The Effects of Warming on Aquatic Insects – Individual to Community Responses

Dachin N. Frances

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Ecology and Evolutionary Biology University of Toronto

Dachin N. Frances

Doctor of Philosophy

Department of Ecology and Evolutionary Biology University of Toronto

2018 Abstract

Climate change is increasing temperatures globally as well as the frequency and severity of extreme events such as heat waves. Given that the majority of animals on Earth are ectothermic, and therefore physiologically linked to environmental temperatures, it is imperative to predict the impacts of future warming on species. The physiological effects of warming on ectotherms have been studied extensively, yet, species interactions can also be affected by temperature. My thesis assessed how differences in species’ responses to warming affect their interactions and ultimately the structure and dynamics of communities. I first asked if phenological and early developmental responses to warming differed among and within dragonfly species. Warming accelerated growth and developmental rates, however, differences were greater within as opposed to between species in these responses. I also examined whether differences in behavioural responses to warming among species could determine which species became the superior IG predator. Foraging and IGP rates increased with warming but activity level changes were not predictive of the outcome of these interactions. Next, I asked how the interactive effects of warming and predator presence affected prey foraging decisions. Warmer conditions appeared to have a greater impact on prey behaviour that the predator presence, even though the prey’s risk of being eaten increased with temperature. Lastly, I manipulated developmental conditions in pond mesocosms to understand how warming and heat waves affect community structure.

Heat waves similarly increased species’ evenness in communities as warming. However, heat waves further boosted survival compared to ambient and warmed conditions, strengthening consumption on primary consumers within these food webs. Together, my thesis demonstrates the importance of understanding complex interactions between species to make more accurate predictions of the effects of future warming on communities.

iii

Acknowledgments

First and foremost, I would like to thank my supervisor, Shannon McCauley. Her enthusiasm and passion for her work is contagious, which is what initially drew me in at the start of my PhD. Since then, her constant support and endless wisdom have been pivotal in guiding me along this academic journey. In addition, she and Steph Schneider have provided hours of conversation, humor, stories, and advice. Thank you both for your generosity, kindness, and van-shuttling! I also owe Karen Mabry a huge thank you, as she is the one who introduced me to Shannon in the first place. Thanks Karen, for hiring me as a technician my first summer after undergrad and being a great collaborator in the years following.

In addition, I would like to generally thank the University of Toronto’s Department of Ecology and Evolutionary Biology for funding me all these years, the Department of Biology at the University of Toronto Mississauga for generous conference funding and support for BGSS, the office of the Vice Dean Graduate for their generous funding to graduate students, all the graduate students involved in UTMAGS who worked hard to improve graduate student life, and the American Museum of Natural History, the Animal Behavior Society for funding part of my research and trips to conferences.

More specifically, I would like to thank members of my doctoral committee: Doug Currie, Benjamin Gilbert, and Marc Johnson for their feedback and advice on my PhD research. Also, thank you, Marc, for reading and commenting on my thesis while traversing the globe on sabbatical. Thanks also to Sanja Hinić-Frlog, for being a great mentor and teaching me so much about pedagogy and birds! I would also like to thank Helen Rodd for her tireless and sincere commitment to graduate students in EEB. Thank you to Darryl Gwynne for being on my appraisal committee and generally just being an interesting person to talk to. Also, thank you, Mary O’Connor, for being on my defense committee and asking thorough and excellent questions that made me think broadly about my work and how it intersects with the field.

I also thank members of the McCauley lab: Sarah French, Celina Baines, Rosemary Martin, Ilia Ferzoco, Ros Murray, Chris Searcy, Shantel Catania, Tharusha Wijewardena, Tammy Duong, Jy Moon, David Przelicki, Sam Hasbum, and Rosalynn Mitchell. I feel so fortunate to be part of such a supportive, creative, and brilliant group. Thanks also to every ROP/intern/work

iv study/thesis undergraduate student that put in hours of field and lab work – this thesis would have taken years longer if not for you!

Next, I thank the graduate students in the department whom I have grown close with over the past few years: Audrey Reid, Laura Junker, Kyla Ercit, Michelle DiLeo, Becci Rooke, Mark McDougall, Aaron LeBlanc, Krystal Nunes, Amy Wong, Nash Turley, Yara Haridy, Lauren Barth, Ernest Prack, Ruth Rivkin, Molly Hetherington-Rauth, Diego Carmona, James Santangelo, Jessica Brown, Cylita Guy, Arjan Banerjee, Kayla Dias, Alex Paquette, Mireille Golemiec, Chris Wong, Jade Atallah, Andrew Catalano, and many more. Thank you all for making my experience here an amusing and unique time!

Lastly, I’d like to thank my family. Thank you all for your constant support and unconditional love. Living in a different country than you was challenging at times, but I always looked forward to seeing you all once or twice a year and I cherish all of those memories. Of course, I can’t end the acknowledgements section without thanking my husband. Thank you, Connor, for being my foundation, my sentinel, and my number one, unwavering supporter. I am so thankful that we could do this journey together and I’m looking forward to embarking on our future path together after this chapter in our lives.

Table of Contents

Acknowledgments ...... iv

Table of Contents ...... vi

List of Tables ...... x

List of Figures...... xi

List of Appendices ...... xiii

General Introduction ...... 1

1 Introduction ...... 1

1.1 Present and future climate change...... 1 1.1.1 Species responses to climate change ...... 2 1.1.2 Phenological shifts ...... 3 1.1.3 Range shifts ...... 3 1.1.4 Shifts in body size – The temperature-size rule ...... 4

1.2 Study system ...... 5 1.2.1 Odonate responses to climate change ...... 6 1.2.2 Odonates as structuring agents of communities ...... 8

1.3 Thesis overview ...... 11 1.3.1 Chapter 2 ...... 11 1.3.2 Chapter 3 ...... 12 1.3.3 Chapter 4 ...... 12 1.3.4 Chapter 5 ...... 13

1.4 References ...... 13

Effects of warming on early life history ...... 19

2 Abstract ...... 19

2.1 Introduction ...... 19

2.2 Materials and methods ...... 23 2.2.1 Study system ...... 23 2.2.2 Experiment 1: Interspecific comparisons of egg development rate ...... 24 2.2.3 Experiment 2: Effects of warming on early life stages of Le. intacta ...... 25 2.2.4 Statistical analyses: ...... 26

2.3 Results ...... 27 2.3.1 Experiment 1: interspecific comparisons of egg development rate ...... 27 2.3.2 Experiment 2: Effects of warming on early life stages of Le. intacta ...... 27

2.4 Discussion ...... 32 2.4.1 The effects of temperature on inter- and intraspecific egg development rate ....33 2.4.2 The effect of temperature on body size, growth rate and survival of Le. intacta 34

2.5 Conclusion ...... 36

2.6 References ...... 37

Warming drives higher rates of prey consumption and increases rates of intraguild predation ...... 44

3 Abstract ...... 44

3.1 Introduction ...... 44

3.2 Materials and methods ...... 48 3.2.1 Study system ...... 48 3.2.2 Study 1: Examining the influence of temperature on foraging behaviour ...... 49 3.2.3 Study 2: Determining the impact of temperature on intraguild predation ...... 51

3.3 Statistical analyses...... 51 3.3.1 Behavioural assays ...... 51 3.3.2 Maximum realized temperature and behavioural responses to temperature .....52 3.3.3 Intraguild predation trials ...... 53

3.4 Results ...... 53 vii

3.4.1 Temperature and foraging behaviour ...... 53 3.4.2 Changes in foraging were related to maximum realized temperatures ...... 54 3.4.3 Temperature increased IGP and altered the size distribution of dragonfly larvae 57

3.5 Discussion ...... 60 3.5.1 Dragonfly larvae increased the number of prey eaten at warmer temperatures 60 3.5.2 Temperature did not change dragonfly larval activity levels or microhabitat preferences ...... 61 3.5.3 Maximum realized temperatures were associated with behavioural changes ....62 3.5.4 Temperature increases rates of IGP in a species-specific way ...... 62 3.5.5 Temperature increased the frequency of predation on smaller individuals ...... 64

3.6 Conclusions ...... 65

3.7 References ...... 66

How will prey respond to predation risk in a warmer world? ...... 71

4 Abstract ...... 71

4.1 Introduction ...... 71

4.2 Materials and methods ...... 73 4.2.1 Statistical analyses ...... 74

4.3 Results ...... 75

4.4 Discussion ...... 78

4.5 References ...... 79

Experimental warming and heat waves modify pond community composition and trophic dynamics ...... 82

5 Abstract ...... 82

5.1 Introduction ...... 82

viii

5.1 Methods ...... 86 5.1.1 Statistical analyses ...... 89

5.2 Results ...... 91 5.2.1 Temperature differences between treatments ...... 91 5.2.2 Survival ...... 91 5.2.3 Evenness ...... 92 5.2.4 Development ...... 92 5.2.5 Zooplankton ...... 93 5.2.6 Trophic position ...... 93

5.3 Discussion ...... 99 5.3.1 Effects of heat waves and warming on survival, evenness, and development ....99 5.3.2 Stable isotopes and food web structure ...... 101

5.4 Conclusion ...... 102

Acknowledgements ...... 103

5.5 References ...... 103

General Discussion ...... 109

6 Discussion ...... 109

6.1 Species’ responses to warming...... 110 6.1.1 Inter- and intraspecific responses to warming ...... 110 6.1.2 The temperature-size rule in dragonflies ...... 111 6.1.3 Climate change experiments: warming regimes, scale, and trade-offs ...... 113 6.1.4 Mechanisms driving community patterns ...... 114

6.2 Conclusions ...... 115

6.3 References ...... 116

Appendices ...... 119

List of Tables

Table 1.1 Odonate species used in this thesis, taxonomic family, geographic distribution, and the chapters in which they are found………………………………………………………………9

Table 2.1 Results from the log-likelihood ratio tests on the effects of fixed and random factors on egg development rate, hatchling head width, and hatchling growth rate based on a linear mixed-effects (LME) model…………………………………………………………………..28

Table 5.1 Comparison of the calculated trophic positions of dragonfly larvae in ambient, warmed conditions, and those exposed to heat waves. Equation used to calculate trophic position is adapted from Post (2002): Trophic position = [(!15Nconsumer !15Nbase) / Δ15N] + "; where λ = 2, and a ∆δ15N = 2.0‰. Range indicates minimum and maximum values calculated for each species in each treatment…………………………………………………………….94

List of Figures

Figure 1.1. Photographs of odonate species used in this thesis in their adult form. (a) Male common whitetail, (b) female eastern pondhawk, (c) male boreal bluet, (d) male autumn meadowhawk, (e) male dot-tailed whiteface, (f) female blue dasher. Photos used under Creative Commons Attribution-Share Alike 2.0 Generic license. Credits for specific photos: (a) CC BY Bruce Marlin; (b) CC BY M. Nazelrod; (c) CC BY Mike Ostrowski; (d,f) CC BY Judy Gallagher; (e) CC BY Gary Yankech...... 10

Figure 2.1 Effects of temperature on egg development rate in four odonate species: Celithemis elisa (squares); Leucorrhinia intacta (circles); Libellula luctuosa (triangles); Libellula pulchella (diamonds). Temperature means of these treatments were 22.2, 24.9, 27.0, and 30.7° C. Error bars are ±1 SE. Points are jittered on the x-axis. n = 20, 21, 21, and 20 egg clutches for each temperature, respectively...... 29

Figure 2.2 The effect of temperature on Le. intacta hatchling head width, mm (a) and growth rate, mm per day (b) in constant (23.4, 27.5, 30.4 °C) and variable (27.4 °C variable) temperatures. n = 40, 36, 38, and 39 larvae, respectively...... 31

Figure 2.3 Kaplan–Meier survival curves for the mean proportion of surviving Le. intacta larvae in constant (23.4, 27.5, 30.4 °C) and variable (27.4 °C – variable) temperature treatments over time (days) since hatching. Different lines represent treatment temperatures...... 32

Figure 3.1. The relationships between temperature and (a) the number of mosquito prey eaten per hour and (b) the total number of moves by a larva for each of the dragonfly species. Points represent means ± 1 SE. Species codes are the first two letters of the genus and species. Sample size varied by species (Erythemis simplicicollis: n = 20, Plathemis lydia: n = 22, Libellula pulchella: n = 32, Sympetrum vicinum: n = 45, Leucorrhinia intacta: n = 32, and Epitheca canis: n = 20) ...... 55

Figure 3.2. The relationship between species’ maximum realized temperatures and their foraging responses to temperature. Foraging response is measured as the change in number of prey eaten per hour from 28º to 32º C. Maximum realized temperatures are median maximum temperatures of the warmest month extracted from point locations across species’ ranges. Species codes are the first two letters of the genus and species ...... 56

Figure 3.3. The proportion of dragonfly larvae killed during IGP trials performed at three temperatures (23º, 28º and 32º C). Initial starting abundance was 8 larvae of each species per tank (5 tanks replicated at each temperature). Each IGP trial lasted for 72-hours. Points are jittered on the x-axis. Points represent means ± 1 SE. Species codes are the first two letters of the genus and species ...... 58

Figure 3.4. The effect of IGP on mean initial (point on left) and final (right) head widths of dragonfly larvae before and after trials at different temperatures. Stars denote significant differences between initial and final head widths (** = P < 0.001; * = P < 0.01; • = P < 0.10). Points are jittered on the x-axis. Points represent means ± 1 SE. Species codes are the first two letters of the genus and species ...... 59

Figure 4.1. Bar plots displaying the effects of temperature and predation risk on Enallagma boreale damselfly larvae total number of feeding attempts (a), and the proportion of total feeding attempts that were successful (b). Blue and red colours represent control (21º) and warmed (27ºC) treatments, respectively. No predator treatments indicate no predation risk (empty cage) and Predator treatments indicate caged (non-consumptive) Anax predator was present. Bars display means + 1 SE ...... 76

Figure 4.2. Bar plots illustrating the effects of temperature and predation risk on larval Enallagma damselfly conspicuous activities. Temperature increased frequency of turning (a) and crawling/stalking (b). The combined effect of predation risk and warming significantly increased distance moved (c), measured as the number of squares the damselfly moved through during the assay. Blue and red colours represent control (21º) and warmed (27ºC) treatments, respectively. No predator treatments indicate no predation risk (empty cage) and Predator treatments indicate caged (non-consumptive) Anax predator was present. Bars display means + 1 SE ...... 77

Figure 5.1 Daily average temperatures in the mesocosms across the 26-week developmental period. Colours represent different treatments conditions. A = ambient (natural diurnal fluctuations); W = warmed (+3º C above ambient); V = heat waves (+1 and + 5º C above ambient, fluctuating weekly)...... 95

Figure 5.2 The proportion of surviving dragonfly larvae raised in different temperature conditions. Treatments are: A = ambient (natural diurnal fluctuations); W = warmed (+3º C above ambient); V = heat waves (+1 and + 5º C above ambient, fluctuating weekly). Points are means ± SE and different shapes/colours represent different species ...... 96

Figure 5.3 The average proportional abundance of dragonfly larvae at the start of the experiment (initial) and after development in different temperature treatments W = warmed (+3º C above ambient); V = heat waves (+1 and + 5º C above ambient, fluctuating weekly). Initial abundances for each mesocosm were n = 20 larvae per species...... 97

Figure 5.4 Mean growth rate (mg × days-1) of dragonfly larvae developing in different temperature conditions. Growth rate was calculated as change in estimated individual dry mass from beginning to end of experiment over the development period (183 days). Treatments are: W = warmed (+3º C above ambient); V = heat waves (+1 and + 5º C above ambient, fluctuating weekly). Points are means ± SE and different shapes represent different species. *Note: L. luctuosa and Pl. lydia growth rates in ambient are from single individuals...... 98

xii

List of Appendices

Appendix I: Supplementary material for chapter 2 131

Appendix II: Supplementary material for chapter 3 136

Appendix III: Supplementary material for chapter 4 142

Appendix IV: Supplementary material for chapter 5 146

xiii

General Introduction 1 Introduction

Ecological communities are shaped by species’ responses to the complex interplay between abiotic and biotic forces acting at varying spatial and temporal scales. Early ecologists first observed the effects of these forces by relating patterns of species distributions across different environmental gradients (Andrewartha & Birch, 1954; MacArthur, 1958; Hutchinson, 1959; Fischer, 1960). Foundational theories were developed to explain these patterns and tested using experiments designed to disentangle the role of often co-varying factors on the distribution of species and community structure (Connell, 1961; Paine, 1966). Now, in an era of rapid environmental change, ecologists are building upon these theories to develop unifying frameworks to better predict how communities will be shaped by future climate change (e.g., Gilman et al., 2010; Webb et al., 2010; Niehaus et al., 2012; Gilbert et al., 2014; Sinclair et al., 2016). In my thesis, I use experiments to test some of the predictions of these frameworks to understand how climate change affects responses of organisms, and ultimately, their interactions and community structure.

By performing experiments that test theoretical predictions of individual and community responses to warming, I integrate concepts across several subfields in ecology to provide a comprehensive perspective of how species and their communities are expected to respond to future climate change. In the next sections, I provide a general review of the common responses to current climate change and how my dissertation research addresses gaps in those fields. I also describe the life history and documented effects of warming on my study organisms, and finally, outline the goals of my data chapters (chapters 2-5).

Present and future climate change

The Intergovernmental Panel on Climate Change’s Fifth Assessment Report concludes that global (land and ocean surface) temperatures have increased by approximately 0.85º C since 1880. The period of 1980 to 2012 was likely the warmest 30-year period of the last 1400 years, and the rate of temperature increase since this time is much faster than it has been in over 1000 years (IPCC, 2014). Varying climatic processes occurring in different regions, such as El Niño

1 2 events, have led to a large degree of regional variation in the magnitude of temperature change and this pattern will continue, with some regions warming more quickly than others. For instance, Canada and other high latitude regions have experienced temperature increases almost double that of the global mean (Canada: + 1.6º C, global: + 0.7º C, since 1948; Environment and Climate Change Canada, 2016). Temperatures are expected to continue increasing, with global projections of 1.0 – 3.7º C increases by 2100 (IPCC, 2014). In addition to warming, an increase in the number of extreme weather and climate events such as heat waves, heavy precipitation, drought, and coastal flooding have been observed since the 1950’s (IPCC, 2014). Periods of sustained high temperatures, i.e., heat waves, are predicted to become more severe and last longer in the future (Meehl, 2004).

1.1.1 Species responses to climate change

In the last century, organisms, populations, and communities globally have experienced rapidly changing environmental conditions due to climate change. Climate change poses a large threat to biodiversity (Ceballos et al., 2015) and is expected to accelerate the loss of species with an estimated one in six species going extinct as temperatures continue to rise (Urban, 2015). While these predicted losses are alarming, some species may be able to persist by responding in one or more of the following ways: species can shift in time (phenological shift) which reduces their exposure to unsuitable climate; species can shift in space (range shift) to track their preferred climatic niche; and some species respond plastically or adaptively in ways that allow them to tolerate different climatic conditions (Hughes, 2000; Walther et al., 2002; Charmantier et al., 2008). While there is considerable variation in the plasticity of species to respond in these ways, researchers have identified some of the most common documented responses to recent climate change. Those responses are: (i) earlier timing of spring events, (ii) range shifts upward in elevation and latitude (Parmesan & Yohe, 2003), and (iii) decreasing size at maturity (Daufresne et al., 2009). These responses have primarily been identified with thorough analysis of historical datasets. However, historical data alone are limited in deciphering the mechanisms through which climate change will act. Our ability to create useful predictive frameworks hinges on understanding the mechanisms through which climate change will act. Experiments are therefore needed that can improve existing theoretical frameworks by rigorously testing their hypotheses and in tandem with long-term data sets, will provide the best knowledge base for predicting how species, communities, and ecosystems will be affected by future warming.

1.1.2 Phenological shifts

Species’ shifts occurring in both time (phenology) and space (range) were first synthesized in the early 2000’s (Hughes, 2000; Walther et al., 2002; Parmesan & Yohe, 2003). These comprehensive meta-analyses found that the majority of documented phenological events such as time of breeding, nesting, first flower, tree budburst, and insect emergence had occurred earlier in the spring than historically recorded. Since these first meta-analyses, many more studies have found the same pattern: springtime events are occurring earlier (Menzel et al., 2006; Schwartz et al., 2006). The change in timing of phenological events is thought to be the result of species either responding directly to temperature (Polgar & Primack, 2011), or indirectly responding to earlier resource availability (Réale et al., 2003).

While there is a strong shift towards earlier spring phenological events, there is considerable variation in the magnitude and even direction of this response among species (Parmesan, 2007). Idiosyncrasies in sensitivity to temperature changes or other seasonal cues can lead to de- coupling or mismatch in the timing that species would normally interact (Both et al., 2009). This can be problematic for species with tight mutualisms with other organisms (Rafferty et al., 2015), however, organisms that reduce competition via seasonal temporal segregation may also be affected by unequal phenological responses to warming if rising temperatures leads to increased overlap between competing species (CaraDonna et al., 2014). Additionally, species that significantly advance springtime arrival with rising temperature may gain an advantage by being the first to colonize a patch or establish territory (Gilman et al., 2010). Conversely, early emergence can be risky, as this often places organisms into environments where they can experience rapid abiotic changes (e.g., frost, Inouye, 2008). With a clearer understanding of the traits underlying species’ sensitivities to rising temperatures, we can better predict how species interactions may be affected by future climate change.

1.1.3 Range shifts

In the search for coherent signatures of climate change, Parmesan and Yohe (2003) identified significant range shifts in response to climate warming. The most commonly identified range shifts are species’ range expansions, especially towards the poles. However, they also identified that some species' ranges may be contracting, and that some species are moving in the opposite direction than predicted. Interestingly, a more recent meta-analysis found evidence that range

4 shifts are occurring at a more rapid pace than previously reported (Chen et al., 2011). Additionally, there is evidence that range shifts are occurring in nearly all taxonomic groups and at all spatial scales (e.g., microclimates to continental Hickling et al., 2006). The movement of species in response to climate change will have large impacts on communities (Urban et al., 2012). For instance, species invasions of new habitats can create novel interactions that can ultimately affect the structure of communities (Gilman et al., 2010). Understanding how the strength and type of interaction are affected by climate change via differential species’ range shifts will be critical for predicting how communities will be affected by future warming.

1.1.4 Shifts in body size – The temperature-size rule

In 1847, based on observations of intraspecific variation in body size across latitude, Carl Bergmann proposed that selection favors larger body size in endotherms that live at high latitudes because a larger size can better conserve heat in these colder climates, i.e., smaller ratio of surface area to body mass reduces heat loss. This reasoning, however, was insufficient to explain this pattern in ectotherms, which are dependent on their environment to determine body temperature. In 1994, David Atkinson summarized what was known at the time for the relationship between ectotherm body size and temperature. His meta-analysis found support for previous theories that ectotherm developmental temperatures are important in determining final body size, with over 80% of cases observing smaller size at maturity when developmental conditions are warmer – a phenomenon he deemed the temperature-size rule (Atkinson, 1994).

