THE INFLUENCE OF URBANIZATION ON WATER DEMAND AND LIPID AND PROTEIN CONSUMPTION IN MESIC ENVIRONMENTS

Jamie Becker

A Thesis

Submitted to the Graduate College of Bowling Green State University in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE

December 2017

Committee:

Kevin McCluney, Advisor

Shannon Pelini

Karen Root ii

ABSTRACT

Kevin McCluney, Advisor

Water is vital for terrestrial life, but water sources are often scarce, and environmental conditions are often desiccating. For example, the urban heat island effect causes areas with high impervious surfaces to have much hotter and drier conditions than rural locations. In chapter one,

I examined the frequency of water demand behavior across variable landscapes in mesic NW

OH, along with the effects of environmental factors and taxonomic identity. Overall, water demand occurred 11% of the time, but some areas experienced it 53.8% of the time. Ants accounted for much of the response (38% overall, but 62.8% and 57.4% in both high-impervious surface sites). Water demand behavior increased when soil moisture declined, especially below

30%. These results suggest that invertebrates experience water demand in a cool mesic region, even those living outside of urban areas. Further, invertebrates inhabiting cooler sites were most susceptible to periodic droughts or sudden weather changes, which could be due to at those locations having fewer xeric-adapted traits.

When water is scarce, arthropods can obtain water by metabolizing dry food. Because lipids provide twice as much metabolic water as other macromolecules, eating high-lipid foods can reduce the need to forage for moist food. In chapter two, I examined the effects that arthropod water demand and impervious surface have on macronutrient consumption within 32 sites across Toledo, OH. Two artificial diets high in lipid or protein were offered at six trees at each site, three of which had wet water pillows and three had dry pillows. Lipid demand (L: P)

was positively associated with impervious surface (χ2 = 8.36, df = 1, p < 0.01), and water iii pillows reduced the magnitude of this effect; this suggests that water balance likely played a role in driving lipid demand, matching predictions. Ants accounted for much of the response in high impervious surface areas. Because ants play key roles in food webs and ecosystems, increased demand for lipids with urbanization or climate change could have major consequences for urban food webs and ecosystem services. Future work should investigate how shifts in L: P preferences alter food webs. iv

ACKOWLEDGEMENTS

My committee, Kevin McCluney, Shannon Pelini, and Karen Root, have been invaluable in their insights into my work and supporting me to help cultivate my skills in research. I thank

Kevin for his kind and honest criticism that has helped me to become a more rigorous, disciplined writer and researcher. His collaborative nature and emphasis on the importance of academic family has made me a more well-rounded individual. Shannon’s insights into alternative analytical methods have invoked clarity, and she has helped me think about the

importance of brown food webs and the effects that climate change has on my results. Karen’s

influence has improved the way I think about how landscape features influence patterns and

processes at multiple scales. Her background and advice has inspired me to pursue a career in

conservation and applied research. I also thank Mary Gardiner and Arianne Cease for their

academic advice and support.

I collectively thank all members of the McCluney lab for offering help with projects,

giving advice, and offering academic and mental support. Lab meetings and get-togethers have

given me a sense of camaraderie – something that is lacking in many labs, and something for

which I am very grateful. I especially thank Ashley Everett, Haley Ingram, Nadejda

Mirochnitchenko, Lily Murnen, Melanie Queener, and Matt Zach who have all aided me in my

research, from running some of my experiments to engaging in extremely tedious tasks.

I thank those who have given permission to conduct my field work on their property:

Toledo Parks and Recreation, Wood County Parks and Recreation, City of Northwood, City of

Perrysburg, City of Rossford, City of Walbridge, Springfield Township, Calvary Assembly of

God, Calvary Cemetery, Cedar Creek Church, Grace United Methodist Church of Perrysburg, v

Grove Patterson Academy, Rosary Cathedral Parish, St. Frances de Sales School, Sunrise

Banquet Center, Walbridge Apartments, and Winterfield Venture Charter Academy.

Finally, I wish to thank my family for their encouragement and support. I thank my sister,

Lauren, and her husband, Matt, for sharing their experiences of graduate school and offering career advice. Finally, I thank J.D. for being an awesome husband.

vi

TABLE OF CONTENTS

Page

CHAPTER I. FREQUENCY OF ARTHROPOD WATER DEMAND IN MESIC

ENVIRONMENTS ...... 1

Introduction ...... 1

Materials and methods ...... 5

Results ...... 8

Discussion ...... 9

CHAPTER II. CLIMATE-INDUCED CHANGES IN WATER DEMAND DRIVE

HIGHER LIPID AND LOWER PROTEIN CONSUMPTION AMONG URBAN

ARTHROPODS ...... 12

Introduction ...... 12

Materials and methods ...... 15

Lab experiments ...... 15

Field experiments ...... 15

Results ...... 19

Lab experiments ...... 19

Field experiments ...... 19

Discussion ...... 20

REFERENCES ...... 23

APPENDIX A. TABLES ...... 33

APPENDIX B. FIGURES ...... 50 1

CHAPTER I. FREQUENCY OF ARTHROPOD WATER DEMAND IN MESIC

ENVIRONMENTS

Introduction

Water is vital for terrestrial life because it is a universal solvent, aids in material

transport, maintains cell membrane fluidity, maintains turgor pressure, and regulates biochemical

reactions (Hadley 1994). Thus, severe dehydration disrupts an individual’s ability to function

normally. In some organisms, dehydration could decrease muscle performance (Claussen et al.

2000), growth rate (Jindra and Sehnal 1990, McCluney and Date 2008), survival (Dinh et al.

1988, Finkler 1999), and reproduction (Coe and Rotenberry 2003, Tieleman et al. 2004). A lack of adequate water availability can ultimately affect species distributions (Buckley and Jetz 2007), species richness (Hawkins et al. 2005, Keil et al. 2008), community composition (Schowalter et al. 1999, McCluney and Sabo 2012), and trophic interactions (Lensing and Wise 2006, Spiller and Shoener 2008, Sabo et al. 2008, Allen et al. 2014, McCluney et al. 2012, McCluney and

Sabo 2009, 2016, Deguines et al. 2016).

Multiple factors can influence water availability and water balance (difference between

water gain and loss), including distribution of water in space and time and organismal traits

(Chown et al. 2011, McCluney 2017). Water availability can depend upon regional and seasonal

differences in precipitation, distribution of above-ground water bodies, and groundwater stores.

Differences in land use can also influence water availability (Groffman et al. 2014, Steele and

Heffernan 2014). Overuse of machinery and improper soil management techniques compact soil in both agricultural (Hamza and Anderson 2005) and urban (Gregory et al. 2006) landscapes.

Urban areas also have a high density of impervious surfaces that limit groundwater infiltration

(Forman 2014). Since soil compaction and impervious surfaces prevent infiltration, water is 2 redistributed to other locations. While some land use types are dry, certain land use types also

provide water, like irrigated croplands, urban greenspaces such as parks, yards, gardens, and

commercial landscaping. Other water sources along an urban-rural gradient include artificial

ponds, swimming pools, and water fountains. Since several factors influence water availability,

maintaining water balance can be a challenge.

Water is gained in organisms through ingestion, water vapor absorption, and release of

water through metabolism (Hadley 1994). When free water is limited, might spend more

time searching for moist areas or moist food (Lewis and Bernays 1985, McCluney and Sabo

2009). If the animal cannot obtain water locally, it may be forced to move to more distant

locations, but this is not without costs (Loarie et al. 2009, Dias et al. 2013). Water ingestion is

the primary way in which animals gain water, but some rely on other methods. Water vapor is

passively absorbed through the cuticle, spiracles, and alimentary canal when relative humidity is

high, but arthropods can also expend energy to actively absorb water vapor (Hadley 1994).

Finally, animals gain water by metabolizing certain macromolecules. Lipid and carbohydrate

metabolism produces water, while proteins require water to produce nitrogenous waste (Hadley

1994).

