodon metcalfi 6 2 4.81 2.23 35.60 -83.11 1318.33 14.96 15.88 19.97 89.85 Plethodon montanus 19 7 4.18 1.89 36.18 -82.00 1176.95 15.09 16.36 20.62 91.65 Plethodon teyahalee 3 2 5.37 3.92 35.69 -82.88 941.33 17.40 19.37 22.13 96.73 Plethodon yonahlossee 2 2 4.30 2.18 36.34 -81.95 1114.50 15.80 17.60 20.25 89.35 Plethodon wehrlei 10 3 5.29 3.13 38.59 -79.78 1217.20 11.18 12.93 18.05 67.17 Plethodon welleri 6 1 3.00 0.58 36.14 -82.35 675.33 9.06 10.90 23.28 46.33 Plethodon dorsalis 12 1 3.41 0.94 35.94 -82.92 389.50 12.03 15.43 25.20 37.66 Desmognathus 64 27 3.37 1.24 36.66 -81.80 1169.59 14.45 15.83 20.68 83.56 Desmognathus wrighti 17 7 2.19 0.28 35.64 -83.02 1379.88 14.55 15.35 19.96 88.85 Eurycea 22 15 3.08 0.64 36.78 -81.37 882.23 15.20 17.59 21.53 83.67 Pseudotriton ruber 6 3 6.21 7.68 36.97 -80.66 559.17 16.77 18.66 22.83 83.70 Gyrinophilus porphyriticus 3 3 7.92 9.92 36.36 -81.99 1275.33 14.23 15.87 20.63 92.50 178

Appendix 2: continued % Coarse % Fine Soil Leaf % Green % Bare Species Woody Woody % Rock % Bark Canopy Aspect Slope Moist. Litter Vegetation Ground Debris Debris Tension Depth Plethodon cinereus 13.59 9.21 3.36 2.83 0.68 1.46 94.25 200.54 8.83 630.85 3.05 Plethodon hoffmani 5.14 3.86 2.14 4.86 0.00 0.00 95.84 159.00 9.89 475.52 3.32 Plethodon hubrichti 10.00 7.40 4.07 9.53 7.13 0.00 97.61 269.20 18.10 411.40 2.83 Plethodon nettingi 50.00 8.78 3.67 11.33 12.11 0.33 94.80 70.78 3.75 1148.37 0.62 Plethodon richmondi 9.38 8.00 4.75 4.25 0.50 1.00 94.74 214.00 9.38 928.75 2.91 Plethodon serratus 7.36 6.64 4.28 1.28 0.56 2.36 93.95 197.88 12.08 251.13 3.08 Plethodon sherando 8.30 8.60 1.80 4.30 0.00 0.90 90.67 150.30 11.69 361.83 2.98 Plethodon cylindraceus 11.89 8.68 3.21 8.21 0.32 0.53 94.13 216.32 12.26 683.72 3.05 Plethodon glutinosus 17.50 11.00 5.25 0.33 0.00 2.00 95.39 183.67 10.35 456.53 3.42 Plethodon jordani 25.28 13.68 5.20 0.64 1.88 0.52 94.11 225.40 17.50 200.86 3.00 Plethodon metcalfi 7.83 8.33 4.83 0.00 0.00 2.50 95.84 185.33 17.04 400.39 3.63 Plethodon montanus 15.37 12.32 2.74 0.37 1.16 1.74 96.13 201.26 11.14 501.36 2.41 Plethodon teyahalee 11.33 8.00 1.33 0.00 0.00 1.67 95.23 242.67 9.50 30.53 2.63 Plethodon yonahlossee 17.00 12.00 2.50 0.00 2.50 15.00 93.89 308.00 14.00 920.28 2.08 Plethodon wehrlei 11.10 4.10 2.90 8.30 0.00 7.50 97.52 295.00 11.23 433.86 3.53 Plethodon welleri 12.83 9.33 3.67 0.00 1.00 0.50 98.27 211.00 4.58 973.01 4.96 Plethodon dorsalis 12.92 13.58 2.00 0.00 0.00 0.00 64.84 156.17 1.10 11.55 2.11 Desmognathus sp. 19.00 9.69 3.28 4.80 2.72 4.09 95.95 219.55 12.11 338.52 2.56 Desmognathus wrighti 29.76 8.24 4.29 0.88 0.88 0.53 96.36 237.00 15.03 292.81 2.60 Eurycea sp. 17.14 7.00 3.41 3.77 1.23 2.14 95.15 202.77 11.44 295.21 2.99 Pseudotriton ruber 29.83 4.33 2.33 1.00 0.00 0.67 98.66 233.00 17.04 813.15 2.43 Gyrinophilus porphyriticus 31.67 8.33 6.00 14.33 0.00 0.00 96.19 193.33 13.00 16.86 3.13

! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !

Thesis and Dissertation Services ! !