Since Atkinson’s 1994 publication, there have been several mechanisms proposed to explain the temperature-size rule (TSR). Proximately, the TSR may simply be a consequence of physiological and biokinetic constraints (Atkinson & Sibly, 1997). Rates of cellular differentiation and growth are temperature-dependent, yet importantly, these rates can vary based on differences in the temperature sensitivity of rate-limiting steps, and ultimately lead to faster maturation at smaller size in warmer conditions (van der Have & de Jong, 1996). There is also evidence that adaptive plasticity of body size to temperature could be driving observed clines in body size. For instance, warm temperatures can favour early maturation, especially because delayed maturation can be costly in warm environments (reviewed by Angilletta et al., 2004). Another potential mechanism of this adaptive plasticity suggests a critical factor is the reduced availability of oxygen in warmer conditions (Portner, 2002), especially in aquatic systems

(Forster et al., 2012). Oxygen availability and therefore temperature, would constrain cell size because it takes longer for oxygen to diffuse and penetrate large cells with large volumes relative to their surface areas (Woods, 1999).

The TSR has become particularly relevant in recent decades, as researchers have found substantial evidence that recent climate warming has led to significant reductions in adult body size of many ectothermic organisms (reviewed by Gardner et al., 2011; Sheridan & Bickford, 2011). Body size often plays a crucial role in demographic processes (Peters, 1986), and therefore warming-induced reductions may affect important traits such as dispersal capability, survival, and fecundity (Post et al., 1997; McCauley & Mabry, 2011; Benard, 2015). More recently, a review of climate-driven changes in body size found that observed decreases in body size are not only determined by physiological processes occurring within individuals, but can also be mediated by interactions with other species (Ohlberger, 2013). Especially in systems that are strongly shaped by size-structured predation, such as aquatic communities, differences in the plasticity of this response among species could severely change the strength and direction of these interactions and presumably, the composition of these communities. Therefore, understanding how differential developmental responses to temperature, body size, and their effects on size-structured interactions are affected by warming will be critical for predicting community-level impacts of climate change.

Study system

In my thesis, I worked with several odonate species (dragonflies and damselflies) in their larval form. I chose to use larval odonates because: (i) they are voracious predators and therefore important in structuring aquatic communities; (ii) they are ectothermic and therefore respond to environmental warming; and (iii) because their regional distributions across aquatic habitats are well known. Ancestors of modern odonates arose during the Carboniferous around 325 Mya. This period is characterized by extremely high levels of atmospheric oxygen, a tropical climate, and globally high temperatures. Subsequent periods of global cooling and warming led to range contractions and expansions away from the equator (Corser et al., 2014). Modern temperate species arose from groups evolving in both tropical and temperate climates. It is thought that their biphasic life history allowed these thermophilic organisms to colonize and persist in temperate environments with more extreme thermal variation than the tropics (Kalkman et al.,

2008). All modern odonates develop in aquatic habitats with a larval period typically lasting 3-11 months for univoltine species in temperate climates. The species I used in my dissertation all occur in temperate latitudes but vary in their distributions across the continent (Table 1.1; figure 1.1).

1.2.1 Odonate responses to climate change

Odonates have shifted in multiple dimensions in response to recent climate change. For instance, they have shifted their ranges (Hickling et al., 2005) and their phenology (Hassall et al., 2007) in the last several decades. As such, Hassall (2015) identified them as important indicators or “barometers” of climate change. In both studies, Hickling et al., (2005) and Hassall et al. (2007), surveyed odonates in Britain and identified that the majority of species showed patterns similar to the global responses to climate change: poleward range shifts and advancing spring phenologies. However, these studies also identified significant variation in species’ responses. For instance, Hassall et al. (2007) found that species with a larval diapause stage (overwinter as larvae) exhibited greater advances in emergence phenology than those with an egg diapause stage, potentially due to the differences in temperature and photoperiod cues used by larvae (Corbet, 1999; Hassall et al., 2007). In my thesis, I determined if differences among species in the temperature-sensitivity of egg development affected their order of hatching, and thus the potential for differential phenological responses to affect species interactions. While I did not explicitly examine the effects of climate change on range shifts on odonates, the implications of such shifts are critical to understanding the other ways in which this insect order will be affected by future climate change.

There is currently mixed evidence that demonstrate odonates as a group display the third "universal" response to climate change, decreasing body size at maturity. An early field survey examining the importance of warming on adult odonate body size was by Cothran and Thorpe (1982). The researchers measured exuvial size (length of the promentum) of multiple dragonfly species along a thermal effluent gradient and found that eight of the nine most abundant species were smaller as adults when emerging from warmer water (four of which were significantly smaller, Cothran and Thorp, 1982). In the damselfly, Lestes viridis, which completes larval development faster than the dragonflies above (< 3 months), experimental increases in temperature significantly reduced size at emergence (De Block & Stoks, 2003). However, recent

7 experimental evidence suggests that larval dragonflies reared in artificially warmed pond environments do not exhibit this pattern of smaller size when raised in warmer conditions (McCauley et al., 2015, 2018). Instead individuals of both species emerged earlier (5-6 days/ 1º C increase), but as the same size as individuals raised in ambient conditions. Conflicting findings, potentially due to a lack of common garden and standardized sampling methods complicates the effects of temperature on odonate size at maturity but also could suggest that indirect processes may be important in shaping odonate size at maturity (e.g. increasing size- biased IGP, Wellborn et al., 1996). Given that body size is an important predictor of fitness and dispersal ability, both factors that will affect the ability of species to persist under future climate scenarios, it is imperative that this paradox is resolved. I examine the importance of body size in structuring species interactions during the larval stage, in particular, how growth and developmental responses to temperature interact to affect size-structured interactions within these communities. I explore these questions in chapters 2, 3, and 5.

Several recent studies have documented patterns of community change in odonates through time and across spatial and climatic gradients. Ball-Damerow et al. (2014) re-sampled freshwater sites in California and Nevada of those initially sampled by C. H. Kennedy in 1914 and 1915. While odonate species richness had not changed considerably over the century, they observed an increase in community similarity across sites (Ball-Damerow et al., 2014a). They attributed this homogenization to a few factors including increasing occurrences of highly labile species, a decrease in specialist species, and a decrease in the number of species that have an overwintering diapause stage. Their results are intriguing as they indicate that changes in community composition may be driven by species-specific differences in tolerance for changing conditions. In other studies, researchers have found strong relationships between odonate species assemblages and climatic variables such as growing degree days, summer temperature, and precipitation (Ball-Damerow et al., 2014b; Boersma et al., 2016). Altogether, current data indicate that odonate communities are strongly affected by temperature and that future climate change may play a significant role in restructuring their communities. I investigate the potential community consequences of future warming in chapters, 3,4 and 5.

1.2.2 Odonates as structuring agents of communities

Intraguild predation (IGP), or the killing and consuming of potential competitors (Polis et al., 1989), is widespread across taxonomic groups in different habitats (Arim & Marquet, 2004). Intraguild predation, which also includes cannibalism, can reduce competition for shared resources through the act of killing and removing a competitor. In this way, IGP is an extreme form of interference competition (Polis, 1981). In odonate larvae, the frequency of cannibalism/IGP is positively related to the magnitude of size differences among individuals (Van Buskirk, 1989; Wissinger, 1988; Anholt, 1994). These body size differences can arise between individuals due to developmental asynchrony within generations, i.e., multiple cohorts of eggs are oviposited by adults females at varying times during the summer (Wissinger, 1988b). Thus, the timing of breeding phenology in adult odonates is important in determining the strength and direction of IGP (Padeffke & Suhling, 2003). Similarly, the rate of egg development, which is strongly influenced by temperature, may also determine whether hatching larvae are more likely to be IG predators or IG prey through priority effects (Snyder & Hurd, 1995). Any interspecific differences in the temperature-dependence of egg development rates could re-order relative hatching times among species, and thus, affect trophic dynamics between IG predators. I examine this hypothesis in Chapter 2.

In theory, IGP can be both a stabilizing and destabilizing force in trophic webs depending on the strength of the interaction (Wootton, 2017). IGP can increase stability within food webs, when those interactions are weak (McCann et al., 1998). However, IGP, like predation rate, may increase with rising temperatures associated with climate change due to increasing encounter rates and higher energetic demands. Increasing predation rates can lead to trophic cascades, which ultimately affects community structure (Hairston et al., 1960; Polis et al., 1996). However, increasing IGP could also lead to the loss of IG predator diversity, in turn dampening cascading effects (Finke & Denno, 2005). Understanding how IGP is affected by warming in guilds of predacious odonates will allow us to better predict how communities characterized by strong size-selective IGP will be affected by future climate change. I explore these ideas in chapters 3 and 4.

Table 1.1. Odonate species used in this thesis, taxonomic family, geographic distribution, and the chapters in which they are found. Species Family Distribution† Chapter

Leucorrhinia intacta Hagen Nova Scotia west to the Pacific coast. Libellulidae South to Kentucky, Colorado and mid- Chps 2,3 Dot-tailed whiteface California Celithemis elisa Hagen New Brunswick west to Minnesota. South Libellulidae Chp 2 Calico pennant through Texas Pachydiplax longipennis Maine west to Minnesota, Nebraska, and Burmeister Libellulidae Oklahoma. Pacific coast north to BC and Chp 5 Blue dasher south through California; west to Texas Erythemis collocata Hagen Pacific coast north to southern BC and Libellulidae Chp 5 Western pondhawk south to Baja California. East to Colorado

Erythemis simplicicollis Say New Brunswick west through Ontario, Libellulidae Nebraska, and Texas. South to Florida, Chp 3 Eastern pondhawk Cuba, and Belize

Sympetrum vicinum Hagen Nova Scotia west to southern Manitoba, Libellulidae Oklahoma, and Texas. Pacific coast north Chp 3 Autumn meadowhawk to southern BC, south to mid-California

Plathemis lydia Drury Nova Scotia west to southern Manitoba, Libellulidae Oklahoma, and Texas. Pacific coast north Chps 3,5 Common whitetail to southern BC, south through California

Libellula luctuosa Burmeister Nova Scotia west to Minnesota, Nebraska, Libellulidae and Texas. BC south through California; Chp 2,5 Widow skimmer Arizona and New Mexico Nova Scotia west to North Dakota, Libellula pulchella Drury Nebraska and Texas; south to South Libellulidae Ch 2,3 Twelve-spotted skimmer Carolina. Throughout California and Nevada

Epitheca canis McLachlan Nova Scotia west to southern Manitoba. Corduliidae South to Ohio. Central California to Ch 3 Beaverpond baskettail southern BC Anax junius Drury West to Pacific coast. Southern Aeshnidae Chp 4 Common green darner Saskatchewan south to central Mexico Enallagma boreale Selys Newfoundland west to Alaska. South to Coenagrionidae Chp 4 Boreal bluet Illinois †Species’ distribution data from the Global Biodiversity Information Facility (gbif.org)

Thesis overview

The objectives of my thesis are to assess how differences in species’ responses to warming affect interactions in communities and how these individual responses ultimately influence community structure and composition. I carried this out by conducting laboratory and mesocosm experiments with various larval odonate species to better understand how communities will be affected by future climate change. While climate change is an umbrella term that encapsulates multiple changes in abiotic conditions, I focus on the effects of warming and extreme temperature events. A more detailed description of each chapter is present below.

1.3.1 Chapter 2

Differential responses to warming among species may affect competitive or predatory interactions among species, however the degree to which ecologically similar species differ in their plasticity to respond to warming is not well known. In addition, the degree to which individuals within a population vary in the plasticity of responses to warming may be crucial for determining the species’ resilience to future warming yet is often over-looked. I examined how warming affected interspecific variation in egg development rate to assess whether differential responses could lead to changes in the relative timing of hatching. In addition, I asked how warming and periodic heat waves could affect performance and growth at early life history stages in a single species. Warming resulted in faster egg development rates, yet there were no differences in the strength of this response among species. Warming also led to smaller body size at hatching, faster growth, and higher mortality rates with high levels of intraspecific variation.

This study was carried out in collaboration with Dr. Shannon McCauley and Jy Moon, a Research Opportunity Program undergraduate student whom I mentored. Shannon contributed to the ideas and experimental design. Jy and I performed one of the studies presented in this chapter together; I performed the other study and collected the data. I analyzed all the data and wrote the manuscript and Shannon provided edits. This paper is published in the Canadian Journal of Zoology and has been included in this thesis with permission from the publisher.

Frances, D.N., Moon, J., & McCauley, S.J. (2017). Effects of environmental warming during early life-history on libellulid odonates. Canadian Journal of Zoology, 95, 373–382.

1.3.2 Chapter 3

Few studies consider species-specific behavioral responses to warming when making predictions about how species interactions will be affected by warming. I first determined if species had different behavioural responses to warming and asked if these behavioral responses could accurately predict the outcome of intraguild predation interactions in warmer conditions. From a macro-physiological perspective, I also examined if species’ behavioural responses to temperature were correlated to their geographic range positions, with the assumption that geographic range limits are correlated with species’ physiological tolerance parameters, and therefore their ability to tolerate future warming. Warming increased feeding rates by 42% on average across species but had no effect on activity rate. The magnitude of change in feeding rate was positively correlated with the maximum temperatures species experience across their ranges. Lastly, warming increased rates of IGP twofold, however, species’ behavioral responses alone were not predictive of their susceptibility to become IG prey of other larvae at warmer temperatures.

This chapter is in collaboration with Shannon McCauley. Shannon and I thought of the idea and designed the experiments. I performed the experiments, collected and analyzed the data, and wrote the manuscript and Shannon provided edits. This paper is published in Oecologia and has been included in this thesis with permission from the publisher.

Frances, D.N. & McCauley, S.J. (2018). Warming drives higher rates of prey consumption and increases rates of intraguild predation. Oecologia. doi: 10.1007/s00442-018-4146-y

1.3.3 Chapter 4

While the non-consumptive effects of predators have been well documented, relatively little is known about how the threat of predation interacts with warming to affect behavioural decisions of prey. This is particularly important because warming can induce opposing effects on activity than predation risk, in which prey generally suppress activity to reduce their exposure to predators. I asked how the combination of warming and predation risk affect behavioural response of prey using larval damselflies. Prey generally responded to temperature more strongly than to predation risk by increasing activity, even though their risk of being eaten increased at warmer temperatures. Interestingly, warming and predation risk had cascading effects on

13 damselfly prey, in that damselflies became more accurate and efficient at feeding on zooplankton when predation threat and temperatures were high.

This chapter is in collaboration with Shannon McCauley. Shannon contributed to the conceptual design of this experiment. I performed the experiments, collected and analyzed the data, wrote the manuscript, and Shannon provided edits. This chapter is in preparation to be sent to Biology Letters.

1.3.4 Chapter 5

The effects of future warming and heat waves on community structure and composition have been little explored, especially in shallow aquatic habitats. Using pond mesocosms, I simulated realistic future warming heat wave conditions and asked how aquatic invertebrate communities are affected by these developmental conditions. Warming heat waves had similar effects on species’ evenness in communities, in which all four odonate species were able to coexist. Heat waves increased larval survival to a greater degree than warming alone and the primary consumers in these conditions were suppressed in terms of abundance, potentially as a result of strengthened trophic cascades in these communities.

This chapter is in collaboration with Drs. Shannon McCauley, Karen Mabry (New Mexico State University), and John Hammond (Marian University) who were involved in establishing the research infrastructure and conceptual underpinnings of this work; Shannon and I designed the experiment; I collected the data, performed the analyses, and wrote the manuscript. Shannon, Karen, and John provided edits for the manuscript. Chapter 5 is in preparation to be sent to Global Change Biology.

References Andrewartha HG, Birch LC (1954) The Distribution and Abundance of Animals. University of Chicago Press, Chicago. Angilletta MJ, Steury TD, Sears MW (2004) Temperature, growth, and body size in ectotherms: fitting pieces of a life-history puzzle. Integrative and Comparative Biology, 44, 498–509. Anholt BR (1994) Cannibalism and early instar survival in a larval damselfly. Oecologia, 99, 60–65. Arim M, Marquet PA (2004) Intraguild predation: a widespread interaction related to species biology. Ecology Letters, 7, 557–564.

Atkinson D (1994) Temperature and organism size-a biological law for ectotherms? Advances in Ecological Research, 25, 1–58. Atkinson D, Sibly RM (1997) Why are organisms usually bigger in colder environments? Making sense of a life history puzzle. Trends in Ecology & Evolution, 12, 235–239. Ball-Damerow JE, M’Gonigle LK, Resh VH (2014a) Changes in occurrence, richness, and biological traits of dragonflies and damselflies (Odonata) in California and Nevada over the past century. Biodiversity and Conservation, 23, 2107–2126. Ball-Damerow JE, M’Gonigle LK, Resh VH (2014b) Local and regional factors influencing assemblages of dragonflies and damselflies (Odonata) in California and Nevada. Journal of Insect Conservation, 18, 1027–1036. Benard MF (2015) Warmer winters reduce frog fecundity and shift breeding phenology, which consequently alters larval development and metamorphic timing. Global Change Biology, 21, 1058–1065. Boersma KS, Nickerson A, Francis CD, Siepielski AM (2016) Climate extremes are associated with invertebrate taxonomic and functional composition in mountain lakes. Ecology and Evolution, 6, 8094–8106. Both C, Van Asch M, Bijlsma RG, Van Den Burg AB, Visser ME (2009) Climate change and unequal phenological changes across four trophic levels: Constraints or adaptations? Journal of Animal Ecology, 78, 73–83. Van Buskirk J (1989) Density-dependent cannibalism in larval dragonflies. Ecology, 70, 1442– 1449. CaraDonna PJ, Iler AM, Inouye DW (2014) Shifts in flowering phenology reshape a subalpine plant community. PNAS, 111, 4916–4921. Ceballos G, Erlich PR, Barnosky AD, Garcia A, Pringle RM, Palmer TM (2015) Accelerated modern human – induced species losses: entering the sixth mass extinction. Sciences Advances, 1, 1–5. Charmantier A, McCleery RH, Cole LR, Perrins C, Kruuk LEB, Sheldon BC (2008) Adaptive phenotypic plasticity in response to climate change in a wild bird population. Science, 320, 800–804. Chen I, Hill JK, Ohlemuller R, Roy DB, Thomas CD (2011) Rapid range shifts of species associated with high levels of climate warming. Science, 333, 1024–1026. Connell JH (1961) The influence of interspecific competition and other factors on the distribution of the barnacle Chthalamus stellatus. Ecology, 42, 710–723. Corbet PS (1999) Dragonflies: Behaviour and Ecology of Odonata, Revised Ed edn. Harley Books, Colchester, Essex. Corser JD, White EL, Schlesinger MD (2014) Odonata origins, biogeography, and diversification in an Eastern North American hotspot: Multiple pathways to high temperate forest insect diversity. Insect Conservation and Diversity, 7, 393–404. Cothran, M. L., & Thorp, J. H. (1982). Emergence patterns and size variation of Odonata in a thermal reservoir. Freshwater Invertebrate Biology, 1(4), 30–39.

Daufresne M, Lengfellner K, Sommer U (2009) Global warming benefits the small in aquatic ecosystems. Proceedings of the National Academy of Sciences, 106, 12788–93. De Block, M., & Stoks, R. (2003). Adaptive sex-specific life history plasticity to temperature and photoperiod in a damselfly. Journal of Evolutionary Biology, 16(5), 986–995. Environment and Climate Change Canada (2016) Climate data and scenarios for Canada : Synthesis of recent observation and modelling results. Gatineau, QC, 1-28 pp. Finke DL, Denno RF (2005) Predator diversity and the functioning of ecosystems: The role of intraguild predation in dampening trophic cascades. Ecology Letters, 8, 1299–1306. Fischer AG (1960) Latitudinal variations in organic diversity. Evolution, 14, 64–81. Forster J, Hirst AG, Atkinson D (2012) Warming-induced reductions in body size are greater in aquatic than terrestrial species. Proceedings of the National Academy of Sciences of the United States of America, 109, 19310–4. Gardner JL, Peters A, Kearney MR, Joseph L, Heinsohn R (2011) Declining body size: A third universal response to warming? Trends in Ecology and Evolution, 26, 285–291. Gilbert B, Tunney TD, Mccann KS et al. (2014) A bioenergetic framework for the temperature dependence of trophic interactions. Ecology Letters, 17, 902–914. Gilman SE, Urban MC, Tewksbury J, Gilchrist GW, Holt RD (2010) A framework for community interactions under climate change. Trends in Ecology and Evolution, 25, 325–331. Hairston, N. G., Smith, F. E., & Slobodkin, L. B. (1960). Community structure, population and competition control. The American Naturalist, 879, 421–425. Hassall C, Thompson DJ, French GC, Harvey IF (2007) Historical changes in the phenology of British Odonata are related to climate. Global Change Biology, 13, 933–941. van der Have TM, de Jong G (1996) Adult size in ectotherms: temperature effects on growth and differentiation. Journal of Theoretical Biology, 183, 329–340. Hickling R, Roy DB, Hill JK, Thomas CD (2005) A northward shift of range margins in British Odonata. Global Change Biology, 11, 502–506. Hickling R, Roy DB, Hill JK, Fox R, Thomas CD (2006) The distributions of a wide range of taxonomic groups are expanding polewards. Global Change Biology, 12, 450–455. Hughes L (2000) Biological consequences of global warming: is the signal already apparent? Trends in Ecology & Evolution, 15, 56–61. Hutchinson GE (1959) Homage to Santa Rosalia or why are there so many kinds of anaimals? The American Naturalist, 93, 145–159. Inouye DW (2008) Effects of climate change on phenology frost damage, and floral abundance of montane wildlife. Ecology, 89, 353–362. IPCC (2014) Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. In: Core Writing Team, Pachauri RK, Meyer LA (eds) IPCC, Geneva, Switzerland, 151 p., 1–112.