Though multiple mechanisms influence an organism’s ability to obtain water, there are

also traits that influence water loss. Cuticular evaporation, respiration, and excretion are ways in

which an organism loses water. Because small organisms have high water loss rates due to their

high surface area: volume ratio, cuticular pathways account for the largest portion of water loss

in most arthropods (Hadley 1994). To reduce water loss, arthropods thus utilize a protective layer

of lipids on the surface of their epicuticles. Arthropods adapted to arid environments tend to have

thick, high quality cuticular lipids made primarily of hydrocarbons, whereas mesic-adapted 3 arthropods commonly have more permeable cuticles (Hadley 1994, Schilman et al. 2005, Benoit and Denlinger 2010). The relative importance of respiratory water loss (RWL) compared to

cuticular water loss is debated (Hadley 1994, Chown 2002, Benoit and Denlinger 2010). Some

have found that RWL only makes up a small portion of total water loss (TWL) in many

arthropods and can be ignored (Hadley 1994, Chown 2002, Lighton et al. 2004). However, RWL

can account for a larger portion of TWL in arthropods with less cuticular permeability (Lighton

et al. 1993, Schilman et al. 2005, 2008). can regulate water loss by strategically opening

and closing their spiracles during respiration (Hadley 1994, Lighton et al. 1993), but some argue

that many insects do not rely on this behavior to conserve water (Chown 2002, Schilman et al.

2008) except during flight (Lehmann 2001).

When losses exceed gains, organisms redistribute energy stores to produce antidiuretic

hormones (O’donnell and Spring 2000, Coast et al. 2002), electrolytes, and sugars (Bayley 1999)

to conserve water. Additionally, some arthropods respond behaviorally to increased water loss

rates by seeking refugia with more favorable microclimates (Bell et al. 1986). However,

adaptations to reduce water loss are not always adequate. Not only do higher ambient

temperatures increase arthropod metabolic rate, it also raises vapor pressure differences between

the body and the atmosphere, increasing cuticular transpiration (Hadley 1994, Roberts et al.

1994). When temperatures are sufficient to alter the stability of cuticular hydrocarbons, water

losses can increase substantially (Hadley 1994). After considerable water loss, an organism with

little desiccation tolerance may suffer severe health consequences.

Studies of ecological and evolutionary responses to desiccation have primarily focused

on organisms inhabiting arid ecosystems. However, less is known about organisms that are

frequently exposed to desiccating conditions in other regions. Mesic regions can have locations 4 that are more arid or these regions may experience droughts. For example, water availability can

be low in areas with sandy soils, despite moderate ambient temperatures and lush vegetation.

Urban areas may also increase desiccating conditions. Impervious surfaces not only reduce local water availability via surface runoff, but dark surfaces tend to absorb and radiate heat, while a lack of canopy cover and decreased vegetation prevent the cooling effects of evapotranspiration

(Taha 1997, Zhao et al. 2014). Thus, areas near the city core in mesic regions tend to be hotter

and drier than surrounding landscapes – a trend known as the urban heat island effect (Brazel et

al. 2000, Imhoff et al. 2010, Groffman et al. 2014, Zhao et al. 2014). Arthropods in mesic

regions could therefore experience frequent desiccation as do desert arthropods.

In this study, I ask how water demand behavior might be influenced by a variety of

landscapes within a mesic region. I specifically test the following hypotheses: 1) water demand

behavior is more frequent in high impervious surface areas because these areas are hotter and

have limited moist soils, promoting desiccation; 2) certain taxa could exhibit water demand

behavior more than others, due to differences in physiological, behavioral, or ecological traits; 3)

water demand behavior is related to measured temperature and soil moisture, because these

variables can alter water loss rates and water availability.

5

Materials and methods

I selected three regions within Northwest Ohio that differed in population size as of U.S.

Census 2010: Toledo (population: 287,208), Bowling Green (BGSU; population: 30,028), and

Kitty Todd Nature Preserve within the Oak Openings area (population within the preserve: ~0).

In Toledo and BGSU, I contrasted street trees [Toledo (41°39'17.3"N, 83°32'04.9"W), BGSU

(41°22'54.0"N, 83°38'28.8"W)] to trees in greenspaces [Toledo (41°39'23.4"N, 83°32'09.0"W),

Bowling Green (41°22'49.9"N, 83°38'29.3"W)]. Within Oak Openings, I selected trees at a site which had sandy soil (41°37'45.91"N, 83°47'5.45"W), and at a site which had clay soil

(41°37'44.09"N, 83°48'45.89"W).

I then placed paired wet and dry water pillows (small pouches filled with a polymer that absorbs water; Cricket water pillows, Zilla, Franklin, WI) on the ground and in the branches of ten trees at each location. Wet water pillows were fully hydrated with deionized water for up to

30 min, placed near (but not touching) each other, with the water accessible side up, attached to a binder clip to prevent pillows from blowing away (or to attach to a branch). From 5 June to 8

August 2014, I visited each region once every three days (13 visits), with each visit including examination of each water pillow during the late afternoon and at night. I did not visit or place pillows during storm conditions. At each visitation, I photographed arthropods on water pillows for later identification, noting those that had escaped prior to taking a photograph. Using a hand- held weather station (WS-HT350, Ambient Weather, Chandler, AZ), I measured shaded temperature and relative humidity measurements once per site at each visit. I then measured soil moisture three times per tree (SM 150 soil moisture sensor, Dynamax, Houston, TX ) at the highest, lowest, and medium points of uneven soil, and used the maximum soil moisture per tree 6 in further analyses. I calculated percent impervious surface per tree at 25m buffers using

National Land Cover Database 2011 (NLCD 2011).

Because some observations included several individuals of the same taxonomic group

(namely, ants), while other observations included only a single individual, using abundance data tended to inflate invertebrate visitation at each tree and rendered our statistical comparisons more difficult. Thus, I calculated water demand in two ways. First, I calculated a frequency of water demand per tree using the number of observations with at least one arthropod present, out of 13

visits per tree (hereafter “observation frequency per tree”). This metric allowed examination of

effects of impervious surface and average site-level climatic conditions. Second, I calculated a

frequency of water demand per day using the number of trees with at least one arthropod present,

out of 10 trees per site (hereafter “observation frequency per day”). This metric allowed

examination of effects of daily variation in weather patterns (“daily soil moisture” or “daily

temperature”). For both metrics, I calculated observation frequency per tree and per day by any

invertebrate and separately by taxonomic group.

To examine effects of differences between sites (impervious surface, average climate), I

used likelihood ratio tests of linear mixed effects models with observation frequency per tree as

the response and water pillow type (wet, dry) and percent impervious surface as fixed factors,

and region (Toledo, BGSU, Oak Openings), site, and tree number as nested random factors.

Fixed factors were dropped one at a time from a full model using Bolker et al. (2009) source

code which tested significant changes in likelihood (i.e., if the model becomes significant once a

fixed factor is dropped, that variable significantly influences the response variable).

To examine differences with daily weather events, I used a linear mixed effects model,

but one that examined interactive or additive effects of pillow type, daily maximum soil 7 moisture, and daily maximum temperature per site as fixed effects, and region and site as nested

random effects. But, I also included average climatic conditions for each site (soil moisture and

temperature) as interactive or additive fixed effects in this model, to assess whether arthropods at

sites that were generally hotter/cooler or drier/wetter responded more strongly on days when

weather conditions became hotter/drier. For these daily statistical models, I compared alternative

temporal autocorrelation structures – those that assumed compound symmetry or autoregressive

variance-covariance – and picked the best model, using AIC, for subsequent analysis.

Assumptions of normality and equality of variance were checked using normal probability plots

on residuals and graphs of residuals vs. fitted estimates, respectively. All analyses were

conducted in R v. 3.3.2, with nlme and lme4 packages.

8

Results

Across all locations, the average observation frequency per tree was 23.3% on wet pillows vs 12.3% at dry pillows (Fig 1a), a difference of 11%, indicating that increased water demand occurs ~11% of the time across this mesic region, but with values varying from 0% or 53.8% among sites and trees. The observation frequency of individuals on wet and dry pillows per tree significantly declined with percent impervious surface (χ2 = 8.69, df = 1, p < 0.01; Fig 1b, Table

2), which could be partially explained by a decline in abundance with increasing impervious

surface (df = 1, p < 0.01, r2 = 0.318; Fig 1c).

Observation frequency per day was influenced by a 4-way interaction between daily soil

moisture, daily temperature, mean site soil moisture, and mean site temperature (χ2 = 30.51, df =

1, p < 0.01; Table 3). In general, there were fewer observations of arthropods on wet pillows on

days with higher soil moisture (Fig 2), especially above ~30%. The effect of daily changes in soil

moisture and temperature on water demand was strongest at sites that were generally cooler (Fig

3).

Across all regions, ants accounted for 38% of the total response on wet and dry pillows.

In Bowling Green and Toledo, ants accounted for 62.76% and 57.4% of the total response,

respectively (Fig 4). The overall response at Oak Openings was more diverse (Table 4) and

primarily dominated by Opiliones, , Hymenoptera, , and Collembola. Of the

15 Orders observed in this study, Oak Openings contained 93% of the taxa (lacking Dermaptera),

with 50% at BGSU and 43% in Toledo.

9

Discussion

Though animal water balance has been well studied in xeric regions, little research investigates the possibility of water demand in mesic regions. Water is not uniform in mesic regions and can vary across landscapes. For instance, soil type might influence accessible soil moisture. And the urban heat island effect can cause areas with high impervious surface to

experience hotter, drier conditions than rural areas. Thus, invertebrates living within cities in

mesic regions could be more likely to be water-limited. However, I did not find support for this hypothesis. Instead, I found that (1) water demand is just as strong or stronger in undeveloped areas in in mesic regions – here NW OH; (2) variation exists between taxa in response to wet

pillows, with ants being important in urban systems; and (3) water demand behavior is linked to

combination of longer-term climatic patterns and shorter-term daily weather (both temperature

and soil moisture).

The initial hypothesis was that water demand behavior can occur in mesic regions and

that it should increase with impervious surface because such areas are typically hotter and drier

than rural areas. Water demand (i.e., the frequency of individuals on wet pillows per day or per

tree) can occur in mesic regions—11% of the time across a 2-month period (5 June to 8 August

2014) and that it can occur as much as 53.8% of the time in some locations in this region. Yet,

contrary to expectations water demand declined with increasing impervious surface. This can

partly be explained by a decline in abundance, but not entirely. Contrary to expectations, there

was strong water demand behavior among arthropods in the undeveloped Oak Openings regions,

with 0% impervious surface. The results of the analysis of environmental factors provides some

insight. Water demand behavior was higher when soil moisture declined at sites that were

generally cooler, like Oak Openings. This result might be expected if arthropods that reside in 10 cooler, more mesic environments, do not have traits allowing resistance to desiccation and thus

must seek water when conditions become drier (hypothesized in McCluney 2017). Urban

arthropods, on the other hand, may have more desiccation resistance and thus may be less prone

to seek water under desiccating conditions.

Most of the water demand behavior in urban environments was driven by ants. These

taxa may have intermediate desiccation tolerance. For example, Camponotus and Brachymyrmex build their nests under loose debris, with access to soil moisture. But, their ability to reside above ground in harsher environments suggests that they are somewhat desiccation tolerant (Hood and

Tschinkel 1990). Thus, a slight change in soil moisture could cause water demand behavior in

this group.

This research shows periods of increased water demand and others have demonstrated

that this increased water demand can alter direct and indirect species interactions (McCluney and

Sabo 2016). Additionally, I documented a strong response among ants, and ants have been

shown to play important roles in food webs and ecosystems in and outside of cities (Uno et al.

2010, Youngsteadt et al. 2014, Penick et al. 2015). Therefore, I suggest that animal water

balance could alter food webs in and outside of cities in mesic regions.

In summary, this study gives evidence that invertebrates inhabiting a cool, mesic region

(NW OH) can be frequently water-limited at certain times and locations, even those living in

undeveloped areas. Moreover, water demand behavior increases when soil moisture declines

(especially below 30%) and, counterintuitively, does so more strongly at cooler sites and those

with less impervious surface. This suggests that the strongest effects of droughts on food webs

should be expected in cooler, more mesic regions, even if desiccating conditions are experienced 11 less frequently than in hotter, drier regions. Future work is needed to more closely link organismal traits and environmental conditions with water demand behavior. 12

CHAPTER II. CLIMATE-INDUCED CHANGES IN ANIMAL WATER DEMAND

DRIVE HIGHER LIPID AND LOWER PROTEIN CONSUMPTION AMONG URBAN

ARTHROPODS

Introduction

Having a reliable water source is essential for all terrestrial organisms. Yet, obtaining this resource can be a challenge at times, and many organisms employ physiological and behavioral mechanisms to prevent dehydration. One behavioral mechanism is to obtain water from food, either from dietary water input or potentially from metabolic water production. For example,

field crickets (Gryllus alogus) along the San Pedro River, AZ consumed 31 times more moist

food when the river was dry than when the river was flowing two weeks later (McCluney and

Sabo 2009).

In addition to moist food, water can also be gained by consuming and catabolizing dry

food. Per gram, lipids produce twice as much water when they are metabolized compared to

carbohydrates and proteins, while a large portion of metabolized protein results in nitrogenous

waste, which requires additional water for excretion (Hadley 1994). It is argued that since fats

contain more energy, and less energy is required to gain the caloric equivalent as carbohydrates,

less metabolic water is produced per unit weight (Frank 1988, Hadley 1994). However, some

insects living in dry environments, or forced to consume dry food, sacrifice growth by increasing

their metabolism to gain water (Jindra and Sehnal 1990, Hadley 1994). And some flying insects

(e.g. bees) rapidly produce metabolic water from lipids (Nicolson 2009). Thus, metabolic water

from lipids can be a significant water source in some situations or for some taxa and waste

excretion associated with proteins can be a significant loss of water. 13

Studies of nutritional physiology and ecology, nutritional demands can differ depending

on aspects of physiology Simpson and Raubenheimer 2012). But despite the large body of

literature that investigates nutrient regulation and substantial research on osmoregulation, the two

are rarely linked—could water demand interact with nutrient intake targets (i.e., the quantity of

and ratio of protein, carbohydrate, and lipid)? Recently, Frizzi et al. (2016) discovered that

Mediterranean ants (Crematogaster scutellaris) will alter their relative intake of nutrients and

water depending on environmental conditions, and plague locusts (Chortoicetes terminifera) will

regulate their intake of protein, carbohydrate, and water (Clissold et al. 2014). Another study

found that Drosophila regulated their protein and carbohydrate consumption and consumed more water when given yeast than when given sucrose (Fanson et al. 2012). These observations suggest that a variety of arthropods regulate their intake of water along with nutrients. However, relatively little is understood about how changes in animal water balance (gains vs losses) might

influence nutrient demand, especially under field conditions.

Here I tested the hypothesis that animals will seek to maintain water balance under

desiccating conditions by reducing consumption of protein (because nitrogenous waste increases

water loss via excretion) and increasing consumption of lipids (because lipids produce metabolic

water). Specifically, I designed lab and field experiments to alter water balance and measure the

relative amount of protein and lipid consumed among diets of contrasting ratios. In the lab, I

conducted consumption trials with crickets, in environmental chambers with altered water

availability. In the field, I altered water availability across an urbanization gradient and measured

rates of consumption by the entire community of arthropods. I suspected that urbanization should

influence dietary targets because the urban heat island effect causes high % impervious surface areas to have hot, dry microclimates, promoting desiccation (Taha 1997, Brazel et al. 2000, 14

Imhoff et al. 2010, Groffman et al. 2014, Zhao et al. 2014). However, alternatively, urbanization might promote lipid consumption due to direct effects of increased temperatures on metabolic rates of ectotherms, by increasing energy demand. Since lipids are high energy content, they might be preferred solely due to temperature. However, I isolated the influence of desiccation from the direct effect of temperature on metabolic rate by adding water sources along gradients of urbanization and temperature. Thus, if the addition of water reduces effects of urbanization or temperature on lipid consumption, it suggests that animal water balance is driving observed patterns. 15

Materials and methods

Laboratory experiments

For this study, I used agar-based artificial diets (Table 5) that were either high-lipid (1P:

1C: 3L) or high-protein (3P: 1C: 1L). The protein components were composed of three different

foods, because each food item did not offer an even and complete suite of amino acids. The

diet’s final amino acid profile was validated by Lebensmittel Consulting Co, Fostoria, OH. Diets

were deposited into bottlecaps and fully dried at 50°C in a drying oven (100L Gravity Oven,

model 51030520, Fisher Scientific, Hampton, NH). Thirty cages were placed into an

environmental chamber (HPP750, Memmert, Deutschland, Germany) set to 30°C and 30 %RH

with a 14L: 10D photoperiod. Twenty-five cages each contained five, fourth-instar house

crickets (Acheta domesticus), and five cages without crickets were used to quantify moisture

absorption of diets. Cages were placed into the chamber in random locations in case cages near

the wall differed in temperature or humidity. Crickets underwent six trials. For the first two trials, crickets were given a choice between a high-lipid and high-protein diet with and without

access to water. In the remaining trials, crickets were only given one of either diet with and

without water. Each trial lasted 24 hours, and cages were periodically checked to remove cricket

carcasses. After removing frass and cricket parts, the ratio and quantity of diet consumption was

measured after controlling for moisture absorption. Data was analyzed using a general linear

model with a gaussian distribution.