Kalkman VJ, Clausnitzer V, Dijkstra KDB, Orr AG, Paulson DR, Van Tol J (2008) Global diversity of dragonflies (Odonata) in freshwater. Hydrobiologia, 595, 351–363. Kratina P, Greig HS, Thompson PL, Carvalho-Pereira TSA, Shurin JB (2012) Warming modifies trophic cascades and eutrophication in experimental freshwater communities. Ecology, 93, 1421–1430. MacArthur RH (1958) Population ecology of some warblers of northeastern coniferous forests. Ecology, 39, 599–619. McCann K, Hastings A, Huxel GR (1998) Weak trophic interactions and the balance of nature. Nature, 395, 794–798. McCauley SJ, Mabry KE (2011) Climate change, body size, and phenotype dependent dispersal. Trends in Ecology and Evolution, 26, 554–555. McCauley SJ, Hammond JI, Frances DN, Mabry KE (2015) Effects of experimental warming on survival, phenology, and morphology of an aquatic insect (Odonata). Ecological Entomology, 40, 211–220. McCauley SJ, Hammond JI, Mabry KE (2018) Simulated climate change increases larval mortality, alters phenology, and affects flight morphology of a dragonfly. Ecosphere, 9, e02151. Meehl GA (2004) More intense, more frequent, and longer lasting heat waves in the 21st century. Science, 305, 994–997. Menzel A, Sparks T, Estrella N et al. (2006) European phenological response to climate change matches the pattern. Global Change Biology, 12, 1969–1976. Niehaus AC, Angilletta MJ, Sears MW, Franklin CE, Wilson RS (2012) Predicting the physiological performance of ectotherms in fluctuating thermal environments. Journal of Experimental Biology, 215, 694–701. Ohlberger J (2013) Climate warming and ectotherm body size - from individual physiology to community ecology. Functional Ecology, 27, 991–1001. Padeffke T, Suhling F (2003) Temporal priority and intra-guild predation in temporary waters: An experimental study using Namibian desert dragonflies. Ecological Entomology, 28, 340–347. Paine RT (1966) Food web complexity and species diversity. The American Naturalist, 100, 65– 75. Parmesan C (2007) Influences of species, latitudes and methodologies on estimates of phenological response to global warming. Global Change Biology, 13, 1860–1872. Parmesan C, Yohe G (2003) A globally coherent fingerprint of climate change impacts across natural systems. Nature, 421, 37–42. Peters RH (1986) The Ecological Implications of Body Size. Cambridge University Press, 329 pp. Polgar CA, Primack RB (2011) Leaf-out phenology of temperate woody plants: From trees to ecosystems. New Phytologist, 191, 926–941.

Polis, G. A., & Strong, D. R. (1996). Food web complexity and community dynamics. The American Naturalist, 147(5), 813–846. Polis GA (1981) The evolution and dynamics of intraspecific predation. Annual Review of Ecology and Systematics, 12, 225–251. Polis GA, Myers CA, Holt RD (1989) The ecology and evolution of intraguild predation: potential competitors that eat each other. Annual Review of Ecology and Systematics, 20, 297–330. Portner HO (2002) Climate variations and the physiological basis of temperature dependent biogeography: Systemic to molecular hierarchy of thermal tolerance in animals. Comparative Biochemistry and Physiology - A Molecular and Integrative Physiology, 132, 739–761. Post E, Stenseth NC, Langvatn R, Fromentin J-M (1997) Global climate change and phenotypic variation among red deer cohorts. Proceedings of the Royal Society of London B: Biological Sciences, 264, 1317–1324. Rafferty NE, CaraDonna PJ, Bronstein JL (2015) Phenological shifts and the fate of mutualisms. Oikos, 1, 14–21. Réale D, McAdam AG, Boutin S, Berteaux D (2003) Genetic and plastic responses of a northern mammal to climate change. Proceedings of the Royal Society B: Biological Sciences, 270, 591–596. Schwartz MD, Ahas R, Aasa A (2006) Onset of spring starting earlier across the Northern Hemisphere. Global Change Biology, 12, 343–351. Sheridan JA, Bickford D (2011) Shrinking body size as an ecological response to climate change. Nature Climate Change, 1, 401–406. Sinclair BJ, Marshall KE, Sewell MA et al. (2016) Can we predict ectotherm responses to climate change using thermal performance curves and body temperatures? Ecology Letters, 1372–1385. Snyder WE, Hurd LE (1995) Egg-hatch phenology and intraguild predation between two mantid species. Oecologia, 104, 496–500. Urban MC (2015) Accelerating extinction risk from climate change. Science, 348, 571–574. Urban MC, Tewksbury JJ, Sheldon KS (2012) On a collision course: competition and dispersal differences create no-analogue communities and cause extinctions during climate change. Proceedings of the Royal Society B: Biological Sciences, 279, 2072–2080. Walther GR, Post E, Convey P et al. (2002) Ecological responses to recent climate change. Nature, 416, 389–395. Webb CT, Hoeting JA, Ames GM, Pyne MI, LeRoy Poff N (2010) A structured and dynamic framework to advance traits-based theory and prediction in ecology. Ecology Letters, 13, 267–283. Wellborn, G. A., Skelly, D. K., & Werner, E. E. (1996). Mechanisms creating community structure across a freshwater habitat gradient. Annual Review of Ecology and Systematics, 27(1996), 337–363.

Wissinger SA (1988a) Effects of food availability on larval development and inter-instar predation among larvae of Libellula lydia and Libellula luctuosa (Odonata: Anisoptera). Canadian Journal of Zoology, 66, 543–549. Wissinger SA (1988b) Life history and size structure of larval dragonfly populations. Journal of the North American Benthological Society, 7, 13–28. Woods HA (1999) Egg-mass size and cell size: effects of temperature on oxygen distribution. American Zoologist, 252, 244–252. Wootton KL (2017) Omnivory and stability in freshwater habitats: Does theory match reality? Freshwater Biology, 62, 821–832.

Effects of warming on early life history

Published as Frances, D. N., Moon, J., & McCauley, S. J. 2017. Effects of environmental warming during early life-history on libellulid odonates. Canadian Journal of Zoology, 95, 373– 382.

2 Abstract

Climate warming affects ectotherms globally, yet we know little regarding the variability in species’ responses to warming, particularly in early life stages. Additionally, intraspecific variation in response to warming is understudied but may determine species’ resilience to warming. To assess how temperature affects egg development rate in co-occurring dragonfly species, we manipulated temperature (range: 22–31° C) and measured time to hatching. Warming decreased egg development time across all species, indicating that while climate warming will advance hatching phenology, maintained synchrony in hatching order will likely not affect species interactions. Our second experiment examined early life-history responses to warming in the dot-tailed whiteface (Leucorrhinia intacta) dragonfly. We measured time to hatching, hatchling size, growth rate, and survival at four temperatures (23–30° C), including a treatment with increased thermal variation. Warming resulted in smaller hatchlings with increased growth and mortality rates, whereas higher thermal variation did not have effects different from those of warming alone. We observed significant intraspecific variation in the responses to warming in both egg development time and hatchling size and some of this variation was correlated with oviposition date. High levels of intraspecific variation may be important in buffering populations from the effects of climate warming.

Introduction

Increasing temperatures resulting from climate change have had large ecological impacts across a broad range of taxonomic groups. In response to warming, organisms are becoming smaller (Gardner et al. 2011, Sheridan and Bickford 2011), altering the timing of life history events (Parmesan 2006), and shifting their range margins (Sunday et al. 2012). Ectotherms are especially vulnerable to the effects of climate change (Chown et al. 2010) because their body temperature and metabolic processes are tightly linked to environmental temperature, therefore

20 warming directly influences ectotherm performance (Huey and Kingsolver 1989). Additionally, many ectotherms use abiotic cues such as changes in temperature to signal onset of life history events. For example, fish time their spawning (Genner et al. 2010) and bees break winter diapause (Forrest and Thomson 2011) in response to temperature signals. Because temperature has large effects on ectotherm life history and phenology, it is crucial to understand how these responses will be affected by a rapidly changing climate.

Shifts in phenology are one of the most commonly documented responses to climate change (Parmesan 2006). For example in freshwater systems, adult odonates (dragonflies and damselflies) are advancing the timing of their emergence from aquatic habitats earlier in the season (Hassall et al. 2007, Dingemanse and Kalkman 2008). Experimental warming of water temperatures has also been observed to result in earlier emergence in the libellulid dragonfly Pachydiplax longipennis (McCauley et al. 2015). Increasing temperatures can also hasten egg development rate, therefore advancing the timing of hatching (Howe 1967, Elliott 1978, Leggott and Pritchard 1985). While faster egg development in response to temperature is well documented, not all species respond in the same way and we still have little information about how much this response varies within and among species for many animals

Understanding how rates of egg development are affected by temperature and how these may respond to climate warming is critical because for many ectotherms the timing of egg hatching is a key phenological event. The timing of egg laying and hatching determines the abiotic conditions (Visser and Holleman 2001) and the biotic interactions of juveniles (Rudolf and Singh 2013). Larval odonates are voracious predators and the order and relative timing in which species hatch can determine body size advantages (Rasmussen et al. 2014), a crucial force structuring species interactions in aquatic systems (Werner and Gilliam 1984). Unequal responses to warming could affect the rank order of hatching within groups of co-occurring species, thereby contributing to these body size advantages or disadvantages (Guo et al. 2009). These size changes induced by temperature could alter interactions between co-occurring species and ultimately create “winners and losers” in response to climate change (Cahill et al. 2012). As a general response, odonates have faster egg development when reared in higher temperatures (Pritchard et al. 1996) yet we know little about the plasticity in egg development in response to temperature found within co-occurring species. We addressed this gap by experimentally

21 manipulating temperature and comparing the timing of egg hatching in four species of dragonfly (Odonata: Anisoptera) across temperatures from 22-31° C.

In ectotherms, smaller body size at maturity is a common effect of developing at higher temperatures because warmer conditions increase metabolic and developmental rates (Atkinson 1994, Sibly and Atkinson 1994). This effect is widespread with 75% of terrestrial and over 90% of aquatic ectotherms exhibiting this pattern (Atkinson 1995, Sheridan and Bickford 2011). Body size can affect dispersal ability (De Bie et al. 2012), range size (Rundle et al. 2007), strength of trophic interactions (Rudolf 2011) and fitness (Sokolovska et al. 2000, Kingsolver and Huey 2008). However, many of these studies have focused on these effects in later ontogeny, while early life history stages may be equally or even more sensitive to temperature (Klockmann et al. 2016). Odonate body size at hatching has been under-reported in the literature and therefore it remains unclear whether faster egg development will lead to smaller hatchling size in this group. Body size at hatching or other early stages of development can determine the size and type of prey that animals can consume as well as their vulnerability to predators. Additionally, juvenile body size has been directly correlated with adult fitness for amphibians (Semlitsch et al. 1988) and some insects (Carroll and Hoyt 1986). Body size post-hatching is smaller when incubation temperatures are warmer as shown in reptiles (Gutzke and Packard 1987, Van Damme et al. 1992), snails (Collin and Salazar 2010), beetles (Ernsting and Isaaks 1997), soil arthropods (Liefting et al. 2010) and butterflies (Fischer et al. 2003), however the opposite pattern has been observed for one damselfly species (Van Doorslaer and Stoks 2005b). It therefore remains unclear how hatchling size is affected by temperature in other odonates. We addressed this question in our study by rearing eggs of a single dragonfly species in different thermal environments and measuring size at hatching.

Temperature can also affect ectotherm survival. While climate change is expected to increase environmental temperatures, most of these temperatures will likely remain below upper lethal limits of many species (Li et al. 2013). Therefore, mortality as a direct result of surpassing critical thermal limits is unlikely (Rohr and Palmer 2013). However, higher temperatures can increase mortality risk directly or through interactions with stressors such as pollutants, disease or other abiotic environmental changes (Folt et al. 1999, Sokolova and Lannig 2008, Deschaseaux et al. 2010, Janssens and Stoks 2013, Cole et al. 2016). Additionally, in many studies survival data are censored, measuring survival only at the end of development. Therefore

22 it remains unclear at what stage higher temperatures are causing observed increases in mortality (Klockmann et al. 2016). In one dragonfly species, higher mortality occurred when larvae were reared in warmer temperatures, however survival was only measured at metamorphosis (McCauley et al. 2015). The timing of mortality in odonates exposed to warming will influence aquatic food webs; mortality early in development has very different effects on the role of these predators in aquatic systems than if most of this mortality is occurring at or near metamorphosis. Therefore, we wanted to know if odonates experienced high rates of mortality during early stages of development when raised in warmer temperatures. Understanding at what stage odonates are most vulnerable to the effects of warming can also help direct future odonate conservation efforts.

We examined the effects of temperature on early life-history stages in odonates, an important group of predatory freshwater ectotherms, using a multi-level approach combining inter- and intraspecific comparisons. For the interspecific comparisons, we assessed how temperature affects egg hatching phenology across four species in the same family that co-occur in lakes and ponds in Eastern North America. We asked: does hatching time among species vary with increasing temperature, thus changing the order of hatching? We predicted more rapid egg development and thus earlier hatching in response to warming, however, we expected the magnitude of these responses to vary among species and lead to changes in the rank order of hatching.

There is a growing body of research indicating that variation within species can affect important ecological dynamics (Bolnick et al. 2011) such as the ability to colonize new habitats (Dibble et al. 2014), community interactions (Duffy 2010), and population stability and persistence (Agashe 2009), especially with environmental perturbations (Oney et al. 2013). For our intraspecific comparisons, we measured the degree to which responses to temperature varied both within and among clutches of the odonate species, Leucorrhinia intacta (Hagen, 1861). Specifically, we quantified how temperature affected egg development time and size at hatching and then followed these larvae to assess the effects of temperature on rates of growth and survival during this critical early life phase. In addition to warming, we included one treatment with increased thermal variation because along with mean increases in temperature, increasing diel and seasonal fluctuations in temperature are expected to become more frequent with climate change (Easterling et al. 2000, Field 2012). A growing body of literature has examined ectotherm

23 responses to fluctuating as opposed to constant temperatures and have found changes in physiology, stress tolerance, life history traits and fitness (Colinet et al. 2015). We wanted to know: i. how are early life history stages affected by warming and by increased thermal variability? ii. How much do these responses vary within a single species? Based on previous studies, we predicted that warming would increase egg development rate, reduce body size at hatching and increase growth and mortality rates (Pritchard et al. 1996, Suhling et al. 2015).

Materials and methods

2.2.1 Study system

Animals used in this study were collected from the Koffler Scientific Reserve, KSR (King City, Ontario, 44° 1' 47.136"N, 79° 32' 0.4662"W). Experiments were conducted in the lab at University of Toronto, Mississauga (UTM). In the first study, we used four odonate species (Anisoptera: Libellulidae): 1. Celithemis elisa (Hagen, 1861), 2. Leucorrhinia intacta (Hagen, 1861), 3. Libellula luctuosa (Burmeister, 1839), and 4. Libellula pulchella (Drury, 1770). Larvae of these species commonly co-occur in freshwater ponds in the study region (Paulson 2011). We chose these species because they are relatively common and dominant members of odonate communities in this region. Additionally, libellulids are the most diverse odonate family in these lentic systems. We selected species from this family that occur across a range of habitats, such as those with differing predators and habitat permanence (McCauley 2008). All of these species are univoltine in this region with a larval period lasting approximately nine months (Corbet et al. 2006).

Eggs of these species are typically laid in the beginning of June and oviposition continues for varying durations across the summer (Paulson 2011). These species are exophytic: eggs are laid on the water surface in clumps and then sink to the bottom and attach to the surface of algae or aquatic vegetation (Walker 1953, Corbet 1999). Eggs develop in the littoral zone of freshwater ponds and lakes—a shallow region that tracks air temperatures closely (Schneider and Mauser 1996). Maximum air temperatures in the study region were 30.1º, 33.6º, and 32.8º C for June, July and August 2015, respectively (Environment Canada, retrieved 12 Dec 2016 from climate.weather.gc.ca). The 2050 climate prediction for this region is a + 2-2.5º C increase over the average current summer temperatures (IPCC 2012). Water has higher thermal conductivity than air, and therefore is more buffered from fine-scale variation in temperature, yet shallow

24 waters such as the littoral zone can still warm and cool mirroring maximum and minimum air temperature (Abrahams et al. 2007).

2.2.2 Experiment 1: Interspecific comparisons of egg development rate

For the first experiment, we collected eggs from five C. elisa, seven Le. intacta, four L. luctuosa and four L. pulchella adult females caught at KSR. Females were captured with aerial insect nets either while mating or while flying in the vicinity of the ponds. Eggs were collected by dipping the females’ abdomen into pond water-filled 120-mL plastic sample cups. After egg collection, we marked forewings of females with a permanent marker to avoid collecting from the same female more than once. Eggs were collected between 21 June and 22 July 2014.

We transported eggs in the water-filled sample cups to the lab at UTM on the same day as collection. They were kept cool and shaded during transit. We sorted eggs the following day after checking for signs of fertilization (eggs darken in color when fertilized). Using a split-brood design, we separated fertilized clutches of each female into four separate 350-mL plastic containers (AMAC © 2009 Plastic Products, Petaluma, California) with approximately 30 eggs per container. We randomly assigned egg containers to treatments and tanks so that eggs of each female were present in every temperature treatment. Treatment tanks were 20-liter glass aquaria. Each tank was filled with de-chlorinated, oxygenated water with submersible water heaters (Visi-Therm® Deluxe, Marineland Aquarium Products, Cincinnati, Ohio, USA) placed in every tank (for study design see Appendix I, Fig. S2.1). We had four temperature treatments (means ± SD): 22.2 ± 0.61º C, 24.9 ± 0.43 º C, 27.0 ± 0.85 º C, and 30.5 ± 1.40º C. There were five replicates of each temperature treatment for a total of 20 treatment tanks. These treatments created a range of thermal environments similar to natural conditions in this region as well as simulating warmer temperatures that eggs may experience in the future with climate change (Feltmate and Thistlethwaite 2012). The highest temperature treatment used is well below the lethal limits recorded for odonates (Garten and Gentry 1976, Dallas and Rivers-Moore 2011). Photoperiod was set to 15L:9D to simulate day length in June in the study region. Data loggers (HOBO Pendant® Onset Computer Corporation, Cape Cod, Massachusetts, USA) recorded water temperature in the treatment tanks every four hours. This logging interval was chosen because the thermal inertia of water meant that a finer scale logging interval was not considered necessary. We visually inspected egg containers every day for newly hatched larvae. We counted

25 hatchlings on the first and second day of hatching. By the second day, more than half of the larvae had hatched from the group (average for all treatments: 68.5% ± 0.2). For this reason, we recorded the second day of observed hatching as ‘day of hatching’ for the whole group.

2.2.3 Experiment 2: Effects of warming on early life stages of Le. intacta

For the second experiment, we measured the effects of temperature on embryonic development time, body size at hatching, growth rate and survival in a single species, Le. intacta. We collected eggs from 10 mating Le. intacta females near ponds at KSR using the same methods as above. We transported egg clutches to the lab at UTM on the same day as collection. Eggs were collected between 26 May and 9 June 2015. Treatment tanks were 20-liter glass aquaria filled with dechlorinated and oxygenated water and heated with submersible water heaters (Visi- Therm® Deluxe). Our experiment had three constant treatment temperatures: 23.4 ± 0.28º C, 27.5 ± 0.54º C, and 30.4 ± 0.45º C. Additionally, to explore the effects of thermal variability on larval performance, we had a treatment with a mean temperature of 27.4 ± 3.22º C, but manipulated variation by alternating between 23º and 30º C on a weekly basis. We chose this variable regime to simulate periodic heat waves that are becoming increasingly common with climate change (Rahmstorf and Coumou 2011). Data loggers (HOBO Pendant®) recorded water temperature every four hours in experimental tanks. Each temperature treatment was replicated four times for a total of 16 tanks. Photoperiod was variable in this experiment, however, larvae were exposed to light at least 8-hours per day.

Again, using a split-brood design, we divided clutches from 10 females into four equal groups and placed each group into a different thermal treatment, so that eggs from each female were included in every treatment (n = 10 for all treatments, except 27.4º C – variable, n = 9). Eggs were visually inspected every day for hatching. Again, we chose the second day of observed hatching as ‘day of hatching’ for each group. After a group hatched, we randomly selected 10 individuals from each group and photographed that sub-sample to estimate hatchling head width (Nikon D3200 HD-SLR camera, Tamron 90mm f2.8 macro lens). Head widths are a common metric used to measure body size in odonates (Corbet 1999). Next, the photographed larvae were individually placed into 100-mL plastic cups filled with dechlorinated, oxygenated water. Each cup was fitted with a polystyrene ring and floated in a water bath in the treatment tank (Appendix I, Fig. S2.2). Larvae were held separately to monitor individual growth rates (n = 40,

38, 39, and 36 larvae in 23.4º, 27.5º, 30.4º and 27.4º C – variable, respectively). We replaced water lost to evaporation every two days with dechlorinated, oxygenated water so that cups were always ~90% full.

Larvae were fed a diet of washed and rinsed Artemia nauplii in addition to small zooplankton (Daphnia species) twice weekly. This feeding regime is considered low compared to other odonate lab studies (see: Suhling et al. 2015), however we did observe treatment effects (see RESULTS), indicating that the feeding rate was sufficient to allow for growth and for us to observe the effects of temperature on growth. Additionally, food limitation may be quite common for larvae developing in ponds making the comparisons of growth responses across different feeding levels of interest. Waste that accumulated in the bottom of cups was removed weekly with 5-mL plastic pipettes to prevent hypoxia. We photographed all larvae again either after they had died, or at the end of the experiment. Head widths of larvae were measured from these photographs using Image J (U. S. National Institutes of Health, Bethesda, Maryland, USA), with a 3-centimeter ruler for calibration. Growth rate was calculated as the difference in head widths (ln-transformed), then divided by the time period to get a rate of mm × day-1 (Hoffman and Poorter 2002). This experiment lasted for 75 days between June and August 2015.

2.2.4 Statistical analyses:

We used a linear mixed effects model (LME) with the function lmer from the R package ‘lme4’, version 1.1-9) to assess the effects of temperature, species identity, the interaction between temperature and species, and date laid (oviposition date) on egg development rate (1/days to hatching). Oviposition date was included to account for temporal heterogeneity in traits of eggs or larvae. We included tank, and the maternal identity (female) as random effects. Starting with a fully parameterized model, we sequentially removed random effects and compared that model to the full model using log-likelihood ratio tests (log-LRT) and eliminated random effects with p > 0.05. We then used the Anova function from the R package ‘car’ (version 2.1-6) to estimate the significance of each fixed effect.

Similar to above, for our second experiment we used a LME model to analyze the effects of treatment temperature, oviposition date, and their interaction on egg development rate for the species Le. intacta. Temperature was included as a fixed effect, because we were not just interested in the mean temperature conditions, but also the effects of temperature fluctuations.

Model selection procedure followed the same steps as above, in which we sequentially removed variables and performed log-likelihood ratio tests to assess their importance in the final model. This same procedure was used to separately analyze the effects of treatment temperature, oviposition date, the interaction between treatment and oviposition date on hatchling head width and larval growth rate female and the interaction between temperature and female on hatchling head width and larval growth rate. Tank and female identity were included as random effects. To determine the effects of temperature treatment on survival, we used a survival analysis using the package “survival” (version 2.38) and “coxme” (version 2.2-5). All analyses were performed in R version 3.2.3 (R Development Core Team 2016). Alternate versions of these analyses can be found in the supplemental materials (Appendix I, Table S2.1).