Field experiments

Artificial diets for the field experiments had greater macronutrient differences compared

to the laboratory experiments to ensure that arthropods could detect these differences. Here, the

high-lipid diet composed of 1P: 1C: 5L and the high-protein diet had a 5P: 1C: 1L ratio (Table 16

4). Diets were placed in bottlecaps and dried as before. Bottlecaps were then attached to small

petri dishes (Fig 5) with a non-toxic glue dot (0.5” removable dot, Glue Dots Intl., Germantown,

WI). The bottlecaps were used for simplicity, while the small petri dish captured particles of food displaced from the bottlecap (modified from Clissold et al. 2014). For the field experiments, food items were placed inside small labeled plastic bags and then weighed to 0.01 mg (Micro

Balance, model XPE56, Mettler Toledo, Columbus, OH). When diets were collected from the field, they were placed into the same bags as before. Diets were then dried inside their open plastic bags, dirt and frass was removed, then they were weighed a final time within their bags.

From June – August 2016, I selected eight areas with high (> 50%) and eight with low (<

50%) impervious surface, calculated within 500-m radius circles. Selected areas were spread

evenly throughout Toledo, OH (Fig 6), and each area was no more than 15 km from the city

center and no less than 3 km from each other. Within each 500-m radius circle, I selected two

smaller high and low impervious surface sites (50% cut-off), calculated within 50-m radius

circles (ensuring a difference of at least 30% impervious surface between the high and the low

site). Thus, I examined high and low levels of impervious surface at a local scale nested within

high and low levels of impervious surface at a much coarser scale (Table 4, Fig 7). The

categorical methods used to determine an appropriate site were necessary due to Toledo’s

landscape features, accessibility of sites, and obtained permissions. I used the 50-m radius

impervious surface categories in certain graphs, to improve interpretability.

Within each site, I selected six trees that were no less than 10 m apart, and I placed one

dietary trial setup at each tree (Fig 5). One high-lipid diet and one high-protein diet and either a

wet (filled with deionized water) or dry water pillow (small pouches filled with a polymer that

absorbs water; Cricket water pillows, Zilla, Franklin, WI) was placed inside cages to exclude 17 mammals. Cages were made of hardware wire laced together with green floral wire and were fixed to the ground with landscape staples. To prevent rain or UV radiation from altering the pillows and diet, I placed a lid made of a large petri dish sprayed with translucent UV protectant

(Model 1305 Gallery Series, Krylon Products Group, Cleveland, OH), completely covering the top of the cage. The lid could also be removed to make observations with minimal disturbance to the arthropods inside. Treatments were assigned to dietary trial setups in a stratified order. The addition of a water source had two purposes: to determine 1) if water availability altered the ratio or quantity of consumption of high-lipid or high-protein diet and 2) whether the effect of water on lipid and protein consumption varied with impervious surface.

I measured temperature and humidity for the entire sampling period using three data

loggers (Thermochron iButton, model DS 1923, Maxim Inc., San Jose, CA), placed within cages,

spread evenly within a site. Each iButton was attached to a Styrofoam covering that protected

from solar radiation while allowing exchange with the atmosphere. I also measured soil moisture

(SM 150 soil moisture sensor, Dynamax, Houston, TX) and canopy cover (Mobile application

software, HabitApp v. 1.1, Scrufster) three times, within 0.5 m of each cage. The mobile

application was used to assess canopy cover to save time in the field, and it was verified to be

comparable to a densiometer before use. I took photographs at each visitation and noted what

and where an arthropod was located within a cage if it moved after we lifted the lid and I was not

able to photograph it.

For each of 16 site pairs, I visited each pair in random order during the first sampling

period, in reverse order during the second sampling period, and in the original order during the

third sampling period. Each site was visited for three days for each sampling period: cages were

placed on the first day, sites were visited during the morning and at night on the second day, and 18 cages were removed on the third day. Visitation on the first and third days required an entire day, and it was not practical to visit sites twice on these days. Cages were not placed during a storm, but cages were visited on the second and third day, regardless of weather conditions.

I first calculated the lipid and protein consumption for each cage on each date and then

took the mean of these values across cages and dates, generating a single value per site to avoid

pseudoreplication and reduce complexity of the statistical analyses. I then tested my hypotheses

using two statistical models. First, to determine if L: P consumption changed with increased

impervious surface, I used likelihood ratio tests of linear mixed effects models with percent

impervious surface and water pillow wetness as fixed effects and site as a random effect. Second,

to determine which environmental factor is most associated with diet consumption, I used a

linear mixed effects model that examined interactive effects of water pillow wetness, site mean

soil moisture, site mean temperature, and canopy cover as fixed effects, and site as a random

effect. Fixed factors were dropped one at a time from each of the full models above using Bolker

et al. (2009) source code which tested significant changes in likelihood. Normality and equality

of variance were checked using normal probability plots of residuals and graphs of residuals vs.

fitted estimates, respectively. Average L: P consumption was log-transformed to satisfy these

assumptions. All analyses were conducted in R v. 3.3.2, with nlme and lme4 packages.

19

Results

Laboratory experiments

For our laboratory feeding trials, insufficient levels of consumption prevented strong conclusions, but we discovered a trend that matched our hypothesis; water-limited crickets tended to reduce their dry food consumption (df = 1, p < 0.01) and select the diet high in lipids

(Fig 8; df = 1, p < 0.43).

Field experiments

In the field, L: P consumption was predicted by additive effects of impervious surface

(within 50 m radius) (χ2 = 8.36, df = 1, p < 0.01; Fig 9) and marginally by water pillow type (χ2

= 3.79, df = 1, p = 0.05). As predicted, greater urbanization coincided with greater lipid consumption relative to protein and wet water pillows seemed to reduce the slope of the relationship (Fig 9).

Dietary changes were also influenced by environmental factors (Table 8). Temperature and canopy cover had a significant interactive effect on diet (χ2 = 4.17, df = 1, p = 0.04), and water pillow wetness had an additive effect (χ2 = 4.79, df = 1, p = 0.03). Though difficult to parse all environmental factors, higher temperatures led to increased lipid consumption relative to protein, but only when canopy cover was also high (Fig 11), and addition of water reduced lipid consumption relative to protein. Temperature (df = 1, p < 0.01, r2 = 0.34) and soil moisture

(df = 1, p = 0.01, r2 = 0.03) were positively correlated with impervious surface, but canopy cover

was not (df = 1, p = 0.25, r2 = 0.002).

20

Discussion

Findings from nutritional physiology and ecology (Simpson and Raubenheimer 2012) suggest that organisms regulate their intake targets (protein: carbohydrate: lipid consumption) and that this may vary based on their physiological state. Here, intake targets of lipids relative to proteins vary with urbanization, temperature, canopy cover, and water availability. Specifically, lipid demand increased with impervious surface or the combined influence of temperature and canopy cover, but water addition reduced lipid consumption. These results support the hypothesis that desiccation causes animals to alter their nutritional demand in a manner that

helps maintain water balance.

Much of the observed response in high impervious surface areas could be attributed to

ants, while crickets, isopods, and harvestmen dominated in rural areas and city parks (J. Becker

and K. McCluney, personal observations). Ants play important roles in food webs and

ecosystems in and outside cities (Uno et al. 2010, Youngsteadt et al. 2014, Penick et al. 2015):

ants aerate the soil when building nests, are important generalists that consume decaying organic

matter (DOM), kill pests such as termites and tent caterpillars (Tilman 1978), tend and protect

aphids, membracids, and staphylinid beetles (Coovert 2005) from other predators, facilitate seed

dispersal (Howe and Smallwood 1982), and serve as prey to other arthropods (Howard et al.