Results

2.3.1 Experiment 1: interspecific comparisons of egg development rate

Increasing temperature significantly increased egg development rates across all species (log- % LRT: #$ = 250.34, p < 0.0001; Fig. 2.1, Table 2.1). We also found a significant effect of species % identity on egg development rate (#& = 11.24, p = 0.01; Fig. 2.1, Table 2.1). Celithemis elisa had slower egg development compared to all other species (Tukey’s HSD for all: p < 0.01). We found no significant interaction between temperature and species identity on egg developmental % rate (#& = 2.69, p = 0.44; Fig. 2.1, Table 2.1).

2.3.2 Experiment 2: Effects of warming on early life stages of Le. intacta

Similar to results in our first experiment, we found that increasing temperature significantly % increased egg development rate (log-LRT: #& = 1537.26, p < 0.0001; Table 2.1). Eggs in the warmest temperatures (30.4º C) developed the fastest (8.9 ± 1.0 days) compared to eggs developing in coolest temperatures, 23.4º C (mean: 14.0 ± 1.1 days). All pairwise comparisons between treatments were significantly different (Tukey’s HSD with Bonferroni correction: p < 0.01 for all comparisons; Appendix I, Table S2.2). In addition to temperature, we found a significant interaction between oviposition date and temperature on egg development rate, with a % trend of faster development when oviposition date was later (#' = 176.60, p < 0.0001; Table 2.1). Eggs laid earlier tended to develop the fastest in warmer conditions, except for clutches collected mid-sampling, in which eggs developed significantly faster in variable temperatures.

Because we sampled over a total of three days, we have low power to be able to decipher this pattern further, however, it is an area that future research could explore.

Table 2.1 Results from the log-likelihood ratio tests for the effects of fixed and random factors on egg development rate, and hatchling head width based on a linear mixed-effects (LME) model. ANOVA was used to assess Le. intacta larval growth rates, as random effects contributed no significant variation to this response.

Response variable Factor Predictor df (2 P Egg development rate Fixed Temperature 1 250.34 < 0.0001 Species 3 11.24 0.01 Temperature × 3 2.69 0.44 Species Oviposition date 7 9.06 0.25 Random Female 1 0 1 Tank 1 10.854 0.0009

Le. intacta egg development rate Fixed Treatment 3 1537.26 < 0.0001 Oviposition date 2 31.47 < 0.0001 Treatment × 6 176.60 < 0.0001 oviposition date

Random Female 1 26.37 < 0.0001 Tank 1 0 1 Treatment 3 159.12 < 0.0001 Le. intacta Fixed hatchling head width Oviposition date 2 7.73 0.02 Treatment × 6 132.93 < 0.0001

oviposition date Female 1 28.23 < 0.0001 Random Tank 1 0 1 Response variable Predictor df, resid df F P ANOVA: Le. intacta growth rate Treatment 3, 126 9.53 < 0.0001 Oviposition date 2, 126 4.54 0.01 Treatment × 6,126 3.18 0.006 oviposition date

Note: Predictors with p < 0.05 (set in boldface type) were significant and were included in the final LME model.

29 Fig. 1

Species

) 0.175

-1 C. elisa L. intaca L. luctuosa 0.150 L. pulchella

0.125

0.100

0.075 Development rate (days to hatching to (days rate Development 0.050 22 24 26 28 30 Treatment temperature (oC)

Figure 2.1 Effects of temperature on egg development rate in four odonate species: Celithemis elisa (squares); Leucorrhinia intacta (circles); Libellula luctuosa (triangles); Libellula pulchella (diamonds). Temperature means were 22.2, 24.9, 27.0, and 30.7° C. Error bars are ±1 SE. Points are jittered on the x-axis. n = 20, 21, 21, and 20 egg clutches for each temperature, respectively.

There was a significant effect of treatment temperature on Le. intacta hatchling head width (log- % LRT: #& = 159.12, p < 0.0001; Fig. 2.2a, Table 2.1). Hatchlings reared at 30.4º C had the smallest head widths, with heads significantly smaller than hatchlings raised in 23.4º C (7% smaller; Tukey’s HSD with Bonferroni correction: p < 0.001). There was no significant difference between head widths of larvae developing in 27.5º C – constant compared to head widths of larvae raised in and 27.4º C – variable (Tukey’s HSD p = 1.0; Appendix I, Table S2.1). Lastly, there was a significant interaction between treatment and oviposition date in which larvae that developed as eggs in warmer temperatures hatched out smaller if their mothers laid eggs % earlier (log-LRT: #' = 132.93, p < 0.0001; Table 2.1). Variation in hatchling head width was % attributed to the mother’s identity (#$ = 28.23, p < 0.0001) and this effect explained around 51% of the total variation.

Temperature had a significant effect on larval growth rate (ANOVA: F3,126 = 9.53, p < 0.0001; Fig. 2.2b, Table 2.1), where growth rates were 29% faster in the warmest temperatures, 30.4º C (0.029 ± 0.01 mm/day), compared to the coolest temperatures, 23.4º C (Tukey’s HSD: p = 0.04; Supplementary Table S2.2). We also found a significant interaction between temperature and oviposition date on growth rates (F6,126 = 3.18, p = 0.006; Table 2.1). Growth rates were slowest in the intermediate 27.5º – constant and 27.4º C – variable treatments – a 39% and 33% reduction, respectively, compared to growth rates in 30.4º C. Larval growth rate was not significantly different between the intermediate 27.5º – constant and 27.4º C – variable temperatures (Tukey’s HSD: p = 1.0; Table S2.2).

% Warmer temperatures had a significant negative effect on survival (LRT: #& = 38.21, p < 0.001; Fig. 2.3). Larvae raised in 30.4º C experienced mortality sooner in the experiment (restricted mean ± SE: 16.1 ± 2.3 days compared to larvae raised at 23.4º C: 40.2 ± 3.64 days). We found no significant differences in survival between intermediate 27.5º – constant and 27.4º C – variable treatments (Tukey’s HSD: p = 0.99; Table S2.2).

(a)

(b)

We examined the effects of warming on odonates in two ways, through examining its effects on egg development rates in a suite of co-occurring, libellulid dragonflies and by comparing early life-history responses to warming in a single dragonfly species. Higher temperatures increased egg development rates in similar ways across the four study species. Our study of the single species, Le. intacta, found that temperature and growth rate were positively related, while size at hatching and survival rates decreased with increasing temperature. We also uncovered two sources of intraspecific variation in Le. intacta responses to thermal conditions, differences among clutches of different mothers and to the date of oviposition. Our results provide a greater

33 understanding into the effects of temperature on early life-history and performance in aquatic ectotherms.

2.4.1 The effects of temperature on inter- and intraspecific egg development rate

In our first experiment, we found that warming significantly accelerated hatching time in the four species we examined and responses among species did not differ (Fig. 2.1). While we expected faster egg development, the consistency with which all species responded to warming was interesting given previous observations that species’ phenological responses to warming are often idiosyncratic (Guo et al. 2009, Diamond et al. 2011, Caradonna et al. 2014). Differential species responses to temperature could affect species interactions and therefore indirectly affect species abundances and community composition (Ohlberger 2013). By making direct comparisons of phenological responses to temperature we can increase our ability to successfully make predictions about the winners and losers of climate change and more broadly, how communities will be affected by warming climates (Ohlberger 2013). In our study, our results suggest that within this subset of libellulid dragonflies the acceleration of egg development in response to temperature may be remarkably consistent.

Timing of egg hatching is crucial in determining when species interact and can therefore have large effects on species interactions, for example intraguild predation (IGP). Perturbations that change the rank order of hatching among species, including differential species responses to temperature, could alter the structure of IGP interactions (e.g. reversing intraguild predator/prey roles), possibly altering population demographics (Guo et al. 2009) or even leading to the loss of certain species (Nakazawa and Doi 2011). The consistent response in egg development rate we found has important implications for this group in temperate regions where climate change is predicted to increase mean summer temperatures—the period in which these species’ eggs develop (Paulson 2011). While increasing mean temperatures will accelerate egg development rates in all of these species, the current order of hatching times will be maintained. Among the species we studied, there was no shift in the rank order of time to hatching. These results suggest that interactions such as IGP, which are structured by body size, will remain relatively consistent even with future climate warming. While our interspecific comparisons captured a small but important portion of odonate life history, future studies should examine how similar these

34 species remain in their responses to warming or if their responses diverge at later ontogenetic stages.

In our second experiment in which we measured early life-history responses to temperature in a single odonate species, Le. intacta, we found that temperature accelerated egg development rates. However, we also found significant intraspecific variation in these developmental responses to temperature with approximately 51% of the total variation in egg development time was attributed to the mother’s identity. Additionally, eggs laid by females that emerged earlier in the season had faster development rates than eggs laid later in the season. Oviposition date is inherently tied to the female and therefore it remains unclear whether females that emerge earlier in the season lay eggs that develop faster, or if this effect is driven by environmental conditions. One hypothesis is that females that emerge earlier in the season have faster growth rates than females that take longer to develop and emerge later in the season. In turn, the eggs of these “fast” females may also have faster development rates. Given that oviposition date appears to explain much of the variation in egg development rate, future studies should explore how developmental responses of offspring from mothers emerging at different times in the season vary.

2.4.2 The effect of temperature on body size, growth rate and survival of Le. intacta

In our second experiment, we measured several variables that relate to early larval performance, including hatchling head width—a measure of larval body size, as well as growth rate and survival. We expected that hatchling head width would be smallest in the warmest temperatures and our results support this prediction. Hatchling head width was inversely related to rearing temperature (Fig. 2.2a). While our study did not control for the maternal environment, we were able to strictly control the environmental conditions for the duration of offspring development. Our results conform to what prior studies have found for other taxa, however they differ from previous findings for odonates. One study found that hatchling size of a damselfly species was positively related to rearing temperature (Śniegula et al. 2016), while our study found an inverse relationship between these variables. Body size at hatching can affect the prey or resources the hatchlings can use and the types of predators they are vulnerable to (Werner and Gilliam 1984, Scharf et al. 2000). Future studies should examine the consequences of hatchling body size, specifically its role in performance at later stages of development.

We also found a significant pattern of smaller head width at time of hatching for larvae developing in 27.4º variable compared to 27.5º constant (Appendix I, Table S2.2). This suggests that temperature variation may have effects different than mean increases in temperature on embryological developmental physiology. As for the other responses (egg development, growth and mortality rate) we found no evidence that the constant and variable temperatures had differing effects on larvae (Tukey’s HSD: p > 0.10; Appendix I, Table S2.1). Overall, our results suggest that fluctuating temperatures may not have observable effects on insects during early stages of development. However, a growing body of literature examining the effects of increased thermal variation have found relationships between fluctuating temperature and life history traits and performance such as development time (Kingsolver et al. 2009), phenotype (Pétavy et al. 2004, Małek et al. 2015), survival (Ragland and Kingsolver 2008), and fitness (Estay et al. 2011). This suggests that the effects of increased thermal variation are cumulative and may have consequences at later life history stages. Vannote and Sweeney (1980) pointed out that natural variation in stream temperatures can affect aquatic insect populations, that these effects can be mediated by factors aside from temperature induced mortality (e.g. effects of temperature on body size and fecundity), and that it can also affect species’ geographic distributions. Their work may provide a useful template for future studies of the effects of climate change associated thermal variation on species and populations. Our work along with others demonstrate the complexity of thermal variation on life history processes and suggest that thermal variation should continue to be incorporated into future experiments to model the outcome of species’ abilities to persist with future climate change (Vazquez et al. 2017).

We found that intraspecific variation in hatchling size was explained by maternal identity. This suggests that body size may be influenced by other factors in addition to temperature; there may be heritable or maternal effects underlying body size at hatching, thus generating a wide range of hatchling body size. This variation in hatchling size could reduce competition among individuals but also increase rates of cannibalism, as this phenomenon, like IGP, is facilitated by body size variation (Claessen et al. 2000). Taken together, our results imply that the developmental environment of eggs is important in determining body size at hatching and also driving hatchling body size variation between egg clutches of different mothers. Whether these variable responses among families are important in structuring interactions such as intraspecific competition and cannibalism should be explored further.

The general pattern we found of higher growth rates at warmer temperatures corresponds well with previous studies measuring the effects of temperature on odonate growth rates (Krishnaraj and Pritchard 1995, Pritchard et al. 2000, Nilsson-Ortman et al. 2014, Suhling et al. 2015) (Fig. 2.2b). These studies have found that growth rates increase in warmer temperatures (except see: Van Doorslaer and Stoks 2005a), an expected result as ectotherm growth rates are known to scale allometrically with temperature (Angilletta et al. 2004). Faster growth rates are likely to underlie observed patterns of advancing phenology, such as earlier adult emergence in odonates (Hassall et al. 2007, McCauley et al. 2015). Whether accelerated growth rates in the context of a warming environment has net benefits or costs for populations will depend on the species and the ecological context in which these animals exist.

We found that temperature had a negative effect on larval survival at early stages of development (Fig. 2.3). Increases in temperature as a result of climate change could potentially add additional sources of mortality to populations that already have low larval survival (Cornell and Hawkins 1995, Hirst and Kiørboe 2002, Forster and Hirst 2012). Temperature appears to increase odonate mortality when larvae develop and metamorphose in heated conditions (McCauley et al. 2015). These types of studies, however, often do not track survival throughout development and therefore it remains unclear at what stage mortality is occurring. Our results suggest that temperature-induced mortality may occur at early stages of development. This is important for two reasons. First, increases in mortality at early life stages reduces the size of the population entering adult stages of development and therefore could influence population dynamics, as only adults reproduce. Second, an increase in mortality at early stages could change interactions between juveniles. With smaller population sizes, competition as well as cannibalism among individuals could decrease, thereby benefiting those individuals that survive in warmer conditions. It remains to be determined how mortality in early developmental stages will influence odonate populations and food web structure in these systems.

Conclusion

Our results demonstrate that early life history traits and performance are altered at higher temperatures, at both the intra- and interspecific levels. In our interspecific comparisons we found that all species decrease developmental rates in response to temperature, with no significant variation between species responses. We suggest that the absence of variation

37 between species’ responses to warming means that hatching order and thus interspecific interactions between the species examined here will be maintained with future climate change. Our intraspecific comparisons found that egg development and growth rates were directly related to temperature, whereas hatchling size and survival decreased with increasing temperature. Egg development rate in responses to temperature varied among clutches of eggs laid at different dates. Hatchling size varied among clutches of different mothers. The high levels of intraspecific variation we observed in egg development time and hatchling size suggest that even with increased temperatures, populations of this species may be relatively more resilient to future climate warming.

Acknowledgements We thank S. French, C. Baines, R. Martin, and C. Searcy for providing comments on the manuscript; C. Fitzpatrick for help with analyses; S. Catania and D. Przelicki for laboratory assistance; and two anonymous reviewers for their constructive feedback. We also thank the Koffler Scientific Reserve for research support. Funding was provided by the Department of Biology at the University of Toronto, Mississauga, and by a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant to S.J.M. Support to D.N.F. during the writing of this manuscript was provided in part by a Zimmerman–Weis Fellowship from the Koffler Scientific Reserve. Funding was provided to support research by J.Y.M. as part of an undergraduate Research Opportunity Project by the Office of the Dean at the University of Toronto, Mississauga.

References Abrahams, M.V., Mangel, M., and Hedges, K. 2007. Predator–prey interactions and changing environments: who benefits? Philos. Trans. R. Soc. B Biol. Sci. 362(1487): 2095–2104. Agashe, D. 2009. The stabilizing effect of intraspecific genetic variation on population dynamics in novel and ancestral habitats. Am. Nat. 174(2): 255–267. Angilletta, M.J., Steury, T.D., and Sears, M.W. 2004. Temperature, growth rate, and body size in ectotherms: fitting pieces of a life-history puzzle. Integr. Comp. Biol. 44(6): 498–509. Atkinson, D. 1994. Temperature and organism size: a biological law for ectotherms? Adv. Ecol. Res. 25: 1–58. Atkinson, D. 1995. Effects of temperature on the size of aquatic ectotherms: exceptions to the general rule. J. Therm. Biol. 20(1–2): 61–74.

Bolnick, D.I., Amarasekare, P., Araújo, M.S., Bürger, R., Levine, J.M., Novak, M., Rudolf, V.H.W., Schreiber, S.J., Urban, M.C., and Vasseur, D.A. 2011. Why intraspecific trait variation matters in community ecology. Trends Ecol. Evol. 26(4): 183–192. Cahill, A.E., Aiello-Lammens, M.E., Fisher-Reid, M.C., Hua, X., Karanewsky, C.J., Yeong Ryu, H., Sbeglia, G.C., Spagnolo, F., Waldron, J.B., Warsi, O., and Wiens, J.J. 2012. How does climate change cause extinction? Proc. R. Soc. B. Biol. Sci. 280(1750): 20121890– 20121890. Caradonna, P.J., Iler, A.M., and Inouye, D.W. 2014. Shifts in flowering phenology reshape a subalpine plant community. Proc. Natl. Acad. Sci. U.S.A. 111(13): 4916–4921. Carroll, D.P., and Hoyt, S.C. 1986. Some effects of parental rearing conditions and age on progeny birth weight, growth, development, and reproduction in the apple aphid, Aphis pomi (Homoptera: Aphididae). Environ. Entomol. 15(3): 614–619. Chown, S.L., Hoffmann, A.A., Kristensen, T.N., Angilletta, M.J., Jr., Stenseth, N.C., and Pertoldi, C. 2010. Adapting to climate change: a perspective from evolutionary physiology. Clim. Res. 43(1): 3–15. Claessen, D., de Roos, A.M., and Persson, L. 2000. Dwarfs and giants: cannibalism and competition in size-structured populations. Am. Nat. 155(2): 219–237. Cole, V.J., Parker, L.M., O’Connor, S.J., O’Connor, W.A., Scanes, E., Byrne, M., and Ross, P.M. 2016. Effects of multiple climate change stressors: ocean acidification interacts with warming, hyposalinity, and low food supply on the larvae of the brooding flat oyster Ostrea angasi. Mar. Biol. 163(5): 125. Colinet, H., Sinclair, B.J., Vernon, P., and Renault, D. 2015. Insects in fluctuating thermal environments. Annu. Rev. Entomol. 60(1): 123–140. Collin, R., and Salazar, M.Z. 2010. Temperature-mediated plasticity and genetic differentiation in egg size and hatching size among populations of Crepidula (Gastropoda: Calyptraeidae). Biol. J. Linn. Soc. 99: 489–499. Corbet, P.S. 1999. Dragonflies: behaviour and ecology of Odonata. Harley Books, Colchester, Essex, UK. Corbet, P.S., Suhling, F., and Soendgerath, D. 2006. Voltinism of Odonata: a review. Int. J. Odonatol. 9(1): 1–44. Cornell, H.V., and Hawkins, B.A. 1995. Survival patterns and mortality sources of herbivorous insects: some demographic trends. Am. Nat. 145(4): 563–593. Dallas, H.F., and Rivers-Moore, N.A. 2011. Critical thermal maxima of aquatic macroinvertebrates: towards identifying bioindicators of thermal alteration. Hydrobiologia, 679(1): 61–76. De Bie, T., De Meester, L., Brendonck, L., Martens, K., Goddeeris, B., Ercken, D., Hampel, H., Denys, L., Vanhecke, L., Van der Gucht, K., Van Wichelen, J., Vyverman, W., and Declerck, S.A.J. 2012. Body size and dispersal mode as key traits determining metacommunity structure of aquatic organisms. Ecol. Lett. 15(7): 740–747.

Deschaseaux, E.S.M., Taylor, A.M., Maher, W.A., and Davis, A.R. 2010. Cellular responses of encapsulated gastropod embryos to multiple stressors associated with climate change. J. Exp. Mar. Biol. Ecol. 383(2): 130–136. Diamond, S.E., Frame, A.M., Martin, R.A., and Buckley, L.B. 2011. Species’ traits predict phenological responses to climate change in butterflies. Ecology, 92(5): 1005–1012. Dibble, C.J., Hall, S.R., and Rudolf, V.H.W. 2014. Intraspecific priority effects and disease interact to alter population growth. Ecology, 95(12): 3354–3363. Dingemanse, N.J., and Kalkman, V.J. 2008. Changing temperature regimes have advanced the phenology of Odonata in the Netherlands. Ecol. Entomol. 33(3): 394–402. Duffy, M.A. 2010. Ecological consequences of intraspecific variation in lake Daphnia. Freshw. Biol. 55: 995–1004. Easterling, D.R., Meehl, G.A., Parmesan, C., Changnon, S.A., Karl, T.R., and Mearns, L.O. 2000. Climate extremes: observations, modeling, and impacts. Science, 289(5487): 2068– 2074. Elliott, J.M. 1978. Effect of temperature on the hatching time of eggs of Ephemerella ignita (Poda) (Ephemeroptera:Ephemerellidae). Freshw. Biol. 8(1): 51–58. Ernsting, G., and Isaaks, J.A. 1997. Effects of temperature and season on egg size, hatchling size and adult size in Notiophilus biguttatus. Ecol. Entomol. 22(1): 32–40. Estay, S.A., Clavijo-Baquet, S., Lima, M., and Bozinovic, F. 2011. Beyond average: an experimental test of temperature variability on the population dynamics of Tribolium confusum. Popul. Ecol. 53(1): 53–58. doi:10.1007/s10144-010-0216-7. Feltmate, B., and Thistlethwaite, J. 2012. Climate change adaptation: a priorities plan for Canada. Climate Change Adaptation Project, Intact Centre on Climate Adaptation, Waterloo, Ont. Fischer, K., Brakefield, P.M., and Zwaan, B.J. 2003. Plasticity in butterfly egg size: why larger offspring at lower temperatures? Ecology, 84(12): 3138–3147. Folt, C.L., Chen, C.Y., Moore, M.V., and Burnaford, J. 1999. Synergism and antagonism among multiple stressors. Limnol. Oceanogr. 44(3, part 2): 864–877. Forrest, J.R.K., and Thomson, J.D. 2011. An examination of synchrony between insect emergence and flowering in Rocky Mountain meadows. Ecol. Monogr. 81(3): 469–491. Forster, J., and Hirst, A.G. 2012. The temperature-size rule emerges from ontogenetic differences between growth and development rates. Funct. Ecol. 26(2): 483–492. Gardner, J.L., Peters, A., Kearney, M.R., Joseph, L., and Heinsohn, R. 2011. Declining body size: a third universal response to warming? Trends Ecol. Evol. 26(6): 285–291. Garten, C.T., Jr., and Gentry, J.B. 1976. Thermal tolerance of dragonfly nymphs. II. Comparison of nymphs from control and thermally altered environments. Physiol. Zool. 49(2): 206– 213. Genner, M.J., Halliday, N.C., Simpson, S.D., Southward, A.J., Hawkins, S.J., and Sims, D.W. 2010. Temperature-driven phenological changes within a marine larval fish assemblage. J. Plankton Res. 32(5): 699–708.