1990), mammals (Möbius et al. 2008), birds (Davies 1977), and herpetofauna (Toft 1981,

Duellman 1990). Like ants, isopods are an important contributor to the brown food web,

shredding DOM, aerating the soil, and returning nutrients to the soil. Crickets and harvestmen

are also important generalists that consume various prey items, eat DOM, and serve as food for

other organisms. If these key organisms experience water limitation, a change in lipid and

protein requirements could greatly affect local ecological processes. 21

Overall, there was a strong association between impervious surfaces and a preference for high-lipid foods, supporting our hypothesis. Moreover, the response to water addition suggests that part of the effect of impervious surface was driven by water availability. However, other factors could have also driven the increased consumption of lipids. For instance, higher impervious surface is associated with higher temperatures and higher temperatures could lead to higher metabolic rates and thus increase energy demand. Because lipids are of high energy density, increased temperatures could have contributed to higher lipid demand. Indeed, there was an association between temperature and lipid intake relative to protein (Fig 11). But due to reduction of lipid demand with greater water availability, water must partly be driving patterns of macronutrient consumption. Thus, I suggest that both temperature and water likely directly alter nutrient demand.

In summary, ground arthropods in a mesic northern city (Toledo, OH) alter their intake targets in response to impervious surfaces, temperature and canopy cover, and water limitation.

This finding could have implications for trophic dynamics and distribution patterns. Because ants were responsible for much of the response in high impervious surface areas, and because ants play key roles in food webs and ecosystems, increased demand for lipids with urbanization or climate change among this group could have major consequences for urban food webs and ecosystem services. For example, a recent study by Youngsteadt et al (2014) compared food waste removal between street medians and parks within Manhattan, New York. They offered three food items with different nutrient content (cookies, hot dogs, and potato chips – all foods high in fat), and they found that arthropods in street medians were responsible for 2-3 times more food waste removal compared to sites in parks, 71% of which were ants. Thus, ants in street medians could have increased overall consumption of food waste to satisfy lipid demand. Greater 22 investigation of the effects of water balance on nutrient demand and food web dynamics is needed. But these findings suggest that a basic understanding of animal physiology may help predict consequences of environmental change (climate, landscape) for animals and food webs. 23

REFERENCES

Allen, D. C., McCluney, K. E., Elser, S. R., & Sabo, J. L. (2014). Water as a trophic currency in

dryland food webs. Frontiers in Ecology and the Environment, 12(3), 156–160.

http://doi.org/10.1890/130160

Bayley, M. (1999). Water Vapor Absorption in Arthropods by Accumulation of Myoinositol and

Glucose. Science, 285(5435), 1909–1911. http://doi.org/10.1126/science.285.5435.1909

Bell, G. P., Bartholomew, G. A., & Nagy, K. (1986). The Roles of energetics, water economy,

foraging behavior, and geothermal refugia in the distribution of the bat, Macrotus ..., (May

2014). http://doi.org/10.1007/BF01101107

Benoit, J. B., & Denlinger, D. L. (2010). Meeting the challenges of on-host and off-host water

balance in blood-feeding arthropods, 56(10), 1366–1376.

http://doi.org/10.1016/j.jinsphys.2010.02.014.Meeting

Brazel, A., Selover, N., Vose, R., & Heisler, G. (2000). The tale of two climates - Baltimore and

Phoenix urban LTER sites. Climate Research, 15(2), 123–135.

http://doi.org/10.3354/cr015123

Buckley, L. B., & Jetz, W. (2007). Environmental and historical constraints on global patterns of

amphibian richness, (February), 1167–1173. http://doi.org/10.1098/rspb.2006.0436

Chown, S. L. (2002). Respiratory water loss in insects. In Comparative Biochemistry and

Physiology - A Molecular and Integrative Physiology (Vol. 133, pp. 791–804).

http://doi.org/10.1016/S1095-6433(02)00200-3 24

Claussen, D. L., Hopper, R. A., & Sanker, A. M. (2000). The effects of temperature, body size,

and hydration state on the terrestrial locomotion of the crayfish Orconectes rusticus. Journal

of Crustacean Biology, 20(2), 218–223. http://doi.org/10.1651/0278-0372(2000)020

Clissold, F. J., Kertesz, H., Saul, A. M., Sheehan, J. L., & Simpson, S. J. (2014). Regulation of

water and macronutrients by the , Chortoicetes terminifera. Journal

of Physiology, 69, 35–40. http://doi.org/10.1016/j.jinsphys.2014.06.011

Coast, G. M., Orchard, I., Phillips, J. E., & Schooley, D. A. (2002). Insect diuretic and

antidiuretic hormones (pp. 279–409). http://doi.org/10.1016/S0065-2806(02)29004-9

Coe, S. J., & Rotenberry, J. T. (2003). Water availability affects clutch size in a desert sparrow.

Ecology, 84(12), 3240–3249. http://doi.org/10.1890/02-0789

Coovert, G. A. (2005). The ants of Ohio. Ohio Biological Survey, Inc.

Davies, N. B. (1977). Prey selection and social behaviour in wagtails (Aves: Motacillidae). The

Journal of Animal Ecology, 37-57.

Deguines, N., Brashares, J. S., & Prugh, L. R. (2016). Precipitation alters interactions in a

grassland ecological community. Journal of Animal Ecology. http://doi.org/10.1111/1365-

2656.12614

Dias, A. T. C., Krab, E. J., Mariën, J., Zimmer, M., Cornelissen, J. H. C., Ellers, J., … Berg, M.

P. (2013). Traits underpinning desiccation resistance explain distribution patterns of

terrestrial isopods. Oecologia, 172(3), 667–677. http://doi.org/10.1007/s00442-012-2541-3 25

Dinh, V. N., Janssen, A. R. M., & Sabelis, M. W. (1988). The influence of humidity and water

availability on the survival of Amblyseius idaeus and Amblyseius anonymus (Acarina :

Phytoseiidae). Experimental and Applied Acarology, 4, 27–40.

Duellman, W. E. (1990). Herpetofaunas in Neotropical rainforests: comparative composition,

history, and resource use. Four neotropical rainforests, 455-505.

Dussutour, A., & Simpson, S. J. (2008). Description of a simple synthetic diet for studying

nutritional responses in ants. Insectes Sociaux, 55(3), 329–333.

http://doi.org/10.1007/s00040-008-1008-3

Fanson, B. G., Yap, S., & Taylor, P. W. (2012). Geometry of compensatory feeding and water

consumption in Drosophila melanogaster. Journal of Experimental Biology, 215(5), 766–

773. http://doi.org/10.1242/jeb.066860

Finkler, M. (1999). Influence of water availability during incubation on hatchling size, body

composition, desiccation tolerance, and terrestrial locomotor performance in the snapping

turtle Chelydra serpentina. Physiological and Biochemical Zoology, 72, 714–722.

Forman R. T. T. (2014). Urban ecology: science of cities. Cambridge University Press

Frank, L. (1988). Diet Selection by a Heteromyid Rodent: Role of Net Metabolic Water

Production, 69(6), 1943–1951.

Forman R. T. T. (2014). Urban Ecology: Science of Cities. Cambridge University Press. 26

Frizzi, F., Rispoli, A., Chelazzi, G., & Santini, G. (2016). Effect of water and resource

availability on ant feeding preferences: a field experiment on the Mediterranean ant

Crematogaster scutellaris. Insectes Sociaux, 63(4), 565–574. http://doi.org/10.1007/s00040-

016-0500-4

Gregory, J., Dukes, M., Jones, P., & Miller, G. (2006). Effect of urban soil compaction on

infiltration rate. Journal of Soil and Water Conservation, 61(3), 117–124.

Groffman, P. M., Cavender-Bares, J., Bettez, N. D., Grove, J. M., Hall, S. J., Heffernan, J. B., …

Steele, M. K. (2014). Ecological homogenization of urban USA. Frontiers in Ecology and

the Environment, 12(1), 74–81. http://doi.org/10.1890/120374

Hadley, N. F. (1994). Water relations of terrestrial arthropods. Academic Press.

Hamza, M. A., & Anderson, W. K. (2005). Soil compaction in cropping systems: A review of the

nature, causes and possible solutions. Soil and Tillage Research, 82(2), 121–145.

http://doi.org/16/j.still.2004.08.009

Hanks, L. M., & Denno, R. F. (1993). Natural enemies and plant water relations influence the

distribution of an armored scale insect. Ecology, 74(4), 1081–1091.

Hawkins, B. A., Diniz-Filho, J. A. F., & Soeller, S. A. (2005). Water links the historical and

contemporary components of the Australian bird diversity gradient. Journal of

Biogeography, 32(6), 1035–1042. http://doi.org/10.1111/j.1365-2699.2004.01238.x 27

Howard, R. W., Akre, R. D., & Garnett, W. B. (1990). Chemical mimicry in an obligate predator

of carpenter ants (Hymenoptera: Formicidae). Annals of the Entomological Society of

America, 83(3), 607-616.

Howe, H. F., & Smallwood, J. (1982). Ecology of seed dispersal. Annual review of ecology and

systematics, 13(1), 201-228.

Imhoff, M. L., Zhang, P., Wolfe, R. E., & Bounoua, L. (2010). Remote Sensing of Environment

Remote sensing of the urban heat island effect across biomes in the continental USA.

Remote Sensing of Environment, 114(3), 504–513. http://doi.org/10.1016/j.rse.2009.10.008

Jindra, M., & Sehnal, F. (1990). Linkage between diet humidity, metabolic water production and

heat dissipation in the larvae of Galleria mellonella. Insect Biochemistry, 20(4), 389–395.

http://doi.org/10.1016/0020-1790(90)90059-4

Keil, P., Simova, I., Hawkins, B. a, & Va, I. R. E. N. a S. I. M. O. (2008). Water-energy and the

geographical species richness pattern of European and North African dragonflies (Odonata).

Insect Conservation and Diversity, 1, 142–150. http://doi.org/DOI 10.1111/j.1752-

4598.2008.00019.x

Lehmann, F.-O. (2001). Matching Spiracle Opening to Metabolic Need During Flight in

Drosophila. Science, 294(5548), 1926–1929. http://doi.org/10.1126/science.1064821

Lensing, J. R., & Wise, D. H. (2006). Predicted climate change alters the indirect effect of

predators on an ecosystem process. Proceedings of the National Academy of Sciences of the

United States of America, 103(42), 15502–15505. http://doi.org/10.1073/pnas.0607064103 28

Lewis, A. C., & Bernays, E. A. (1985). Feeding behavior: selection of both wet and dry food for

increased growth in Schistocerca gregaria nymphs. Entomological Experimentalis et

Applicata, 37, 105–112. http://doi.org/10.1055/s-0030-1250443

Lighton, J. R. B., Schilman, P. E., & Holway, D. A. (2004). The hyperoxic switch: assesing

respiratory water loss rates in tracheate arthropods with continous gas exchange. J. Exp.

Biol., 207, 4463–4471.

Lighton, J. R. B., Garrigan, D. a, Duncan, F. D., & Johnson, R. a. (1993). Spiracular Control of

Respiratory Water Loss in Female Alates of the Harvester Ant Pogonomyrmex Rugosus.

Journal of Experimental Biology, 179, 233–244.

Loarie, S., & Pimm, S. L. (2009). Fences and artificial water affect African savannah elephant

movement patterns. BIOLOGICAL CONSERVATION, (June 2017).

http://doi.org/10.1016/j.biocon.2009.08.008

McCluney, K. E. (2017). Implications of animal water balance for terrestrial food webs. Current

Opinion in Insect Science, 23, 13-21.

McCluney, K. E., & Date, R. C. (2008). The effects of hydration on growth of the house cricket,

Acheta domesticus. Journal of Insect Science (Online), 8(32), 1–9.

http://doi.org/10.1673/031.008.3201

McCluney, K. E., Belnap, J., Collins, S. L., González, A. L., Hagen, E. M., Nathaniel Holland,

J., … Wolf, B. O. (2012). Shifting species interactions in terrestrial dryland ecosystems 29

under altered water availability and climate change. Biological Reviews.

http://doi.org/10.1111/j.1469-185X.2011.00209.x

McCluney, K. E., & Sabo, J. L. (2009). Water Availability Directly Determines Per capita

Consumption at Two Trophic Levels. Ecology, 90(6), 1463–1469.

McCluney, K. E., & Sabo, J. L. (2012). River drying lowers the diversity and alters the

composition of an assemblage of desert riparian arthropods. Freshwater Biology, 57(1), 91–

103. http://doi.org/10.1111/j.1365-2427.2011.02698.x

McCluney, K. E., & Sabo, J. L. (2016). Animal water balance drives top-down effects in a

riparian forest — implications for terrestrial trophic cascades. Proceedings of the Royal

Society B: Biological Sciences, 283, 20160881. http://doi.org/10.1098/rspb.2016.0881

Möbius, Y., Boesch, C., Koops, K., Matsuzawa, T., & Humle, T. (2008). Cultural differences in

army ant predation by West African chimpanzees? A comparative study of microecological

variables. Animal Behaviour, 76(1), 37-45.

Nicolson, S. W. (2009). Water homeostasis in bees, with the emphasis on sociality. Journal of

Experimental Biology, 212(3), 429–434.

O’Donnell, M. J., & Spring, J. H. (2000). Modes of control of insect Malpighian tubules:

Synergism, antagonism, cooperation and autonomous regulation. Journal of Insect

Physiology, 46(2), 107–117. http://doi.org/10.1016/S0022-1910(99)00119-5 30

Penick, C. A., Savage, A. M., Dunn, R. R., Hoornweg, D., Bhada-Tata, P., Agency, U. S. E. P.,

… Marussich, W. (2015). Stable isotopes reveal links between human food inputs and urban

ant diets. Proceedings of the Royal Society B: Biological Sciences, 282(1806), 20142608.

Roberts, S. P., Quinlan, M. C., & Hadley, N. F. (1994). Interactive effects of humidity and

temperature on water loss in the lubber grasshopper Romalea guttata. Comparative

Biochemistry and Physiology -- Part A: Physiology, 109(3), 627–631.

http://doi.org/10.1016/0300-9629(94)90202-X

Sabo, J. L., Mccluney, K. E., Marusenko, Y., Keller, A., & Soykan, C. U. (2008). Greenfall

Links Groundwater to Aboveground Food Webs in Desert River Floodplains. Ecological

Monographs, 78(4), 615–631.

Schilman, P. E., Lighton, J. R. B., & Holway, D. A. (2005). Respiratory and cuticular water loss

in insects with continuous gas exchange: Comparison across five ant species. Journal of

Insect Physiology, 51(12), 1295–1305. http://doi.org/10.1016/j.jinsphys.2005.07.008

Schowalter, T. D., Lightfoot, D. C., & Whitford, W. G. (1999). Diversity of Arthropod

Responses to Host-plant Water Stress in a Desert Ecosystem in Southern New Mexico. The

American Midland Naturalist, 142(2), 281–290. http://doi.org/10.1674/0003-

0031(1999)142[0281: DOARTH]2.0.CO;2

Simpson, S. J., & Raubenheimer, D. (2012). The nature of nutrition: a unifying framework from

animal adaptation to human obesity. Princeton: Princeton University Press. 31

Spiller, D. A., & Schoener, T. W. (2008). Climatic control of trophic interaction strength: The

effect of lizards on spiders. Oecologia, 154(4), 763–771. http://doi.org/10.1007/s00442-

007-0867-z

Steele, M. K., & Heffernan, J. B. (2014). Morphological characteristics of urban water bodies:

mechanisms of change and implications for ecosystem function, 24(5), 1070–1084.