Guo, K., Hao, S.-G., Sun, O.J., and Kang, L. 2009. Differential responses to warming and increased precipitation among three contrasting grasshopper species. Glob. Change Biol. 15(10): 2539–2548. Gutzke, W., and Packard, G.C. 1987. Influence of the hydric and thermal environments on eggs and hatchlings of bull snakes Pituophis melanoleucus. Physiol. Zool. 60(1): 9–17. Hassall, C., Thompson, D.J., French, G.C., and Harvey, I.F. 2007. Historical changes in the phenology of British Odonata are related to climate. Glob. Change Biol. 13(5): 933–941. Hirst, A.G., and Kiørboe, T. 2002. Mortality of marine planktonic copepods: global rates and patterns. Mar. Ecol. Prog. Ser. 230: 195–209. Hoffman, W.A., and Poorter, H. 2002. Avoiding bias in calculations of relative growth rate. Ann. Bot. 90(1): 37–42. Howe, R.W. 1967. Temperature effects on embryonic development in insects. Annu. Rev. Entomol. 12: 15–42. Huey, R.B., and Kingsolver, J.G. 1989. Evolution of thermal sensitivity of ectotherm performance. Trends Ecol. Evol. 4(5): 131–135. IPCC. 2012. Managing the risks of extreme events and disasters to advance climate change adaptation: a special report of working groups I and II of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK, and New York, N.Y., USA. Janssens, L., and Stoks, R. 2013. Fitness effects of chlorpyrifos in the damselfly Enallagma cyathigerum strongly depend upon temperature and food level and can bridge metamorphosis. PLoS ONE, 8(6): e68107. Kingsolver, J.G., and Huey, R.B. 2008. Size, temperature, and fitness: three rules. Evol. Ecol. Res. 10: 251–268. Kingsolver, J.G., Ragland, G.J., and Diamond, S.E. 2009. Evolution in a constant environment: thermal fluctuations and thermal sensitivity of laboratory and field populations of Manduca sexta. Evolution, 63(2): 537–541. Klockmann, M., Günter, F., and Fischer, K. 2017. Heat resistance throughout ontogeny: body size constrains thermal tolerance. Glob. Change Biol. 23: 686–696. Krishnaraj, R., and Pritchard, G. 1995. The influence of larval size, temperature, and components of the functional response to prey density on growth rates of the dragonflies Lestes disjunctus and Coenagrion resolutum (Insecta: Odonata). Can. J. Zool. 73(9): 1672– 1680. Leggott, M., and Pritchard, G. 1985. The effect of temperature on rate of egg and larval development in populations of Argia vivida Hagen (Odonata: Coenagrionidae) from habitats with different thermal regimes. Can. J. Zool. 63(11): 2578–2582. Li, Y., Cohen, J.M., and Rohr, J.R. 2013. Review and synthesis of the effects of climate change on amphibians. Integr. Zool. 8(2): 145–161. Liefting, M., Weerenbeck, M., Van Dooremalen, C., and Ellers, J. 2010. Temperature- induced plasticity in egg size and resistance of eggs to temperature stress in a soil arthropod. Funct. Ecol. 24(6): 1291–1298.

Małek, D., Drobniak, S., Gozdek, A., Pawlik, K., and Kramarz, P. 2015. Response of body size and developmental time of Tribolium castaneum to constant versus fluctuating thermal conditions. J. Therm. Biol. 51: 110–118. McCauley, S.J. 2008. Slow, fast and in between: habitat distribution and behaviour of larvae in nine species of libellulid dragonfly. Freshw. Biol. 53(2): 253–263. McCauley, S.J., Hammond, J.I., Frances, D.N., and Mabry, K.E. 2015. Effects of experimental warming on survival, phenology and morphology of an aquatic insect (Odonata). Ecol. Entomol. 40(3): 211–220. Nakazawa, T., and Doi, H. 2011. A perspective on match/mismatch of phenology in community contexts. Oikos, 121(4): 489–495. Nilsson-Ortman, V., Stoks, R., and Johansson, F. 2014. Competitive interactions modify the temperature dependence of damselfly growth rates. Ecology, 95(5): 1394–1406. Ohlberger, J. 2013. Climate warming and ectotherm body size — from individual physiology to community ecology. Funct. Ecol. 27(4): 991–1001. Oney, B., Reineking, B., O’Neill, G., and Kreyling, J. 2013. Intraspecific variation buffers projected climate change impacts on Pinus contorta. Ecol. Evol. 3(2): 437–449. Parmesan, C. 2006. Ecological and evolutionary responses to recent climate change. Annu. Rev. Ecol. Evol. Syst. 37: 637–669. Paulson, D. 2011. Dragonflies and damselflies of the East. Princeton University Press, Princeton, N.J. Pétavy, G., David, J.R., Debat, V., Gibert, P., and Moreteau, B. 2004. Specific effects of cycling stressful temperatures upon phenotypic and genetic variability of size traits in Drosophila melanogaster. Evol. Ecol. Res. 6(6): 873–890. Pritchard, G., Harder, L.D., and Mutch, R.A. 1996. Development of aquatic insect eggs in relation to temperature and strategies for dealing with different thermal environments. Biol. J. Linn. Soc. 58(2): 221–244. Pritchard, G., Harder, L.D., Kortello, A., and Krishnaraj, R. 2000. The response of larval growth rate to temperature in three species of coenagrionid dragonflies with some comments on Lestes disjunctus (Odonata: Coenagrionidae, Lestidae). Int. J. Odonatol. 3(2): 105–110. Ragland, G.J., and Kingsolver, J.G. 2008. The effect of fluctuating temperatures on ectotherm life-history traits: comparisons among geographic populations of Wyeomyia smithii. Evol. Ecol. Res. 10(1): 29–44. Rahmstorf, S., and Coumou, D. 2011. Increase of extreme events in a warming world. Proc. Natl. Acad. Sci. U.S.A. 108(44): 17905–17909. Rasmussen, N.L., Van Allen, B.G., and Rudolf, V.H.W. 2014. Linking phenological shifts to species interactions through size-mediated priority effects. J. Anim. Ecol. 83(5): 1206– 1215. R Core Team. 2016. R: a language and environment for statistical computing. Version 3.2.3 [computer program]. R Foundation for Statistical Computing, Vienna, Austria. Available from https://www.r-project.org/.

Rohr, J.R., and Palmer, B.D. 2013. Climate change, multiple stressors, and the decline of ectotherms. Conserv. Biol. 27(4): 741–751. Rudolf, V.H.W. 2012. Seasonal shifts in predator body size diversity and trophic interactions in size-structured predator–prey systems. J. Anim. Ecol. 81(3): 524–532. Rudolf, V.H.W., and Singh, M. 2013. Disentangling climate change effects on species interactions: effects of temperature, phenological shifts, and body size. Oecologia, 173(3): 1043–1052. Rundle, S.D., Bilton, D.T., Abbott, J.C., and Foggo, A. 2007. Range size in North American Enallagma damselflies correlates with wing size. Freshw. Biol. 52: 471–477. Scharf, F.S., Juanes, F., and Rountree, R.A. 2000. Predator size-prey size relationships of marine fish predators: interspecific variation and effects of ontogeny and body size on trophic- niche breadth. Mar. Ecol. Prog. Ser. 208: 229–248. Schneider, K., and Mauser, W. 1996. Processing and accuracy of Landsat Thematic Mapper data for lake surface temperature measurement. Int. J. Remote Sens. 17(11): 2027–2041. Semlitsch, R.D., Scott, D.E., and Pechmann, H.K. 1988. Time and size at metamorphosis related to adult fitness in Ambystoma talpoideum. Ecology, 69(1): 184–192. Sheridan, J.A., and Bickford, D. 2011. Shrinking body size as an ecological response to climate change. Nat. Clim. Change, 1: 401–406. Sibly, R.M., and Atkinson, D. 1994. How rearing temperature affects optimal adult size in ectotherms. Funct. Ecol. 8(4): 486–493. Sniegula, S., Golab, M.J., and Johansson, F. 2016. A large-scale latitudinal pattern of life-history traits in a strictly univoltine damselfly. Ecol. Entomol. 41:459–472. Sokolova, I.M., and Lannig, G. 2008. Interactive effects of metal pollution and temperature on metabolism in aquatic ectotherms: implications of global climate change. Clim. Res. 37(2–3): 181–201. Sokolovska, N., Rowe, L., and Johansson, F. 2000. Fitness and body size in mature odonates. Ecol. Entomol. 25: 239–248. Suhling, F., Suhling, I., and Richter, O. 2015. Temperature response of growth of larval dragonflies — an overview. Int. J. Odonatol. 18(1): 15–30. Sunday, J.M., Bates, A.E., and Dulvy, N.K. 2012. Thermal tolerance and the global redistribution of animals. Nat. Clim. Change, 2(9): 686–690. Van Damme, R., Bauwens, D., Braña, F., and Verheyen, R.F. 1992. Incubation temperature differentially affects hatching time, egg survival, and hatchling performance in the lizard Podarcis muralis. Herpetologica, 48(2): 220–228. Van Doorslaer, W., and Stoks, R. 2005a. Growth rate plasticity to temperature in two damselfly species differing in latitude: contributions of behaviour and physiology. Oikos, 111(3): 599–605. Van Doorslaer, W., and Stoks, R. 2005b. Thermal reaction norms in two Coenagrion damselfly species: contrasting embryonic and larval life-history traits. Freshw. Biol. 50(12): 1982– 1990.

Vannote, R.L., and Sweeney, B.W. 1980. Geographic analysis of thermal equilibria: a conceptual model for evaluating the effect of natural and modified thermal regimes on aquatic insect communities. Am. Nat. 115: 667–695. Vazquez, D.P., Gianoli, E., Morris, W.F., and Bozinovic, F. 2017. Ecological and evolutionary impacts of changing climatic variability. Biol. Rev. Camb. Philos. Soc. 92: 22–42. Visser, M.E., and Holleman, L.J. 2001. Warmer springs disrupt the synchrony of oak and winter moth phenology. Proc. R. Soc. B Biol. Sci. 268(1464): 289–294. Walker, E.M. 1953. The Odonata of Canada and Alaska. Vol. 1. University of Toronto Press, Toronto, Ont. Walther, G.-R., Post, E., Convey, P., Menzel, A., Parmesan, C., Beebee, T.J.C., Fromentin, J.M., Hoegh-Guldberg, O., and Bairlein, F. 2002. Ecological responses to recent climate change. Nature, 416(6879): 389–395. Werner, E.E., and Gilliam, J.F. 1984. The ontogenetic niche and species interactions in size- structured populations. Annu. Rev. Ecol. Syst. 15: 393–42

Warming drives higher rates of prey consumption and increases rates of intraguild predation

Published as Frances, D. N., & McCauley, S. J. 2018. Warming drives higher rates of prey consumption and increases rates of intraguild predation. Oecologia, doi:/10.1007/s00442-018- 4146-y

3 Abstract

Warming due to climate change is expected to alter species interactions. These interactions are shaped by components of individual behaviour, particularly foraging behaviours. However, few studies consider species’ behavioural responses to warming to predict how species interactions will be affected by warming. We chose two complementary approaches to examine how climate warming may affect the behaviour and interactions of aquatic intraguild predators. First, we measured behavioural responses to warming in six larval dragonfly species, expecting that feeding rate and activity level would increase with temperature. Secondly, we conducted intraguild predation (IGP) trials with three species to understand how temperature affects IGP, and if species’ behavioural responses to warming are indicative of the outcome of IGP interactions. Warming increased feeding rates by 42% on average across species but had no effect on activity level. The magnitude of change in feeding rate was positively correlated with the maximum temperatures species experience across their ranges. Lastly, warming increased rates of IGP two-fold, however, species’ behavioural responses alone were not predictive of their susceptibility to become IG prey of other larvae at warmer temperatures. Our results provide evidence that IGP interactions may be greatly affected by future increases in temperature, however, activity responses to warming alone are weak predictors of the outcomes of these interactions. Future studies should consider other species’ traits when forecasting the effects of climate change on species interactions.

Introduction

In ectotherms, organismal physiology is directly affected by temperature, with metabolic rate increasing exponentially with environmental temperature until an upper critical limit is reached

(Gillooly et al., 2001). Biological parameters scale with metabolic rate including growth (Angilletta & Dunham, 2003), development time (Gillooly et al., 2002), and fitness (Kingsolver & Huey, 2008). Experiencing temperatures outside of an individual’s optimal body temperature can be costly, decreasing performance (Huey & Stevenson, 1979) and scaling up to affect demography (Frazier et al., 2006) and community structure (Dossena et al., 2012). The direct influence of warming on these physiological processes has received a great deal of attention (Dell et al., 2011), especially based on mounting evidence of rapidly increasing temperatures caused by climate change (IPCC, 2014). Unfortunately, many studies fail to incorporate correlated temperature-induced behavioural changes into their predictions of how species will be affected by future climate warming. Specifically, we know little of the impacts of rising temperatures on individual behaviour, and how those behavioural changes may affect species interactions (O’Connor, 2009; Smolinský & Gvoždík, 2014).

Foraging behaviours are of particular importance because they contribute to shaping interactions within communities (Lima & Dill, 1990). Theoretical and empirical work have shown that components of the functional response are not only highly influenced by temperature, but are also highly variable among species (Englund et al., 2011). Feeding rate is expected to increase with warming because animals with higher metabolic rates require more food to satisfy increased energetic demands (Elliott, 1976). Additionally, higher temperatures should theoretically increase activity level, ultimately increasing encounter rates between predators and prey and lead to higher predation rates (Lima & Dill, 1990; Werner & Anholt, 1993). Therefore, understanding the effects of temperature on activity and other behaviours may provide insights into how the strength of predator-prey interactions will be affected by climate change and the consequences of this for community structure.

Theoretical models predict that community responses to climate warming will be a result of the combined differential responses between predators and prey to increasing temperatures (Vasseur & McCann, 2005). Some differential responses have already been documented. For instance, warming can differentially alter phenological responses of interacting species (Rudolf & Singh, 2013), can change the spatial overlap of predators and prey (Smolinský & Gvoždík, 2014), and potentially affect their interaction strengths (Rall et al., 2010). Given the importance of behaviour in shaping interactions, there is a surprising lack of empirical work linking

46 behavioural responses to warming and species interactions. Additionally, little is known about how a different type of predator-prey interaction, intraguild predation (IGP), is affected by rising temperatures. Intraguild predation, the killing and consuming of potential competitors (Polis et al., 1989), occurs broadly across many systems and is important in structuring the demography and dynamics within communities (Arim & Marquet, 2004). A study by Barton and Schmitz (2009) found that warming can increase the spatial overlap of terrestrial predators, therefore increasing encounter rates and consequently IGP rates. Some theoretical work predicts that warming will increase connectance and interaction strengths within omnivorous food webs, leading to more unstable biological networks (Sentis et al., 2014). Additionally, some work has found cannibalism rate, like predation increases with warming (e.g. Crumrine 2010; Start 2017). It still remains unclear, however, how warming will affect trophic interactions between multiple species within the same guild. This remains an unanswered, yet important question, given that species often differ in their responses to warming and this could have large effects on IGP dynamics. Here we examined the degree to which behavioural responses to warming varied among a suite of co-occurring dragonfly species and then determined whether differential behavioural responses accurately predicted the outcome of species interactions at warmer temperatures.

Aquatic dragonfly larvae are ideal study organisms to address questions about the effects of warming on behaviour and species interactions for several reasons, including that they respond to both short and long-term changes in their thermal environment. For instance, temperature can affect the growth (Suhling et al., 2015), development (Flenner et al., 2010; Frances et al., 2017), and phenology of dragonflies and damselflies (McCauley et al., 2015). Additionally, dragonfly larvae are intraguild (IG) predators, competing with as well as preying on other dragonfly larvae (Polis et al., 1989), allowing us to test how the strength of IGP changes with temperature. IGP is generally driven by body size differences, with the magnitude of difference between individuals increasing the frequency of IGP (Polis et al., 1989) and therefore allowing for testable predictions of how size differences among species should affect IGP dynamics. Finally, the dragonfly species used in our experiments overlap in their distributions at our field site but have varying ranges across North America, allowing us to test for correlations between species’ climatic niches relative to the magnitude of their behavioural responses to warming. While factors such as competition and predation are important in shaping species’ distributions

(Connell, 1961), abiotic factors such as temperature remain highly important in determining where species can exist (Sunday et al., 2011). Previous work suggests that adaptation to different climates can contribute to shaping thermal tolerance (Addo-Bediako et al., 2000) with species’ northern and southern range limits correlating with the upper and lower bounds of their thermal tolerances (Calosi et al., 2010). While the upper lethal limit is generally conserved across insects, little is known about how performance at high temperatures but still within the sub-lethal range varies among species across different latitudes. Because this assemblage of dragonfly species varies in range distributions, and presumably thermal tolerance, they are ideal for assessing the relationship between macrophysiology and behavioural responses to warming.

We conducted two experiments to address how temperature affects behaviour and species interactions. First, we conducted behavioural assays on six dragonfly species at three different temperatures to assess how activity, feeding rate, and microhabitat preference changed with warming. We predicted that temperature would increase feeding rate and activity level, and larvae would alter microhabitat preference by disassociating from structure more frequently to hunt and sustain higher feeding demands. We also tested for correlations between species’ behavioural plasticity to warming and the maximum temperatures species experience within their ranges. We predicted that warm-adapted species (those with more central latitudinal ranges) would be likely to increase feeding and activity rates with temperatures, as their physiological tolerance may have evolved to allow them to maximize performance at warmer temperatures. In contrast, we expected cold-adapted species (those with higher latitude ranges) to be less able to respond behaviourally to rising temperatures by increasing feeding and/or activity levels, as they would be pushed beyond their thermal stress limits. Next, we examined how IGP predation was affected by warming with three species that varied to the greatest degree in their behavioural responses to warming and also in the thermal ranges they occupy. We predicted that temperature would increase rates of IGP, with the least active responder to temperature being at the greatest risk of IGP as they would be discovered and eaten at a higher rate by species with more behavioural plasticity in activity level in response to warming. Alternatively, the most active responder could be at the highest risk of IGP predation, as higher activity rates, and thus more frequent encounters with other larvae increase their likelihood of being killed by an IG predator. We assessed both of these possibilities in our experiments. Because IGP is typically influenced

48 by size-structure, we also examined how the strength of size-selective IGP changes with warming.

Materials and methods

3.2.1 Study system

Animals used in our experiments were collected from ponds at the University of Toronto’s Koffler Scientific Reserve (KSR) and at the University of Toronto, Mississauga (UTM). For sampling dates for each species, see Appendix II. We used the following species in our experiments: Erythemis simplicicollis (Say, 1839), Epitheca canis (McLachlan, 1886), Libellula pulchella (Drury, 1770), Sympetrum vicinum (Hagen, 1861), Leucorrhinia intacta (Hagen, 1861), and Plathemis lydia (Drury, 1773). These species belong to the superfamily Libelluloidea (Odonata: Anisoptera). We selected these species because they are common residents at our field sites with varying foraging strategies and microhabitat preferences (McCauley, 2008). Epitheca canis, L. pulchella, and P. lydia use sit-and-wait foraging strategies, and use both visual and mechanical senses to detect and capture approaching prey (Pritchard, 1965; Hopper et al., 1996). Leucorrhinia intacta, E. simplicicollis, and S. vicinum are more active predators, climbing on aquatic vegetation to detect and pursue moving prey (Pritchard, 1965; Johansson, 1991).

We kept larvae in our laboratory at UTM with conspecifics of similar size housed together in water-filled containers (5.5 L) with plenty of artificial structure to reduce the frequency of cannibalism (Wissinger, 1992). Larval head width, a standard proxy for total body size (Benke, 1970), varied by species (mean ± SD: E. canis: 4.58 ± 0.65 mm; S. vicinum: 4.26 ± 0.23 mm; E. simplicicollis: 4.22 ± 0.52 mm; L. intacta: 4.06 ± 0.58 mm; L. pulchella: 3.78 ± 0.80 mm; P. lydia: 3.56 ± 0.71 mm; Appendix II, Fig. S3.1). We maintained dragonfly larvae at room temperature (~22º Celsius) with a 14:10 (light: dark) photoperiod and fed them zooplankton (Daphnia and copepods) ad libitum every two to three days prior to starting the experiments. We ran our first experiment, the behavioural assays, from 13-June-2015 through 14-Oct-2015. Due to low sample sizes of E. simplicicollis in 2015, we conducted another round of behavioural assays for this species on 21 and 24-Sept-2016. Our second experiment, determining the effects of temperature on IGP, ran from 26-Aug-2016 to 24-Sept-2016.

3.2.2 Study 1: Examining the influence of temperature on foraging behaviour

Individual behavioural assays – To determine how temperature affects foraging behaviour, we observed behavioural responses to temperature in six different species. Observation tanks were 20-liter glass aquaria filled with aged tap water and heated with submersible water heaters (Visi- Therm® Deluxe, Marineland Aquarium Products, Cincinnati, Ohio, U.S.A.) placed on the bottom of the tank perpendicular to the longest edge. We heated tanks to one of three temperature conditions: 24º, 28º, and 32º C (means ± SD: 23.50 ± 0.31º, 28.21 ± 0.52º, 32.36 ± 0.77º C). Our highest temperatures are based on the projected increases in air temperature in the region for the middle and end of the century (Bush et al., 2014). The littoral zone where these larvae are found is shallow and tracks air temperatures. Average maximum temperatures in the region for June, July, August, September, and October 2015 were 22.6, 27.0, 25.5, 24.7, and 14.0º C, respectively (Environment Canada, retrieved 12-Sept-2017 from http://climate.weather.gc.ca).

We placed larvae individually in each observation tank 16-hours prior to starting the behavioural assays. This period allowed larvae to acclimate to the thermal conditions and standardized their hunger levels. Following the acclimation period, we added twenty mosquito larvae (Aedes and Culex species) into each observation tank as prey for the dragonfly larva. We chose this density in order to exceed the amount of prey that dragonflies can normally consume in the time period of these experiments. Differences in feeding rates among species, if they existed, could then be detected. Dragonfly larvae were randomly assigned to treatments and only used once. To observe larval activity, we used a standard method for quantifying dragonfly behaviour (Johansson, 2000; McCauley, 2008). A 3 x 2 grid taped to the back of the tank and on the sides was used to visually subdivide the tank into 12 cubes (each cube: 5 cm3). We could then determine the position of the larva in three-dimensional space. A single observer (D.N.F.) recorded the position of a larva’s head in the grid and if it was on or touching structure in the tank. Positional data of each larva within every tank was recorded every 5 minutes for 1 hour 15 minutes. After a 1 h hiatus, another round of observations was conducted for an additional 1 h 15 m for a total observation period of 2 h 30 m. We removed dragonfly larvae after they had spent a total of 5.54 ± 0.54 h (mean ± SD) in the tank with the mosquito larvae prey. Dragonfly head widths were measured

50 using Vernier calipers accuracy of ± 0.001mm. To remove remaining prey, we swept fine mesh aquaria nets through tanks at least three times. Tanks were then visually scanned after net- sweeping to identify any prey that had not been collected in the net. Our sample sizes for each dragonfly species varied based on natural abundances in ponds (E. simplicicollis: n = 20, L. pulchella: n = 32, P. lydia: n = 22, L. intacta: n = 32, S. vicinum, n = 45, E. canis: n = 20).

Prey response to temperature – To understand how the prey of dragonfly larvae respond to increasing temperatures, we conducted another experiment measuring the activity level of mosquito larvae at the same temperatures as the behavioural assays above. For detailed methods and results, please see Appendix II.

Relationship between species’ behavioural responses and maximum realized temperatures – We chose to examine the maximum temperatures naturally experienced (hereafter, maximum realized temperatures) by species across their distributions. This parameter is of interest because species’ realized thermal niches have been tightly correlated to physiological tolerance limits (Calosi et al., 2010; Sunday et al., 2011; Stuart-Smith et al., 2017) and acclimatory capacities (Calosi et al., 2008), which may be important in determining species’ responses to climate warming. To determine how maximum realized temperature is correlated to behavioural responses to temperature, we estimated the median maximum temperature of the warmest month experienced by species across their distributions and correlated this with their respective changes in behaviour from 28 to 32º C. Occurrence data for each species were retrieved from the Global Biodiversity Information Facility (http://www.gbif.org) through the R package dismo (version 1.1-4). Range-wide climate data for each species was extracted from Worldclim (Fick & Hijmans, 2017) in R using the raster package (version 2.5-8). We present the results for maximum realized temperature for each species in this paper, however, we also looked at the lowest and average temperatures naturally experienced by species across their distributions (Appendix II, Table S3.1). Results from those tests were qualitatively similar to the tests performed between maximum realized temperature and behavioral responses to temperature (see RESULTS). We therefore chose to focus the discussion only on the warmer temperatures within species’ ranges.

3.2.3 Study 2: Determining the impact of temperature on intraguild predation

Intraguild predation trials – To determine how IGP changes with increasing temperature, we measured how IGP frequency and the larval size distribution changed with temperature. We selected three species from the initial six, choosing species that differed most strongly in their behavioural responses to temperature, particularly in activity level (see RESULTS). Those species were L. intacta and P. lydia, representing species with marginal increases and decreases in activity, respectively, and E. simplicicollis, a species with no mean activity changes with temperature. The trial tanks were 20-L glass aquaria filled with aged tap water, aquarium gravel to line the bottom, and two weighted artificial plants per tank. We included one submersible water heater (Visi-Therm® Deluxe), attached along the bottom of each tank. We heated tanks to one of three temperatures: 23º, 28º and 32º C (means ± SD: 23.36 ± 0.15º, 27.96 ± 0.25º, and 31.54 ± 0.26º C). Data loggers measured temperature every hour (HOBO Pendant® Onset Computer Corporation, Cape Cod, Massachusetts, USA). Fluorescent strip lights above the tanks were set to 14 h light:10 h dark photoperiod. Before the trial began, we measured the head width of every individual. Head widths of larvae varied by species to reflect the natural size variation within ponds during the collection period (mean ± standard deviation: L. intacta: 3.95 ± 0.43 mm, E. simplicicollis: 3.82 ± 0.52 mm, P. lydia: 2.94 ± 0.38 mm). We used eight individuals of each species for a total starting abundance of 24 larvae per tank. Each larva was placed separately into a water-filled 120-mL plastic sample cup. Cups were sealed and floated within the treatment tank so larvae could acclimate individually and hunger levels would be standardized. After a 16 h acclimation period, we introduced the larvae into the tank with the others. We replicated each temperature treatment five times for a total of 15 tanks. The trials lasted for 72 h, after which we removed all surviving larvae and re-measured head widths in order to assess how predation shifted the size distribution of larvae within treatments.

Statistical analyses

3.3.1 Behavioural assays

We measured three different responses related to foraging ecology in the behavioural assays: feeding rate, activity level, and microhabitat preference. We quantified feeding rate as the

52 number of mosquito prey eaten by a larva divided by the time the larvae spent with the prey to get a rate per hour. Larvae will consume small prey whole and only living/moving prey (Pritchard, 1965). Therefore, we excluded any mosquito prey that died during the trial but not from apparent predation by dragonflies. This did not occur frequently, and we found no correlation between temperature and non-predator induced mosquito mortality (GLM: F1,169 = 0.44, P = 0.51). Head width was included as a covariate as it is highly correlated with gape-size – a common predictor of foraging effort in dragonflies (Wissinger 1992, for average head width of each species, see Appendix II, Fig. S3.1). We used a linear mixed model (LME) with the R package lme4 (version 1.1-14) to test for fixed effects of temperature, species, and head with, and interactions between temperature and species and temperature and head width, and a random effect of trial date on feeding rate of the dragonflies. We first created a global model with all parameters and possible interactions. Using the ‘dredge’ function from the R package MuMIn (Bartoń, 2017, version 1.40.0), we compared the fit of all candidate models using Akaike Information Criterion (AIC) and selected the model with the lowest AIC score as the best model, with a cut-off delta AIC value of 2. Estimated P-values of important factors and their interactions were obtained from type II Wald Chi-square tests with the function Anova from the package ‘car’ (version 2.1-6). This procedure was repeated for activity (total number of times a larva moved between squares within the grid during the observation period) using a GLMM with a Poisson error distribution, and for microhabitat preference (the proportion of time the larva was in contact with structure during the observation period) using a GLMM with a binomial error distribution.

3.3.2 Maximum realized temperature and behavioural responses to temperature

We used separate linear regressions to analyze the relationship between maximum realized temperature and species’ respective changes in feeding rate and number of moves from 28º to 32º C (from assays detailed above). We focused on the behavioural differences of these temperatures as we were interested in the conditions dragonfly larvae may experience in future, warmer summers.

3.3.3 Intraguild predation trials

First, we were interested in how rates of IGP changed with temperature and if there were certain species that were at a higher risk of being killed than others. Here, we define IGP as the killing and consuming of guild members, including conspecifics, i.e. cannibalism. Because our observations did not involve scoring the identity of the IG predator, we could not distinguish cannibalism events from mortality caused by interspecific predation. Both predation on other dragonfly species and cannibalism are common for larval dragonflies (Johansson, 1993). We used a GLMM with a binomial error distribution to test for the main effects of temperature, species, and initial head width, and the interactions between temperature and species, and temperature and initial head width, and a random effect of trial start date on the proportion of larvae killed during the trial. Initial head width was included as a predictor to distinguish the effects of variation in head width from other sources of variation due to species identity on observed predation rates. Again, we used the ‘dredge’ function (Bartoń, 2017) to select the best model based on AIC as described above. Next, we investigated how the size distribution of larvae shifted during the experiment in the different temperature treatments to understand how IGP altered the body size spectrum. The time frame of our experiment was short-term, so we did not expect larvae to grow during this period. Rather, we analyzed how the size-distribution of the surviving larvae compared to initial sizes in order to estimate which larval size classes were being killed and determine how size-biased predation changed with temperature. For this analysis, we used the difference in mean head width (final - initial) for each species as the response variable. We used a LME with temperature, species, and the interaction between temperature and species, and a random effect of trial start date to predict the difference in size distribution after IGP. All analyses were performed in R version 3.3.1 (R Development Core Team, 2016). Unless otherwise noted, results are presented as means ± standard errors.

Results

3.4.1 Temperature and foraging behaviour

The feeding rate of larval dragonflies significantly increased with temperature (24º C: 1.16 ± % 0.08 prey/h, 32º C: 1.65 ± 0.08 prey/h; LME: #$ = 32.99, P < 0.0001; Fig. 3.1a). Feeding rate % was positively related to head width (#$ =32.47, P < 0.0001), and feeding rates differed among

% species (#) = 29.06, P < 0.0001, Fig. 3.1a). There was a marginally significant interaction % between temperature and species on feeding rate (#) = 10.27, P = 0.07, Fig. 3.1a). Larval % dragonfly species differed in their levels of activity (GLMM: #) = 61.72, P < 0.0001; Fig. 3.1b), % however, we found no main effect of temperature on larval activity (#$ = 0.42, P = 0.52; Fig. % 3.1b), and a marginally significant interaction between temperature and species on activity (#) = 10.25, P = 0.07; Fig. 3.1b). Lastly, we found a significant interaction between temperature and % species on larval association with structure (#) = 64.22, P < 0.0001; Appendix II, fig. S3.2), with some species associating more with structure as temperature increased (E. simplicicollis and E. canis), while one species disassociated from structure more frequently with warming (S. vicinum; Fig. S3.2).

3.4.2 Changes in foraging were related to maximum realized temperatures

Species differed in their latitudinal range limits, and therefore also differed in their maximum realized temperatures experienced across their range (Appendix II, Table S3.1). Epitheca canis had the highest northern latitudinal range limit (53.07º N followed by L. intacta (50.57º N). Erythemis simplicicollis had the lowest northern latitudinal range limit (44.20º N). Maximum realized temperature was negatively associated with northern range limit (Pearson’s product moment correlation: t = -16.67, df = 4, P < 0.0001, r = -0.99), meaning that species with more northern range limits also had lower maximum realized temperatures and vice versa. We found a positive correlation between maximum realized temperature and change in feeding rate from 28º

2 to 32º C (F1,4 = 10.59, P = 0.03, R = 0.73; Fig. 3.2). In other words, species with higher maximum realized temperatures (e.g. E. simplicicollis and P. lydia; 32.2º and 30.3º C, respectively) increased their feeding rate to a greater degree compared to species with lower maximum realized temperatures (e.g. E. canis and L. intacta; 25.1º and 27.0º C; Fig. 3.2). Changes in activity, quantified as number of moves, were not significantly correlated with 2 maximum realized temperatures (F1,4 = 1.26, P = 0.32, R = 0.24; Appendix II, Fig. S3.3).

3.4.3 Temperature increased IGP and altered the size distribution of dragonfly larvae

Significantly more dragonfly larvae were killed as temperature increased (mean percentage % killed: 23º C:10.8 ± 2.7%, 28º C: 20.0 ± 4.2%, 32º C: 33.3 ± 5.1%; GLMM: #$ = 10.08, P = 0.001; Fig. 3.3), and there were marginally significant differences among species in the risk of being killed (P. lydia: 26.7 ± 4.5%, L. intacta: 20.0 ± 4.9%, E. simplicicollis: 17.5 ± 4.7%; % GLMM: #% = 5.32, P = 0.07; Fig. 3.3). We found a significant interaction between temperature and species, meaning that species had different risks of being killed depending on the thermal conditions. For instance, L. intacta experienced a sharper increase in mortality than the other two species from the coolest to warmest treatments (fold increase in proportion killed: L. intacta: 14- % fold, E. simplicicollis: 2-fold, P. lydia: 1-fold; GLMM: #% = 8.27, P = 0.016; Fig. 3.3).

Temperature significantly increased the difference in head widths across all species from the beginning to the end of the experiment (mean change in head width at the end of the experiment for each temperature: 23º C: 0.06 ± 0.03 mm, 28º C: 0.12 ± 0.04 mm, 32º C: 0.23 ± 0.05 mm;

LME F1,39 = 10.83, P = 0.002; Fig. 3.4), thereby shifting the distribution towards larger-bodied individuals (one-sided paired t-test: t = 2.66, df = 39, P = 0.006). We also found a marginally non-significant difference between species’ head widths before and after the experiment (P. lydia: 0.20 ± 0.04 mm, E. simplicicollis: 0.16 ± 0.05 mm, L. intacta: 0.06 ± 0.02 mm; LME: F2,39 = 3.17, P = 0.053; Fig. 3.4).

Discussion

We addressed a gap in the broad understanding of how climate change may affect species interactions. Predation rates increased at warmer temperatures and this was true whether the predation was on prey at a lower trophic level (mosquitoes) or on other guild members. However, a species’ size, not activity response to temperature, was more informative for predicting the outcome of those predatory interactions. Additionally, our study suggests that species with different latitudinal positions, and thus presumably different thermal adaptations, may differ in the plasticity of their foraging responses to warming.

3.5.1 Dragonfly larvae increased the number of prey eaten at warmer temperatures

Dragonfly larvae feeding rates increased on average by 42% between 24º and 32º C (Fig. 3.1a). Positive associations between feeding rate and temperature have also been observed in other omnivorous insects (e.g., Vucic-Pestic et al. 2011), herbivorous insects (e.g. Birkemoe et al. 2016), and in fish (e.g., Oyugi et al. 2012). This suggests that increasing temperatures could put greater pressure on prey or resources at lower trophic levels (Dell et al., 2014) – potentially leading to stronger top-down effects and influencing energy and nutrient cycling through food webs (Kratina et al., 2012). However, while most species increased feeding rate with temperature, two species (L. intacta and E. canis) appeared to reach their maximal feeding rate at intermediate temperatures (Fig. 3.1a). Differences in feeding rate at warmer temperatures could result from species-specific variation in the functional response (Thompson, 1978). Additionally, components of the functional response such as attack rate, handling time, and digestion rate can have different temperature dependencies (Englund et al., 2011; Sentis et al., 2013), and vary between species and trophic groups, potentially leading to variation in energetic efficiencies (Sentis et al., 2012). Decreased feeding rate could also be a stress-induced response (Quigley & Hinch, 2006) and differences in stress tolerance thresholds could be associated with species- specific thermal tolerances. Future work should investigate the mechanistic basis between species’ differences in foraging response to temperature to be able to predict how these species will be affected by rising temperatures.

3.5.2 Temperature did not change dragonfly larval activity levels or microhabitat preferences

We expected activity to increase with temperature, as some empirical evidence has shown (Gannon et al., 2014; Start et al., 2017), however, only L. intacta responded in this way. One species, P. lydia, slightly decreased activity with temperature, however, most species exhibited no observable activity change in response to warming (E. simplicicollis, L. pulchella, S. vicinum and E. canis; Fig. 3.1b). Higher activity levels in predators are often correlated with higher encounter rates with their prey, thus increasing chances of predation, however, encounter rates could also increase if the prey become more active in response to temperature. When we looked for this effect in the dragonfly’s prey, we found no significant impact of warming on mosquito activity when exposed to predation risk (for methods and results, see Appendix 2). While the mechanism underlying increased predation at higher temperatures in this system is unknown, it is clear temperature increases the strength of predation. This indicates that climate change may have a major impact on the strength of predator-prey interactions and potentially influence food web structure and energy flux through these systems (Shurin et al., 2012). Indeed, we see that feeding rates increase with temperature, but not necessarily due to observable changes in activity. While activity level is an important component of feeding, here it does not appear that activity level scales with temperature even though feeding rates does. Activity can also be influenced by other factors such as the threat of predation (McPeek, 1990a; Strobbe et al., 2011). Because our behavioural assays were performed on individual dragonfly larvae without the threat of predation, the behaviours observed are likely representative of the maximum levels of activity expression in these species. Future studies should consider the ecological contexts in which behaviour is measured in order to compare how multiple factors including temperature and predator presence affect activity (e.g. Smolinský and Gvoždík 2014).

We predicted that species would switch to more active foraging modes at higher temperatures, thus they would more frequently be found swimming through open water rather than clinging to structure. However, this was not true of all species; instead we found that warming had mixed effects on the likelihood of larvae being found on structure (Appendix II, Fig. S3.2). The species we examined differed in their microhabitat choices, a result supported by other odonate studies (Crowley & Johnson, 1982; McCauley, 2008). One species decreased association with structure

62 with warming which could have been to more actively pursue prey, however the majority of species either did not change or increased the time spent on structure (5 out of 6 species). This suggests that rising temperatures may impact where dragonflies choose to forage, and future studies should investigate whether this affects their overlap with prey as temperatures increase.

3.5.3 Maximum realized temperatures were associated with behavioural changes

We found that changes in feeding rate from 28 to 32º C were correlated with maximum realized temperature (Fig. 3.2). Specifically, species with higher maximum realized temperatures were most likely to increase their feeding rates at warmer temperatures (E. simplicicollis, P. lydia, L. pulchella, and S. vicinum, mean = + 46%), whereas species with lower maximum realized temperature actually decreased feeding rates in warmer conditions (L. intacta and E. canis, mean = - 14%). This could be attributed to these species’ differences in physiological acclimatization, which Calosi et al. (2008) demonstrated was greater in species with higher upper thermal limits. Thermal plasticity is a critical trait for species coping with climate change (Seebacher et al., 2014) and differences in behavioural responses may be indicative of species’ abilities to tolerate future warming. If species with lower upper temperature tolerances are more limited in feeding rate plasticity, they may be at a higher risk of extinction if feeding rate cannot keep up with metabolic demands. The scope of our study only included measuring behavioural responses in single populations with six species, but the correlations we found suggest that species’ maximum realized temperatures can predict foraging plasticity in response to temperature. Future studies should measure foraging responses across multiple populations and with more species to determine if this pattern is observed at a larger spatial scale. Additionally, measuring thermal tolerances of these species, especially at different parts within their range could determine if the mechanism underlying species’ differential foraging responses to temperature is related to physiological thermal tolerances.

3.5.4 Temperature increases rates of IGP in a species-specific way

We found that temperature significantly increased the frequency of IGP 2-fold between our coolest and warmest temperatures (Fig. 3.3). Theoretically encounter rates between individuals should increase with temperature (Dell et al., 2014), and drive higher rates of predation as

63 predators and prey encounter one another more frequently (Lima & Dill, 1990). However, the results of our activity level assays do not necessarily indicate that encounter rates increase with temperature. It is possible that the behaviours measured in the assays do not match with how larvae behaved in the IGP experiment, however we would expect that the presence of other guild members would lead to lower activity levels, and therefore would not explain why IGP increases with temperature. Presumably, instead of encounter rates increasing at warmer temperatures, the frequency of deadly encounters increased. This may be because increased metabolic and feeding rates at higher temperatures induce more defensive or predatory strikes when individuals encountered each other, or increased hunger levels caused individuals to make riskier decisions, resulting in greater exposure to IG predators (Lima & Dill, 1990).

We found a significant interaction between species and temperature on the frequency of IGP, meaning the proportion of individuals killed varied among species at different temperatures (Fig. 3.3). All species were killed at higher frequencies in the warmest temperatures however, L. intacta experienced the greatest change in mortality with a 14-fold increase from the coolest to warmest treatments (Fig. 3.3). Leucorrhinia intacta has the most northern distribution compared to the other species. Presumably, L. intacta, which also decreased feeding rate with temperature, was the most severely affected by warming because our highest temperature treatment exceeded its thermal optimum. While more work is needed to elucidate the mechanism driving these differences our results have interesting implications for species living close to their thermal margins. Species with high latitude ranges may have a reduced ability to tolerate increasing temperatures associated with climate change, whereas lower latitude species may be better able to handle warming and this could potentially explain why lower latitude species have expanded their ranges faster than higher latitude species since the 1970’s (Hickling et al., 2005). Our study sheds light on how plasticity of behavioural responses to warming in species with different latitudinal range positions may vary in predictable ways and suggests that these species may be differentially affected by warming, especially as novel species interactions arise due to range shifts induced by climate change.

An increase in molt frequency at higher temperatures could also result in increased mortality at higher temperatures. In general, temperature correlates positively with molt frequency in odonates (Corbet, 2002). In our IGP study, we observed individuals that molted more frequently

64 when in warmer temperatures (GLM: P = 0.004; Appendix II, Fig. S3.4). A freshly-molted individual may be at a higher risk of dying from attacks, either because of decreased activity following molting or because the unhardened exoskeleton right after molting makes them more vulnerable to attacks. In systems where species pass through multiple molts, e.g. many freshwater invertebrates, molt frequency could be an important factor in predator-prey interactions and its relationship with temperature should be explored further.

Killing events where dead larvae had fatal wounds but were not eaten, were most common in warmer conditions (29% and 20% at 28º and 32º, respectively, versus 7% at 23° C). Partial prey consumption can occur for several reasons, including when mismatched hunger signals of an empty midgut preclude food intake due to a lag in foregut emptying in arthropods (Johnson et al., 1975). Higher temperatures can increase the rate of gastric emptying (Van Damme et al., 1991), potentially increasing hunger levels in warmer conditions and leading to higher feeding rates as we observed in behavioural assays. Partial prey consumption may also increase if larvae become more defensive and attack more frequently at warmer temperatures when other larvae approach them. While the exact reason for increased partial prey consumption in our study is not yet understood, any increase to the number of killed and partially-consumed animals could have large-scale ecosystem effects by shunting energy away from higher trophic level consumers and instead into the detrital web (Kratina et al., 2012; Shurin et al., 2012).

3.5.5 Temperature increased the frequency of predation on smaller individuals

We found that temperature-induced increases in IGP led to changes in larval size distributions. At 23º C, mortality rates for E. simplicicollis, L. intacta, and P. lydia were low enough that the mean size distributions were not significantly altered across the 72-hour period (Fig. 3.4). However, at 28º and 32º C, predation on smaller individuals increased, shifting the size distributions of the three species towards larger-bodied individuals (Fig. 3.4). At 23º C, predation on the smallest species, P. lydia, was most common (Figs. 3.3 & 3.4), as is typical for populations with overlapping size/age classes (Robinson & Wellborn, 1987). However, at 32º C, we found that small larvae of all three species were eaten at significantly higher frequencies (Figs. 3.3 & 3.4). Our results suggest that the strength of IGP increases with temperature, with smaller individuals being at a higher risk of mortality by larger-bodied IG predators, regardless

65 of species identity. We speculate that this may have larger ecosystem consequences, given the importance of IGP in trophic webs. The potential increase in intensity of IGP could lead to decreases in larval abundance and put pressure on the individuals remaining to compensate for that guild’s function in the food web and potentially have a de-stabilizing effect on food webs (Wootton, 2017).

If warmer temperatures increase the vulnerability of smaller individuals to IGP, future climate warming could have major indirect effects on the size distributions of these populations and communities. Dragonfly larvae, like many ectotherms, typically have faster growth and developmental rates in warmer temperatures (Suhling et al., 2015; Frances et al., 2017). Presumably, these direct physiological effects of warming should lead to smaller size at emergence, yet this phenomenon known as the temperature-size-rule has not been detected in dragonflies (Hassall & Thompson, 2008; McCauley et al., 2015, 2018). We speculate that increasing strength and frequency of size-biased IGP should favour larger individuals, thus why the direct effects of warming on adult size may appear to absent from this taxonomic group. However, more experimental evidence is needed to understand how the direct and indirect effects of warming interact to influence the size structure of these ectotherms.