Taha, H. (1997). Urban climates and heat islands: albedo, evapotranspiration, and anthropogenic

heat. Energy and Buildings, 25(2), 99–103. http://doi.org/10.1016/S0378-7788(96)00999-1

Tieleman, B. I., Williams, J. B., & Visser, G. H. (2004). Energy and water budgets of larks in a

life history perspective: Parental effort varies with aridity. Ecology.

http://doi.org/10.1890/03-0170

Tilman, D. (1978). Cherries, Ants and Tent Caterpillars: Timing of Nectar Production in

Relation to Susceptibility of Caterpillars to Ant Predation. Ecology, 59(4), 686-692.

doi:10.2307/1938771

Toft, C. (1981). Feeding Ecology of Panamanian Litter Anurans: Patterns in Diet and Foraging

Mode. Journal of Herpetology, 15(2), 139-144. doi:10.2307/1563372

Uno, S., Cotton, J., & Philpott, S. M. (2010). Diversity, abundance, and species composition of

ants in urban green spaces. Urban Ecosystems, 13(4), 425–441.

Youngsteadt, E., Henderson, R. C., Savage, A. M., Ernst, A. F., Dunn, R. R., & Frank, S. D.

(2014). Habitat and species identity, not diversity, predict the extent of refuse consumption

by urban arthropods. Global Change Biology. http://doi.org/10.1111/gcb.12791 32

Zhao, L., Lee, X., Smith, R. B., & Oleson, K. (2014). Strong contributions of local background

climate to urban heat islands. Nature, 511(7508), 216–219.

http://doi.org/10.1038/nature13462 33

APPENDIX A. TABLES Table 1. Summary of site characteristics. Values show the mean and standard error

Region Oak Openings Oak Openings BGSU BGSU Toledo Toledo Site Sand Clay Street Greenspace Street Greenspace Observation 0.44 ± 0.04 0.38 ± 0.04 0.08 ± 0.03 0.25 ± 0.06 0.06 ± 0.02 0.19 ± 0.04 frequency per tree Observation 0.33 ± 0.05 0.14 ± 0.04 0.04 ± 0.02 0.11 ± 0.03 0.04 ± 0.02 0.08 ± 0.02 frequency per tree Observation 0.45 ± 0.06 0.38 ± 0.04 0.08 ± 0.03 0.25 ± 0.04 0.07 ± 0.02 0.19 ± 0.04 frequency per day Observation 0.32 ± 0.03 0.15 ± 0.04 0.04 ± 0.01 0.10 ± 0.02 0.04 ± 0.02 0.08 ± 0.02 frequency per day Impervious 0 ± 0 2.91 ± 13.34 74.56 ± 1.97 44.36 ± 2.38 68.57 ± 1.32 45.64 ± 0.70 surface Maximum soil 19.81 ± 1.58 33.90 ± 5.45 29.41 ± 1.86 46.82 ± 1.75 62.46 ± 4.02 42.44 ± 1.70 moisture per tree Minimum soil 0.43 ± 0.24 2.25 ± 0.96 1.46 ± 0.34 12.04 ± 1.12 3.99 ± 1.12 3.22 ± 0.75 moisture per tree Daily maximum 18.83 ± 2.02 37.82 ± 3.90 30.17 ± 1.69 47.18 ± 1.36 49.2 ± 4.54 36.82 ± 1.74 soil moisture Daily minimum 1.25 ± 0.28 3.43 ± 0.68 1.13 ± 0.43 11.27 ± 0.89 4.01 ± 0.95 2.12 ± 0.65 soil moisture Daily maximum 28.75 ± 0.43 28.28 ± 0.48 27.2 ± 0.57 26.96 ± 0.68 28.08 ± 0.59 27.62 ± 0.58 temperature 34

Daily minimum 20.80 ± 0.44 20.15 ± 0.41 22.96 ± 0.75 23.13 ± 0.70 23.98 ± 0.57 24.04 ± 0.57 temperature 35

Table 2. Likelihood ratio test results of water pillow wetness and impervious surface on observation frequency per tree. To find the best fit for our model, we removed one complex interactive effect at a time until a significant environmental effect on L: P is found. Here, water demand significantly declines with increased impervious surface. p < 0.05

Model component removed df ΔAIC LRT (χ2) p-value

- Water pillow wetness * impervious surface 1 6.22 8.21 < 0.01*

- Water pillow wetness 1 23.18 25.19 < 0.01*

- Impervious surface 1 20.27 22.28 < 0.01* 36

Table 3. Likelihood ratio test results of environmental characteristics on observation frequency per day. Water demand is effected by multiple interactions among mean site temperature, mean site soil moisture, daily site soil moisture, and daily site temperature. P < 0.05*, P < 0.01**

Model component removed df ΔAIC LRT (χ2) p-value

- Mean site temperature * Mean site soil moisture * 1 26.76 28.78 < 0.01** Daily site temperature * Daily site soil moisture 37

Table 4. Biodiversity within wet and dry sites in each region. S = richness, H = Shannon index, and H/ln(S) = Evenness.

Region Site S H H/(lnS)

BGSU Dry 6 1.340 0.748

BGSU Wet 6 1.039 0.580

Oak Openings Dry 11 2.263 0.944

Oak Openings Wet 14 2.158 0.818

Toledo Dry 5 1.203 0.745

Toledo Wet 6 1.096 0.612

38

Table 5. Artificial diet ingredients. Ingredients modified from Dussutour and Simpson (2008).

Each cage contained both a high-lipid diet (lab: 1P: 1C: 3L; field: 1P: 1C: 5L) and a high-protein diet (lab: 3P: 1C: 1L; field: 5P :1C: 1L).

Ingredient Manufacturer High-lipid High-protein

Unflavored, Jarrow Whey protein (g) 2.79 28.94 Formulas

Unflavored, True Calcium caseinate (g) 2.75 21.88 Nutrition

Whole egg powder (g) Hoosier Hill Farm 10.87 10.35

Sucrose (g) Pure sugar, Great Value 8.75 6.04

Canola oil (mL) Pure canola oil, Crisco 44.83 2.79

Vanderzant vitamin mixture for Sigma-Aldrich 2 2 insects (g)

Methyl-4 hydroxybenzoate (g) Sigma-Aldrich 1 1

Agar-agar powder (g) Landor Trading Co. 4 4

39

Table 6. Site imperviousness and distance to city center. Site selection necessitated a categorical approach, using % impervious surface that was calculated within buffers. In each image, red squares are high % impervious surface sites, while yellow triangles are low % impervious surface sites. Images obtained from Google Earth, updated December 17, 2015.

Buffer Distance to Impervious Impervious size Site name Coordinates city center Image surface (%) Category (m) (km)

Cemetery 41°41'0.03"N 500 26.10 Low 4.5 and Willys 83°34'36.56"W

Woodlawn 41°40'55.43"N 50 19.67 Low Cemetery 83°34'46.93"W

41°41'3.61"N 50 Willys Park 56.76 High 83°34'26.87"W

Woodlands 41°33'37.18"N 500 and 29.00 Low 12.1 83°36'44.40"W Church 40

Woodlands 41°33'40.23"N 50 20.21 Low Park 83°36'43.38"W

Grace United 41°33'33.88"N 50 Methodist 62.69 High 83°36'45.51"W Church of Perrysburg