Conclusions

Our results demonstrate that ecologically similar species exhibit variable behavioural responses to temperature and these differences are strongly associated with species’ temperature preferences across their geographic range. However, behavioural responses alone were weak predictors of the outcome of IGP interactions and are therefore unlikely to effectively determine the “winner” and “loser” species of climate change (Zarnetske et al., 2012). Our study has implications for other systems in which size-structured predation and/or IGP are common. First, predation on lower trophic levels may increase as feeding rate is positively correlated with temperature, however species’ abilities to increase feeding rate as temperatures rise may be constrained by their thermal tolerance adaptations that are shaped by their climatic niches. This could be an important factor in predicting which species will perform better in warmer temperatures. Second, the strength of IGP may increase as temperatures rise, and potentially select for larger individuals and/or species. Future studies should examine how the impacts of these behavioural changes on prey can affect demographic, community, and ecosystem

66 processes. Such efforts would improve our knowledge of the indirect effects of climate change on communities.

Acknowledgements:

We would like to thank R. Mitchell and J. Moon for their research assistance, and the Koffler Scientific Reserve and S. Schneider for research infrastructure support. We are also thankful for the thoughtful suggestions made by two anonymous reviewers. Funding to DNF was provided by Zimmerman and Weis Graduate Scholarship by the Department of EEB at University of Toronto. Funding to SJM was provided by an NSERC Discovery Grant (# 435614), CFI John R. Evans Leaders Fund, and Ontario Leaders Opportunity Fund.

References Addo-Bediako A, Chown SL, Gaston KJ (2000) Thermal tolerance, climatic variability and latitude. Proc R Soc Lond B 267:739–745. https://doi.org/10.1098/rspb.2000.1065 Angilletta MJ, Dunham AE (2003) The temperature-size rule in ectotherms: simple evolutionary explanations may not be general. Am Nat 162:332–342. https://doi.org/10.1086/377187 Arim M, Marquet PA (2004) Intraguild predation: a widespread interaction related to species biology. Ecol Lett 7:557–564. https://doi. org/10.1111/j.1461-0248.2004.00613.x Bartoń, K. (2017). MuMIn: multi-model inference. R Package Version 1.40.0. Retrieved from https://cran.r-project.org/package=MuMIn Barton BT, Schmitz OJ (2009) Experimental warming transforms multiple predator effects in a grassland food web. Ecol Lett 12(12):1317–1325 Benke AC (1970) A method for comparing individual growth rates of aquatic insects with special reference to the odonata. Ecology 51:328–331 Birkemoe T, Bergmann S, Hasle TE, Klanderud K (2016) Experimental warming increases herbivory by leaf-chewing insects in an alpine plant community. Ecol Evol 6:6955–6962. https://doi. org/10.1002/ece3.2398 Bush EJ, Loder JW, James TS et al (2014) An overview of Canada’s changing climate. In: Warren FJ, Lemmen DS (eds) Canada in a changing climate: sector perspectives on impacts and adaptation. Government of Canada, Ottawa, pp 23–64 Calosi P, Bilton DT, Spicer JI (2008) Thermal tolerance, acclimatory capacity and vulnerability to global climate change. Biol Lett 4:99–102. https://doi.org/10.1098/rsbl.2007.0408 Calosi P, Bilton DT, Spicer JI et al (2010) What determines a species’ geographical range? Thermal biology and latitudinal range size relationships in European diving beetles (Coleoptera: Dytiscidae). J Anim Ecol 79:194–204. https://doi.org/10.111 1/j.1365- 2656.2009.01611.x

Connell JH (1961) The influence of interspecific competition and other factors on the distribution of the barnacle Chthalamus stellatus. Ecology 42:710–723 Corbet PS (2002) Stadia and growth ratios of Odonata: a review. Int J Odonatol 5:45–73. https://doi.org/10.1080/13887890.2002.97481 76 Crowley PH, Johnson DM (1982) Habitat and seasonality as niche axes in an odonate community. Ecology 63:1064–1077 Crumrine PW (2010) Body size, temperature, and seasonal differences in size structure influence the occurrence of cannibalism in larvae of the migratory dragonfly, Anax junius. Aquat Ecol 44:761–770. https://doi.org/10.1007/s10452-010-9314-z Dell AI, Pawar S, Savage VM (2011) Systematic variation in the temperature dependence of physiological and ecological traits. Proc Natl Acad Sci USA 108:10591–10596. https://doi.org/10.1073/ pnas.1015178108 Dell AI, Pawar S, Savage VM (2014) Temperature dependence of trophic interactions are driven by asymmetry of species responses and foraging strategy. J Anim Ecol 83:70–84. https:// doi.org/10.1111/1365-2656.12081 Dossena M, Yvon-Durocher G, Grey J et al (2012) Warming alters community size structure and ecosystem functioning. Proc R Soc Lond B 279:3011–3019. https://doi.org/10.1098/rspb.2012.0394 Elliott JM (1976) The energetics of feeding, metabolism and growth of brown trout (Salmo trutta L.) in relation to body weight, water temperature and ration size. J Anim Ecol 45:923– 948 Englund G, Öhlund G, Hein CL, Diehl S (2011) Temperature dependence of the functional response. Ecol Lett 14:914–921. https://doi. org/10.1111/j.1461-0248.2011.01661.x Fick SE, Hijmans RJ (2017) Worldclim 2: new 1-km spatial resolution climate surfaces for global land areas. Int J Climatol. https://doi. org/10.1002/joc.5086 Flenner I, Richter O, Suhling F (2010) Rising temperature and development in dragonfly populations at different latitudes. Freshw Biol 55:397–410. https://doi.org/10.111 1/j.1365-2427.2009.02289.x Frances DN, Moon J, McCauley SJ (2017) Effects of environmental warming during early life- history on libellulid odonates. Can J Zool 95:373–382. https://doi.org/10.1139/cjz-2016- 0233 Frazier MR, Huey RB, Berrigan D (2006) Thermodynamics constrains the evolution of insect population growth rates: “warmer is better”. Am Nat 168:512–520. https://doi.org/10.1086/506977 Gannon R, Taylor M, Suthers I et al (2014) Thermal limitation of performance and biogeography in a free-ranging ectotherm: insights from accelerometry. J Exp Biol 217:3033–3037. https ://doi.org/10.1242/jeb.104455 Gillooly JF, Brown JH, West GB et al (2001) Effects of size and temperature on metabolic rate. Science 80(293):2248–2251. https ://doi.org/10.1126/science.1061967

Gillooly JF, Charnov EL, West GB et al (2002) Effects of size and temperature on developmental time. Oikos 417:70–74 Hassall C, Thompson DJ (2008) The effects of environmental warming on Odonata: a review. Int J Odonatol 11:131–153. https:// doi.org/10.1080/13887890.2008.9748319 Hickling R, Roy DB, Hill JK, Thomas CD (2005) A northward shift of range margins in British Odonata. Glob Chang Biol 11:502– 506. https://doi.org/10.1111/j.1365- 2486.2005.00904.x Hopper KR, Crowley PH, Kielman D (1996) Density dependence, hatching synchrony, and within-cohort cannibalism in young dragonfly larvae. Ecology 77:191–200. https://doi. org/10.2307/2265668 Huey RB, Stevenson RD (1979) Intergrating thermal physiology and ecology of ectotherms: a discussion of approaches. Am Zool 19:357–366 IPCC (2014) Fifth assessment report—synthesis report. IPCC, Geneva Johansson F (1991) Foraging modes in an assemblage of odonate larvae—effects of prey and interference. Hydrobiologia 209:79–87. https://doi.org/10.1007/BF00006721 Johansson F (1993) Intraguild predation and cannibalism in odonate larvae—effects of foraging behavior and zooplankton availability. Oikos 66:80–87. https://doi.org/10.2307/3545198 Johansson F (2000) The slow-fast life style characteristics in a suite of six species of odonate larvae. Freshw Biol 43:149–159. https:// doi.org/10.1046/j.1365-2427.2000.00532.x Johnson DM, Akre BG, Crowley PH (1975) Modeling arthropod predation: wasteful killing by damselfly naiads. Ecology 56:1081–1093 Kingsolver JG, Huey RB (2008) Size, temperature, and fitness: three rules. Evol Ecol Res 10:251–268 Kratina P, Greig HS, Thompson PL et al (2012) Warming modifies trophic cascades and eutrophication in experimental freshwater communities. Ecology 93:1421–1430. https://doi. org/10.1890/11-1595.1 Lima SL, Dill LM (1990) Behavioral decisions made under the risk of predation: a review and prospectus. Can J Zool 68:619–640. https ://doi.org/10.1139/z90-092 McCauley SJ (2008) Slow, fast and in between: habitat distribution and behaviour of larvae in nine species of libellulid dragonfly. Freshw Biol 53:253–263. https://doi.org/10.1111/j.1365-2427.2007.01889 .x McCauley SJ, Hammond JI, Frances DN, Mabry KE (2015) Effects of experimental warming on survival, phenology, and morphology of an aquatic insect (Odonata). Ecol Entomol 40:211–220. https ://doi.org/10.1111/een.12175 McCauley SJ, Hammond JI, Mabry KE (2018) Simulated climate change increases larval mortality, alters phenology, and affects flight morphology of a dragonfly. Ecosphere 9:e02151. https:// doi.org/10.1002/ecs2.2151 McPeek MA (1990) Behavioral differences between Enallagma species (Odonata) influencing differential vulnerability to predators. Ecology 71:1714–1726. https://doi.org/10.2307/1937580

O’Connor MI (2009) Warming strengthens an herbivore: plant interaction. Ecology 90:388–398. https://doi.org/10.1890/08-0034.1 Oyugi DO, Cucherousset J, Baker DJ, Britton JR (2012) Effects of temperature on the foraging and growth rate of juvenile common carp, Cyprinus carpio. J Therm Biol 37:89–94. https://doi. org/10.1016/j.jtherbio.2011.11.005 Polis GA, Myers CA, Holt RD (1989) The ecology and evolution of intraguild predation: potential competitors that eat each other. Annu Rev Ecol Syst 20:297–330 Pritchard G (1965) Prey capture by dragonfly larvae (Odonata; Anisoptera). Can J Zool 43:271– 289. https://doi.org/10.1139/z65-026 Quigley JT, Hinch SG (2006) Effects of rapid experimental temperature increases on acute physiological stress and behaviour of stream dwelling juvenile Chinook salmon. J Therm Biol 31:429–441. https://doi.org/10.1016/j.jtherbio.2006.02.003 R Core Team (2016) R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria. https://www.R-project.org/ Rall BC, Vucic-Pestic O, Ehnes RB et al (2010) Temperature, predator- prey interaction strength and population stability. Glob Chang Biol 16:2145–2157. https://doi.org/10.1111/j.1365- 2486.2009.02124.x Robinson JV, Wellborn GA (1987) Mutual predation in assembled communities of odonate species. Ecology 68:921–927. https:// doi.org/10.2307/1938363 Rudolf VHW, Singh M (2013) Disentangling climate change effects on species interactions: effects of temperature, phenological shifts, and body size. Oecologia 173:1043–1052. https://doi.org/10.1007/ s00442-013-2675-y Seebacher F, White CR, Franklin CE (2014) Physiological plasticity increases resilience of ectothermic animals to climate change. Nat Clim Chang 5:61–66. https://doi.org/10.1038/nclimate2457 Sentis A, Hemptinne J-L, Brodeur J (2012) Using functional response modeling to investigate the effect of temperature on predator feeding rate and energetic efficiency. Oecologia 169:1117–1125. https ://doi.org/10.1007/s00442-012-2255-6 Sentis A, Hemptinne JL, Brodeur J (2013) Parsing handling time into its components: implications for responses to a temperature gradient. Ecology 94:1675–1680. https://doi.org/10.1890/12-2107.1 Sentis A, Hemptinne JL, Brodeur J (2014) Towards a mechanistic understanding of temperature and enrichment effects on species interaction strength, omnivory and food-web structure. Ecol Lett 17:785–793. https://doi.org/10.1111/ele.12281 Shurin JB, Clasen JL, Greig HS et al (2012) Warming shifts top-down and bottom-up control of pond food web structure and function. Philos Trans R Soc Lond B 367:3008–3017. https://doi. org/10.1098/rstb.2012.0243 Smolinský R, Gvoždík L (2014) Effect of temperature extremes on the spatial dynamics of predator-prey interactions: a case study with dragonfly nymphs and newt larvae. J Therm Biol 39:12–16. https ://doi.org/10.1016/j.jtherbio.2013.11.004

Start D, Kirk D, Shea D, Gilbert B (2017) Cannibalism by damselflies increases with rising temperature. Biol Lett. https://doi. org/10.1098/rsbl.2017.0175 Strobbe F, McPeek MA, De Block M, Stoks R (2011) Fish predation selects for reduced foraging activity. Behav Ecol Sociobiol 65:241–247. https://doi.org/10.1007/s00265-010-1032-y Stuart-Smith RD, Edgar GJ, Bates AE (2017) Thermal limits to the geographic distributions of shallow-water marine species. Nat Ecol Evol 1:1846–1852. https://doi.org/10.1038/s4155 9-017-0353-x Suhling F, Suhling I, Richter O (2015) Temperature response of growth of larval dragonflies— an overview. Int J Odonatol 18:15–30. https ://doi.org/10.1080/13887890.2015.1009392 Sunday JM, Bates AE, Dulvy NK (2011) Global analysis of thermal tolerance and latitude in ectotherms. Proc R Soc Lond B 278:1823–1830. https://doi.org/10.1098/rspb.2010.1295 Thompson DJ (1978) Towards a realistic predator-prey model: the effect of temperature on the functional response and life history of larvae of the damselfly, Ischnura elegans. J Anim Ecol 47:757–767 Van Damme R, Bauwens D, Verheyen RF (1991) The thermal dependence of feeding behaviour, food consumption and gut-passage time in the lizard Lacerta vivipara Jacquin. Funct Ecol 5:507–517 Vasseur DA, McCann KS (2005) A mechanistic approach for modeling temperature dependent consumer resource dynamics. Am Nat 166:184–198. https://doi.org/10.1086/431285 Vucic-Pestic O, Ehnes RB, Rall BC, Brose U (2011) Warming up the system: higher predator feeding rates but lower energetic efficiencies. Glob Chang Biol 17:1301–1310. https://doi.org/10.1111/j.1365-2486.2010.02329.x Werner EE, Anholt BR (1993) Ecological consequences of the trade-off between growth and mortality rates mediated by foraging activity. Am Nat 142:242–272 Wissinger SA (1992) Niche overlap and the potential for competition and intraguild predation between size-structured populations. Ecology 73:1431–1444. https://doi.org/10.2307/1940688 Wootton KL (2017) Omnivory and stability in freshwater habitats: does theory match reality? Freshw Biol 62:821–832. https://doi. org/10.1111/fwb.12908 Zarnetske PL, Skelly DK, Urban MC (2012) Biotic multipliers of climate change. Science 80(336):1516–1518. https://doi. org/10.1126/science.1223365

How will prey respond to predation risk in a warmer world?

I have written this chapter to be submitted to Biology Letters. With the exception of some of the methods, which will be shifted to the supplementary materials to accommodate the length restrictions of this journal, I intend to publish this chapter as: Frances, D.N. and S.J. McCauley. How will prey respond to predation risk in a warmer world?

4 Abstract

Prey often decrease activity levels in accordance with the threat of predation. Yet, this reduction in activity often comes at the cost of missing out on foraging opportunities and temporarily forfeiting growth. While this response to predation risk is well documented, there are few data on how predation risk interacts with other stressors in shaping prey responses. This is particularly important when additional stressors induce opposing effects on activity, such as high temperatures, which typically increase metabolism, requiring higher activity levels to meet foraging demands. We assessed how activity is affected by the combination of warming and predation risk using larval damselflies. We found that prey generally responded to temperature more strongly than to predation risk by increasing activity, even though their risk of being eaten increased at warmer temperatures. Interestingly, warming and predation risk had cascading effects on damselfly prey, in that damselflies became more accurate and efficient at feeding on zooplankton when predation threat and temperatures were high. These findings suggest that predation risk and temperature may interact to affect the vigilance and foraging strategies of damselflies, thereby affecting trophic cascades within these communities.

Introduction

The mere presence of predators can induce a variety of physiological, developmental, morphological, and/or behavioural changes in their prey (Sih 1980; Lima & Dill, 1990; Werner & Anholt, 1993; Werner & Peacor, 2003). These phenotypic traits, while effective in minimizing the risk of being eaten, are energetically costly for prey to produce or maintain (Preisser et al., 2005). To date, little is known about how the effects of climate warming will interact with these non-consumptive effects on the ecological responses of prey. This knowledge gap is especially critical for assumptions of the growth/predation risk trade-off (Lima & Dill, 1990). This model

72 predicts that prey should modify their behaviour (e.g. activity level) in accordance with perceived predation risk (Werner & Anholt, 1993). When perceived risk is high, prey should become less conspicuous to predators by reducing activity (Lima & Dill, 1990; Werner & Anholt, 1993). However, this reduction comes at the cost of slower growth rates and longer developmental periods – both of which can impact fitness (McPeek & Peckarsky, 1998). Prey must balance this trade-off to reduce their likelihood of being eaten while maintaining energetic gains necessary for growth and development.

Warmer temperatures such as those associated with climate change are expected to increase metabolic rates, and therefore energetic demands of many organisms (Elliott, 1976). Non-sessile consumers should then increase activity level, in particular foraging activity, to be able to sustain higher energetic demands (Dell et al., 2014). Models that account for the temperature- dependence of metabolic and ecological rates (e.g. sprint speed, encounter rates, etc.) predict that the strength of trophic interactions, such as predation, will increase with warming (Gilbert et al., 2014). Theoretically, prey should increasingly forgo foraging opportunities at higher temperatures to minimize their risk of predation, although temperature-induced increases in metabolic rates may make this especially costly in these conditions. Evidence from the physiological literature suggests this may be true given that traits associated with escape are under different thermal constraints than attack rates (Dell et al., 2011). Yet, empirical data that examine the interactive effects of warming and predation risk are mixed. In an aquatic system, damselfly larvae increased consumption rates with rising temperatures even when predators were present (Culler et al., 2014). However, in a rocky intertidal and stream system, the combinations of predation risk and temperature had additive non-consumptive effects on prey (Hammock & Johnson, 2014; Miller et al., 2014). Therefore, it still remains unclear how prey behaviourally respond to both predation risk and warming as these studies generally fail to directly observe activity in response to these interacting effects. As a result, the critical link in the growth/predation risk trade-off, activity, has not been explicitly examined with the additional temperature interaction included.

We tested for the combined effects of warming and predation risk on activity in larval Enallagma boreale (Selys, 1875) damselflies. We predicted that (i) warming will increase activity level, especially foraging activity, and (ii) predation risk should suppress prey activity,

73 even in warmer temperatures. Alternatively, if the energetic costs associated with reducing activity in warmer conditions are too great, activity will increase with temperature even when predation risk is high.

Materials and methods

Larval Enallagma boreale (hereafter Enallagma) are found in fishless lakes and ponds across northeastern North America where large-bodied dragonfly larvae such as Anax junius (Drury, 1773) are top predators (McPeek, 1990b). Prior work has demonstrated that this species reduces conspicuous behaviours in the presence of Anax junius (hereafter Anax; McPeek, 1990b). We collected Enallagma and Anax from a fishless pond at the Koffler Scientific Reserve in Ontario, Canada in September 2017. Larvae were transferred to the laboratory at the University of Toronto Mississauga. Enallagma were divided between three glass aquaria filled with dechlorinated tap water and fed with locally collected zooplankton (Daphnia spp.) which were replenished daily so damselflies could feed ad libitum. To reduce the risk of cannibalism in these holding tanks, we also added charophyte algae and mesh screening to each tank to increase refuge availability and decrease encounter rates. Water in each aquarium was oxygenated via an aquarium bubbler with an air-stone to diffuse air into the water. All Enallagma and their predators experienced a 14:10 L:D photoperiod.

Behavioural assays were performed in glass aquaria (9.5 L) filled with dechlorinated and oxygenated tap water. To each one, we added aquarium gravel to cover the bottom, one weighted, plastic aquarium plant, and a weighted datalogger that recorded water temperature every 15-minutes (HOBO Pendant® UA-001-08, Onset Computer Corporation). On the back and sides of each aquarium, we attached a paper grid that visually sub-divided the aquarium into 32 blocks (each block: 7.6 x 5.0 cm3). This standard method allowed us to quantify the damselfly’s position and behaviour in 3-dimensions while also preventing it from seeing conspecifics or predators in adjacent aquaria (Johansson, 2000). Using a fully-crossed design, we manipulated water temperature (mean ± SD, control: 20.74 ± 0.27º C, and warmed: 27.28 ± 0.68º C) with aquarium water heaters (Eheim Jager TruTemp Submersible Heater 50 W), and predator presence (caged Anax or empty cage). Control temperatures reflect mean water temperatures this species experiences during summer larval development; warmed treatments represent predicted

74 temperature for the region by the end of the century (Environment and Climate Change Canada, 2016). Aquaria were randomly assigned a treatment but we did not switch predator treatments between replicates so that residual kairomones, i.e. chemical alarm cues, were not carried over to no-predator treatment on another day. Each treatment was replicated six times over two days and individuals were only used once.

We placed individual Enallagma into aquaria 17 h before the start of the assay to acclimate and standardize hunger levels among individuals. Before the start of each assay, we placed a single Anax (including the water it was housed in) into a plastic cage (350-mL weighted, plastic container) and fed each one an Enallagma to release conspecific kairomones (Hopper, 2001; Stoks et al., 2003). We then introduced the caged Anax predator into the assay aquarium. Mesh screening covering the top of the predator’s cage allowed the focal individual to respond to both olfactory and visual cues of the predator (McPeek, 1990a). In the no-predator treatments, a plastic cage filled with water and covered in mesh screening was added before the start of the assay. We then added 20 cladoceran zooplankton to the assay aquarium and began observing Enallagma behaviour after 5 min. A single observer (DNF) scored the activity of each individual continuously for 30 min. Similar to McPeek (1990b), we observed seven activities: crawling/advances towards prey, turning, swimming, abdominal bending, striking at prey (unsuccessful feeding attempts), eating prey (successful feeding attempts), and total distance traveled (number of blocks an individual passed through) during the observation period. Behavioural analyses took place on 3 and 5 October 2017.

In a separate experiment, we tested whether Anax predators increase predation rates on Enallagma at higher temperatures to determine if Enallagma predation risk increases with temperature. These experiments were conducted in aquaria similar to the set-up described above and at the same temperatures. For details of these assays including statistical analyses, please see the Appendix.