Calvary 41°39'40.36"N 500 Cemetery 32.44 Low 5.2 83°35'56.05"W and School

Cavalry 41°39'35.71"N 50 5.79 Low Cemetery 83°35'52.10"W

St. Francis 41°39'44.21"N 50 de Sales 69.21 High 83°35'59.46"W School

41

Winterfield 41°38'7.82"N 500 Park and 36.16 Low 9.5 83°38'50.80"W School

Winterfield 41°38'13.16"N 50 23.43 Low Park 83°39'0.52"W

Winterfield Venture 41°38'2.33"N 50 68.80 High Charter 83°38'40.92"W Academy

Sunrise 41°34'32.17"N 500 Banquet 38.37 Low 13.3 83°25'17.10"W Center

Sunrise 41°34'32.23"N 50 Banquet 0.00 Low 83°25'10.88"W Center

Sunrise 41°34'31.66"N 50 Banquet 51.56 High 83°25'23.97"W Center

42

Loop and 41°35'26.01"N 500 45.88 Low 8.3 Walbridge 83°29'27.01"W

41°35'20.55"N 50 Loop Park 28.75 Low 83°29'10.99"W

Walbridge 41°35'31.57"N 50 58.93 High Apartments 83°29'42.55"W

Trilby and 41°42'58.08"N 500 47.43 Low 9.7 Wichita 83°37'20.80"W

41°42'54.39"N 50 Trilby Park 22.92 Low 83°37'17.16"W

41°43'1.17"N 50 Wichita Rd. 67.50 High 83°37'24.10"W

43

Byrne and 41°36'0.91"N 500 48.65 Low 9.5 Cedar 83°37'24.32"W

41°35'58.84"N 50 Byrne Park 33.00 Low 83°37'17.52"W

Cedar Creek 41°36'2.87"N 50 84.50 High Church 83°37'31.49"W

East Pt. 41°35'59.67"N 500 and 50.15 High 9.2 83°27'19.72"W Woodville

East Pt. 41°35'46.39"N 50 18.75 Low Blvd. 83°27'13.98"W

Woodville 41°36'11.03"N 50 61.14 High Mall 83°27'25.16"W

44

Cullen 41°42'29.39"N 500 Park and 50.37 High 7.5 83°28'52.87"W Library

41°42'21.21"N 50 Cullen Park 10.15 Low 83°28'32.39"W

Ottawa School 41°42'35.76"N 50 50.13 High Public 83°29'10.83"W Library

Calvary 41°34'33.97"N 500 Assembly 52.18 High 13.5 83°39'32.57"W of God

Calvary 41°34'32.58"N 50 Assembly of 17.80 Low 83°39'24.96"W God

Calvary 41°34'34.82"N 50 Assembly of 79.63 High 83°39'39.37"W God

45

Marvin and 41°40'47.92"N 500 53.26 High 7.2 Middlesex 83°37'4.65"W

Marvin 41°40'51.46"N 50 13.60 Low Playground 83°37'1.98"W

Middlesex 41°40'43.66"N 50 57.69 High Dr. 83°37'7.63"W

Beech and 41°36'37.77"N 500 55.10 High 5.3 Superior 83°33'20.86"W

Beech St. 41°36'29.29"N 50 9.93 Low Park 83°33'22.15"W

41°36'44.46"N 50 Superior St. 79.79 High 83°33'19.88"W

46

Bear and 41°36'41.80"N 500 56.35 High 14.3 Mail 83°41'48.57"W

Bear Creek 41°36'31.50"N 50 0.00 Low Park 83°41'57.30"W

41°36'50.78"N 50 Mail Dr. 69.57 High 83°41'40.84"W

Greenwood 41°43'24.73"N 500 59.45 High 8.2 and Gage 83°34'37.22"W

Greenwood 41°43'31.04"N 50 13.56 Low Park 83°34'38.98"W

41°43'18.58"N 50 Gage Rd. 56.89 High 83°34'35.32"W

47

41°39'20.94"N 500 City Center 79.04 High 0 83°32'14.66"W

Civic Center 41°39'27.06"N 50 44.14 Low Mall 83°32'6.41"W

Library 41°39'14.75"N 50 87.17 High Square 83°32'22.35"W

48

Table 7. Likelihood ratio test results of water pillow wetness and impervious surface on diet.

Terms were removed from the model, iteratively, from more complex to less complex until a significant effect on L: P was found. Here, L: P ratio of consumption is affected by impervious surface and marginally by water pillow wetness. p < 0.05*

Model component removed df ΔAIC LRT (χ2) p-value

- Water pillow wetness * impervious surface 1 0.12 2.12 0.15

- Water pillow wetness 1 1.80 3.79 0.05

- Impervious surface 1 6.37 8.36 0.003* 49

Table 8. Likelihood ratio test results of water pillow wetness and environmental characteristics.

Diet consumption is effected by water pillow wetness and the interaction between canopy cover and site temperature. p < 0.05*

Model component removed df ΔAIC LRT (χ2) p-value

- Water pillow wetness * soil moisture * air temperature 1 -0.53 1.47 0.23 * canopy cover

- Water pillow wetness * soil moisture * air temperature 1 -1.94 0.06 0.81

- Water pillow wetness * soil moisture * canopy cover 1 1.12 3.12 0.08

- Water pillow wetness * air temperature * canopy cover 1 -1.87 0.13 0.72

- Soil moisture * air temperature * canopy cover 1 -1.47 0.53 0.47

- Water pillow wetness * soil moisture 1 -1.76 0.24 0.63

- Water pillow wetness * air temperature 1 -1.46 0.54 0.46

- Water pillow wetness * canopy cover 1 -1.86 0.14 0.71

- Soil moisture * air temperature 1 -0.6 1.4 0.24

- Air temperature * canopy cover 1 2.17 4.17 0.04*

- Water pillow wetness 1 2.79 4.79 0.03*

- Soil moisture 1 -1.68 0.32 0.57

50

APPENDIX B. FIGURES

(a) (b) (c)

Figure 1. Observation frequency per tree with impervious surface. Average percent impervious surface is calculated within a 25m buffer around each tree. A) About 23% of observed frequency were found on wet pillows, and 12.3% occurred on dry pillows, indicating that water demand occurs about 11% of the time in this mesic region. B) Wet and dry pillow frequencies significantly decline with increased impervious surface. C) Arthropod abundance was correlated with impervious surface, which partially, but not fully explains decline in

(b). All pillow frequencies clustered at 0% impervious surface per tree come from the Oak Openings region. 51

Figure 2. Observation frequency per day with increased daily soil moisture across all sites.

Water demand behavior is correlated with daily soil moisture (wet pillows: (df = 1, p < 0.01, r2 =

0.14); dry pillows: (df = 1, p = 0.04, r2 = 0.04).

52

Figure 3. Effects of daily soil moisture fluctuations and mean site temperature (climate) on daily

water demand (observation frequency per day). The effect of daily changes in soil moisture on

water demand is strongest at generally cooler sites.

53

Figure 4. Abundance of invertebrates on wet and dry pillows. Arthropod responses to water

pillows were dominated by Opiliones, Orthoptera, Hymenoptera, and Collembola at the undeveloped sites (Oak Openings), while developed sites (BGSU and Toledo) were primarily

composed of Hymenoptera (mostly ants, especially to Camponotus, Lasius, and Brachymyrmex).

These genera are widespread in OH and were present at all sites. Ants in Toledo and BGSU were

primarily Camponotus pennsylvanicus.

54

Figure 5. Cage contents and position within sites. Each cage included a wet or dry pillow, a

high-lipid diet, and a high-protein diet. Wet and dry cages were in stratified order within a site.

Three of the six cages had an iButton covered with a Styrofoam cup. Protective lid not shown.

Image shows ants clustered on the high-lipid diet at Trilby Park.

55

Figure 6. Large-scale areas used in this research. Yellow triangles are low % impervious sites,

while red squares are high % impervious surface sites. See Table 1 for coordinates and Figure 8

for an example expansion showing small sites. Image obtained from Google Earth, updated

December 17, 2015. 56

Figure 7. An example of a pair of small-scale sites nested within one large-scale area. Yellow triangles are low % impervious sites, while red squares are high % impervious surface sites. See Table 1 for coordinates and Figure 7 for all large-scale site names. Images obtained from Google Earth, updated December 17, 2015. 57

Figure 8. Lipid and protein consumption in the laboratory. Each food rail represents the L+C: P concentration of the high-lipid and high-protein diets that were offered to crickets in the lab experiment. Crickets were given a choice between the high-lipid and high-protein diet, under conditions when water was and was not available. Per-capita intake target tended to favor lipids when crickets were water limited, but no significant differences were observed, possibly due to low levels of consumption overall. No water: [mean ± se: L+C (0.572 ± 0.117): P (0.193 ±

0.071)]; water: [L+C (1.721 ± 0.286): P (1.405 ± 0.234)]. 58

Figure 9. Lipid and protein consumption with impervious surface. L: P consumption

significantly increased with impervious surface, and water addition marginally reduced the slope

of the relationship. 59

Figure 10. Field intake targets within a geometric space. Each food rail represents the L+C: P concentration of the high-lipid and high-protein diet offered in the field experiment. High impervious surfaces (within 25 m radius) significantly shift site-level intake targets to favor

foods higher in lipids. Each point represents the mean and se of wet and dry pillow food

consumption at each site, relative to high-lipid and high-protein nutrient trajectories. Using

categorical values for impervious surface (low < 50%, high > 50%, allowing a difference of at

least 30% between sites) provide more interpretable graphs. 60

Figure 11. Interactive effects of temperature and canopy cover on diet. L: P is higher at warmer sites with higher levels of canopy cover.