4.2.1 Statistical analyses

We excluded swimming and abdominal bending from analyses as these behaviours occurred infrequently (< 3 times total). We performed separate generalized linear mixed models (GLMM) on (i) the total number of feeding attempts; (ii) the proportion of successful feeding attempts;

75 number of times an individual (iii) crawled, (iv) and turned; and (v) the total distance moved with the function lmer from the R package ‘lme4’ (version 1.1-14). We included temperature (continuous), predator presence (binary), and their interaction as fixed effects. Trial date was included as a random effect only when log-likelihood ratio tests indicated significant differences between models with and without the effect. In all GLMMs, a Poisson distribution was used, as these were count data, except the model for the proportion of successful feeding attempts, in which a binomial distribution was used. To control for differences in behaviour due to size, individual wet mass (mg) was included as a fixed effect. P-values of fixed factors and their interactions were approximated through type II Wald Chi-square tests with the Anova function from the ‘car’ package (version 2.1.6). All analyses were performed in R (version 3.3.2).

Results

For many activities, warming combined with predation risk generally had little effect on Enallagma. Instead, temperature alone most strongly influence Enallagma activity, with higher temperatures significantly increasing the proportion of successful feeding attempts (temperature: % % ($ = 7.62, P = 0.006; figure 4.1b), and the frequency of crawling (($ = 9.64, P = 0.002; figure % 4.2b), Warmer temperatures also marginally increased total feeding attempts (#$ = 3.03, P = % 0.08; figure 4.1a) and the frequency of turning (($ = 3.13, P = 0.077; figure 4.2a). Interestingly, predation risk did affect one aspect of behaviour in an unexpected way: Enallagma exposed to predation risk increased their proportion of successful feeding attempts by 325% compared with % Enallagma not exposed to predators (predator treatment: ($ = 7.66, P = 0.006; figure 4.1b). Predation risk also had an effect on the frequency of crawling but in the opposite direction predicted, i.e., the frequency of crawling increased by 125% compared to no-predator treatments % (($ = 3.77, P = 0.052; figure 4.2b). Lastly, Enallagma moved longer distances in warmer % temperatures, but only when predation risk was absent (temperature × predator interaction: ($ = 6.31 P = 0.01; figure 2c).

In the second experiment assessing risk of predation at higher temperatures, Anax increased predation rate on Enallagma by 33% at higher temperatures, however, this effect was only marginally significant, potentially due to the large variation in individual Anax responses % (temperature: ($ = 2.71, P = 0.10; see Appendix III for Fig. S4.2, methods, and all results).

Figure 4.1. The effects of temperature and predation risk on Enallagma boreale damselfly larvae total number of feeding attempts (a), and the proportion of total feeding attempts that were successful (b). Blue and red colours represent control (21º) and warmed (27º C) treatments, respectively. “No predator” treatments indicate no predation risk (empty cage) and “Predator” treatments indicate caged (non-lethal) Anax predator was present. Bars display means + 1 SE

Figure 4.2. The effects of temperature and predation risk on larval Enallagma damselfly conspicuous activities. Temperature increased frequency of turning (a) and crawling/stalking (b). The combined effect of predation risk and warming significantly increased distance moved (c), measured as the number of squares the damselfly moved through during the assay. Blue and red colours represent control (21º) and warmed (27º C) treatments, respectively. “No predator” treatments indicate no predation risk (empty cage) and “Predator” treatments indicate caged (non-lethal) Anax predator was present. Bars display means + 1 SE

Discussion Based on theory of how organisms should balance the growth/predation risk trade-off (e.g. Sih, 1987; Lima & Dill, 1990; Werner & Anholt, 1993), we expected that larval damselflies would reduce activity in the presence of predators to become less conspicuous, even at the cost of foregoing feeding opportunities when conditions were warmer. Instead, we found that larval damselflies increased certain activities in warmer conditions, despite the presence of predators (figures 4.1 and 4.2). Additionally, prey became more effective predators, increasing the accuracy of their feeding attempts when predators were present, especially when temperatures were warmer. Our results show that prey respond more strongly to temperature than predation risk for many behaviours, indicating that prey may be weighing these risks differently in warmer temperatures.

In previous work that examined warming and predation risk, grasshoppers appeared to not sacrifice feeding opportunities at warmer temperatures when their predators were present (Barton et al., 2009). Instead, predator presence and warming induced diet shifts, wherein grasshoppers shifted from eating grasses to forbs, potentially as a way to decrease spatial overlap with predators while maintaining increased energetic demands associated with those conditions (Barton et al., 2009). In the damselfly Enallagma vesperum, larvae increased consumption rates with rising temperatures, however this effect was magnified when their fish predators were present (Culler et al., 2014). Furthermore, both of these studies found that higher feeding rates did not translate to increased prey biomass or growth rates, suggesting that the energetic cost of forgoing foraging opportunities at higher temperatures may outweigh the risk that predators pose. Specifically, at high temperatures, maximal consumption rates cannot keep pace with rising energetic costs associated with faster metabolism, and consequently places ectotherms at risk of starvation in these conditions (Lemoine & Burkepile, 2012). While these experiments did not explicitly measure prey activity, their results and ours demonstrate that prey are not responding to predation risk by down-regulating activity and instead, continue to feed even when predators are present. This could be costly, given that our study and Crumrine (2010) found that predation rates by Anax increased when temperatures were higher, which aligns with theoretical and empirical work in many different systems showing that predation risk increases with warming (Vasseur & McCann, 2005; Sentis et al., 2012; Gilbert et al., 2014).

Activity reductions are one form of anti-predator behaviour, however, prey can also become more vigilant in the presence of predators. Increased vigilance may reduce predation risk, but generally at the cost of reducing feeding rate (Chivers & Smith, 1998; Brown, 1999). However, we found that Enallagma had more successful feeding attempts (figure 4.1) and increased the frequency of crawling when predators were present (figure 4.2b). These responses suggest that Enallagma become more vigilant in the presence of predators by reducing bold, conspicuous behaviours and only eating when their prey are within striking distance. An increase in feeding accuracy may be a strategy that organisms use to reduce their chance of starving in warmer conditions, while still remaining vigilant around predators. Additionally, the interaction between warming and predation risk could then have cascading effects, where organisms at lower trophic levels, such as primary consumers, are indirectly suppressed by the increasing foraging efficacy of their consumers.

Prior work has found that prey may be able to adaptively moderate their activity and foraging levels when exposed to novel predators (Urban, 2007), and there is some evidence that the presence of predators can act synergistically with temperature to modify the evolutionary trajectories of prey (Tseng & O’Connor, 2015). Future work should examine how prey adapt to warmer conditions as their risk of being eaten increases. While the scope of our study did not extend to observing differential growth, our results suggest that predation risk may increase in warmer conditions, and affect the vigilance and foraging strategies of Enallagma, in turn, potentially altering trophic cascades within this system. We suggest that future work explicitly examine how behavioural decisions made under the threat of predation are affected by increasing temperatures, specifically, how animals balance costs and benefits associated with maintaining or increasing certain activities in warmer conditions, especially when predation risk is higher.

Acknowledgements We thank members of the lab and M. Johnson who gave insightful comments in the preparation and writing of this manuscript. This work was supported by a NSERC (Canada) grant (RGPIN 435614) to SJM.

References Barton BT, Beckerman AP, Schmitz OJ (2009) Climate warming strengthens indirect interactions in an old-field food web. Ecology, 90, 2346–2351. Brown JS (1999) Vigilance, patch use and habitat selection: Foraging under predation risk.

Evolutionary Ecology Research, 1, 49–71. Chivers DP, Smith RJF (1998) Chemical alarm signalling in aquatic predator-prey systems: a review and prospectus. Ecoscience, 5, 338–352. Crumrine PW (2010) Body size, temperature, and seasonal differences in size structure influence the occurrence of cannibalism in larvae of the migratory dragonfly, Anax junius. Aquatic Ecology, 44, 761–770. Culler LE, McPeek MA, Ayres MP (2014) Predation risk shapes thermal physiology of a predaceous damselfly. Oecologia, 176, 653–660. Dell AI, Pawar S, Savage VM (2011) Systematic variation in the temperature dependence of physiological and ecological traits. Proceedings of the National Academy of Sciences of the United States of America, 108, 10591–10596. Dell AI, Pawar S, Savage VM (2014) Temperature dependence of trophic interactions are driven by asymmetry of species responses and foraging strategy. Journal of Animal Ecology, 83, 70–84. Elliott JM (1976) The energetics of feeding, metabolism and growth of brown trout (Salmo trutta L.) in relation to body weight, water temperature and ration size. Journal of Animal Ecology, 45, 923–948. Environment and Climate Change Canada (2016) Climate data and scenarios for Canada : Synthesis of recent observation and modelling results. Gatineau, QC, 1-28 pp. Gilbert B, Tunney TD, Mccann KS et al. (2014) A bioenergetic framework for the temperature dependence of trophic interactions. Ecology Letters, 17, 902–914. Hammock BG, Johnson ML (2014) Trout reverse the effect of water temperature on the foraging of a mayfly. Oecologia, 175, 997–1003. Hopper KR (2001) Flexible antipredator behavior in a dragonfly species that coexists with different predator. Oikos, 93, 470–476. Johansson F (2000) The slow-fast life style characteristics in a suite of six species of odonate larvae. Freshwater Biology, 43, 149–159. Lemoine NP, Burkepile DE (2012) Temperature-induced mismatches between consumption and metabolism reduce consumer fitness. Ecology, 93, 2483–2489. Lima SL, Dill LM (1990) Behavioral decisions made under the risk of predation: a review and prospectus. Canadian Journal of Zoology, 68, 619–640. McPeek MA (1990a) Determination of species composition in the Enallagma damselfly assemblages of permanent lakes. Ecology, 71, 83–98. McPeek MA (1990b) Behavioral differences between Enallagma species (Odonata) influencing differential vulnerability to predators. Ecology, 71, 1714–1726. McPeek MA, Peckarsky BL (1998) Life histories and the strength of species interactions: combining mortality, growth, and fecundity effects. Ecology, 79, 867–879. Miller LP, Matassa CM, Trussell GC (2014) Climate change enhances the negative effects of predation risk on an intermediate consumer. Global Change Biology, 20, 3834–3844.

Preisser EL, Bolnick DI, Benard MF (2005) Scared to death? The effects of intimidation and consumption in predator-prey interactions. Ecology, 86, 501–509. Sentis A, Hemptinne J-L, Brodeur J (2012) Using functional response modeling to investigate the effect of temperature on predator feeding rate and energetic efficiency. Oecologia, 169, 1117–1125. Sih, A. (1980). Optimal behavior: can foragers balance two conflicting demands. Science, 210, 1041–1043. Sih, A. (1987). Predators and prey lifestyles: an evolutionary and ecological overview. In Predation. Direct and indirect impacts on aquatic communities (pp. 203–224). Stoks R, McPeek MA, Mitchell JL (2003) Evolution of prey behavior in response to changes in predation regime: damselflies in fish and dragonfly lakes. Evolution, 57, 574–585. Tseng M, O’Connor MI (2015) Predators modify the evolutionary response of prey to temperature change. Biology Letters, 11, 20150798. Urban MC (2007) Risky prey behavior evolves in risky habitats. Proceedings of the National Academy of Sciences, 104, 14377–14382. Vasseur DA, McCann KS (2005) A mechanistic approach for modeling temperature dependent consumer resource dynamics. The American Naturalist, 166, 184–198. Werner EE, Anholt BR (1993) Ecological consequences of the trade-off between growth and mortality rates mediated by foraging activity. The American Naturalist, 142, 242–272. Werner EE, Peacor SD (2003) A review of trait-mediated indirect interactions in ecological communities. Ecology, 84, 1083–1100.

Experimental warming and heat waves modify pond community composition and trophic dynamics

This paper is in preparation to be published as: Frances, D.N., J.I. Hammond, K.E. Mabry, and S.J. McCauley. Experimental warming and heat waves modify pond community composition and trophic dynamics.

5 Abstract

Warming temperatures due to recent climate change have had large impacts on the composition, dynamics, and size structure of ecological communities. Along with warming, extreme climate events such as heat waves are becoming more severe, yet we lack an understanding of how communities will be respond to future extreme events. Especially in shallow aquatic habitats, where water temperature tracks air temperature, our understanding of how these communities will be affected by warming as well as increased frequency of heat waves is limited. In a pond mesocosm study, we simulated realistic future warming and heat wave conditions during larval development in a community composed of four predaceous dragonfly species and their zooplankton prey. We found that warming and heat waves similarly shifted the composition of larval communities towards greater species evenness in abundance. Interestingly, heat waves were correlated with greater larval survival, reduced zooplankton availability, and larvae were found at higher trophic positions relative to those in warmed communities. These results indicate that heat waves and warming may have similar effects on the composition of species in a community, but heat waves may have a greater effect on the trophic dynamics of aquatic food webs.

Introduction

Ecological communities across diverse habitats have changed immensely over the last century. These changes include species compositional shifts (Moritz et al., 2008; Ruhland et al., 2008), changes to the body size spectrum and biomass distribution (Daufresne et al., 2009), and mismatches in the timing of species interactions (Visser & Both, 2005). There is now convincing evidence that these changes are related to increasing temperatures due to climate change (Walther et al., 2002). However, while historical data provide insights into patterns of change,

83 the mechanisms driving these patterns are often difficult to distinguish, given that temporal changes can be correlated with a multitude of other factors (Magurran, 2007). Researchers have therefore used experimental warming approaches simulating future climate scenarios to have a better understanding of the mechanisms underlying these community changes (e.g. Walker et al., 2006; Yvon-Durocher et al., 2015). Yet climate warming not only includes mean increases in temperature, but also extreme temperature events such as heat waves (Klein Tank & Können, 2003; Gershunov et al., 2009), which are expected to become more frequent and severe in the future (Meehl, 2004). Even though these temperature anomalies have been recognized as a threat to biodiversity (Easterling et al., 2000), there have been few studies investigating their impact on community composition (Thompson et al., 2013). In order to have better-informed predictions of the effects of future climate change on ecological communities, experimental data are needed that incorporate both increases in temperature and extreme thermal fluctuations in a realistic manner (Jentsch et al., 2007).

The thermal environment plays a major role in determining regional and local scale patterns of diversity (Peters et al., 2016). However, elucidating the mechanisms driving patterns of community shifts in response to climate change is challenging due to the complex interaction between direct and indirect effects of temperature on species (Ockendon et al., 2014; Lord et al., 2017). In ectotherms, physiological functions such as metabolism are directly impacted by environmental temperature which ultimately affects growth and developmental rates (Brown et al., 2004). The temperature dependence of these biological rates can vary among species, leading to inter-specific variation in response to the same thermal conditions (Dell et al., 2011). Warmer conditions in temperate latitudes are expected to favour warm-adapted species, such as those with equatorial evolutionary origins (Calosi et al., 2008). However, seasonal variation increases with latitude, and will therefore be expected to favour species that can also withstand these extreme fluctuations (Rapoport, 1975; Stevens, 1989; Ragland & Kingsolver, 2008). Ultimately, these differences will shape community composition, favouring the species that can tolerate warmer, as well as more variable temperatures (Deutsch et al., 2008), i.e., the “winner” species of climate change (Williams et al., 2008).

Warming can also indirectly affect communities by altering interactions between trophic groups, e.g. changing the strength and topology of interactions within food webs (Barton et al., 2009; Shurin et al., 2012). These changes are manifested in both bottom-up and top-down processes.

Warming can increase primary productivity, leading to more resources at the base of trophic webs (bottom-up effect). Increasing consumption rates due to rising energetic costs associated with faster metabolic rate in warmer conditions can strengthen top-down effects, i.e., trophic cascades (Kratina et al., 2012), as autotroph metabolism is less sensitive to temperature increases than heterotroph metabolism (O’Connor, 2009). At high temperatures, a lag between metabolic rate and a consumer’s maximal consumption rate is more likely to lead to energetic deficiencies in consumers (Lemoine & Burkepile, 2012), potentially weakening the consumer-resource link. Lastly, due to the hump-shaped relationship between temperature and performance, periodic high temperatures are more likely to surpass ectotherm optimal thermal limits and place them in unfavourable conditions for performance (Dell et al., 2014; Sinclair et al., 2016). Ultimately, being pushed beyond thermal tolerance limits, even for short periods, can have sublethal effects such as reduced fitness (Kingsolver et al., 2013).

Community composition in freshwater habitats has changed over the last century with strong evidence suggesting climate change and anthropogenic influence as the main factors driving these changes (Burgmer et al., 2007; Daufresne & Boet, 2007; Ball-Damerow et al., 2014). Organisms in these habitats are particularly vulnerable to temperature changes given that the majority are ectothermic, meaning that their body temperatures are dependent on environmental temperatures. Theoretical work predicts that changes to the mean and variance in temperature will affect individual performance in a variety of ways (Vasseur et al., 2014). However, it is still unclear what the community level impacts of these changes will be given that temperature can have direct and indirect influences. Boersma and colleagues (2016) found that extreme high temperatures were strongly correlated with particular assemblages of freshwater invertebrate species, suggesting that climatic extremes may play a significant role in determining aquatic community composition by filtering out species that cannot survive those conditions. To date, there is still a lack of experimental work investigating the effects of future, more frequent and severe heat waves on aquatic communities. This is problematic given that organisms inhabiting shallow ponds or the littoral zone of lakes will be exposed to the same maximum temperatures as those in terrestrial systems (Schneider & Mauser, 1996; Abrahams et al., 2007).

Intraguild predation (IGP), or the killing and consuming of potential competitors (Polis et al., 1989), is common in aquatic systems and important in structuring these communities. IGP, which can include cannibalism, is a form of omnivory that, like predation rate, may become

85 more frequent with future warming (Barton & Schmitz, 2009; Start et al., 2017; Frances & McCauley, 2018). Higher frequencies of IGP drive down the abundance of consumers, ultimately reducing competition between guild members (Polis et al., 1989; Wissinger, 1992). However, species can differ in their responses to warming, with temperature potentially acting indirectly as a filter by favouring species that become superior IG predators under those conditions. To date, it is unknown how communities of IG predators will shift in response to future warming, particularly, if there are species that can become “winners” by becoming superior IG predators relative to others.

To understand the community consequences of periodic heat waves and warming on freshwater community structure, we combined larvae of four dragonfly species (Odonata: Anisoptera) in aquatic mesocosms in which we could manipulate the developmental temperatures they experienced. We used larval dragonflies, as they are voracious meso- and top-predators in freshwater habitats, they frequently engage in IGP, and respond physiologically and behaviourally to increases in temperature (Krishnaraj & Pritchard, 1995; Pritchard et al., 1996, 2000; Hassall & Thompson, 2008; McCauley et al., 2015, 2018; Suhling et al., 2015; Frances et al., 2017). We based our expected community outcomes on the combinations of potential direct and indirect effects of heat waves and warming on aquatic communities. To mechanistically determine how trophic interactions within these communities are affected by warming and heat waves (i.e., is IGP increasing or decreasing?), we performed stable isotopes analysis on dragonfly larvae that survived to the end of the experiment to estimate their trophic positions in their respective communities. We predict that: (i) increasing temperatures will increase larval growth rates leading to phenology shifts that alter community stage structure (i.e. earlier shifts to later developmental stages), (ii) species will differ in the magnitude of this shift based on interspecific differences in growth responses to temperature; (iii) warming will increase the abundance of primary consumers (zooplankton, which are prey of dragonfly larvae) via increased primary productivity; hence warming will indirectly sustain higher abundances of larvae (secondary consumers) from increased prey availability. Alternatively, if IGP increases with temperature, warming will indirectly decrease larval abundance, and lead to increased zooplankton abundance. Lastly, (iv) if warming is driving higher levels of IGP, larval dragonflies will be found at higher trophic positions within their food webs. Utilizing a novel warming approach, we were able to disentangle the effects of warmer mean temperatures from those

86 caused by heat waves, both of which are expected with future climate change, yet are rarely experimentally performed and compared side-by-side.

5.1 Methods

We conducted our study at University of California Davis’ Quail Ridge Reserve (Napa County, California, 38º 28′58.72′′ N, 122º 8′58.17′′ W) using larvae of four dragonfly species: Erythemis collocata Hagen (western pondhawk), Libellula luctuosa Burmeister (widow skimmer), Pachydiplax longipennis Burmeister (blue dasher), and Plathemis lydia Drury (common whitetail). These species are in the family Libellulidae and are seasonally abundant in lakes and ponds in the study region (Manolis, 2003). All are univoltine with larval development typically lasting between 9 and 10 months (Corbet, 1999). While these species differ slightly in the length and timing of their summer breeding seasons, all adults of these species overlapped during the time of egg collection.

We collected eggs from mating adult females at ponds at the Wantrup Wildlife Sanctuary (Pope Valley, California, 38º 35′58.07′′ N, 122º 22′13.10′′ W) and Quail Ridge Reserve in June 2015. We caught females with insect nets and gently held them by their wings while dipping their abdomens into 125 mL sample cups filled with pond water until they released their eggs. Dipping initiates oviposition in female dragonflies. These females were then marked with permanent black ink on their wings to prevent collecting eggs from the same female. After collection, we kept eggs cool with ice packs in insulated coolers and transported them to our rearing facility at Quail Ridge Reserve (44 km from Wantrup Wildlife Sanctuary).

A total of 32 fertilized clutches (E. collocata: n = 10, L. luctuosa: n = 5, P. longipennis: n = 11 and Pl. lydia: n = 6) were collected and placed into species-specific rearing containers (1153- liter cattle tanks; two rearing tanks per species). We inoculated mesocosms with a standard, high abundance mixture of crustacean zooplankton collected from two tanks stocked with locally sourced zooplankton cultures. Zooplankton served as the primary food source for dragonflies. Locally collected dry oak leaves were added for allochthonous nutrient addition to sustain phytoplankton growth and support the zooplankton population as well as to provide structural complexity in the benthic portion of tanks. The rearing tanks were covered with shade cloth to minimize colonization of other insects and amphibians. Temperatures in the rearing tanks

averaged 20.27 ± 3.0º C from June 2015 to October 2015, with no significant temperature differences between rearing tanks (ANOVA: F3,436 = 1.16, P = 0.33).

In October 2015, we sorted larvae from the rearing containers into the experimental mesocosms (416-litre tanks). We had three treatments: ambient, warmed, and heat waves. Each treatment was replicated across eight spatial blocks (total of 24 mesocosms). Treatments were randomly assigned within each block. Ambient mesocosms were not thermally manipulated and therefore tracked the natural seasonal and diurnal variation in air temperature at our field site. Electronic sensors in each mesocosm continuously monitored the water temperature (CR1000; Campbell Scientific Inc., Logan, Utah). Aquarium heaters (True Temp 500W; JBJ, Inglewood, California) in thermally manipulated treatments adjusted the water temperature relative to the readings from the ambient mesocosm in each block. This method allowed us to simulate climate warming in a realistic way in which diurnal and seasonal variation in temperature could be maintained while still increasing mean temperature to predetermined levels. Warmed mesocosms were programmed to be 3º C above ambient temperatures. This treatment simulated conditions predicted for this region in the next 50 years under a conservative, low emissions scenario (Cayan et al., 2008). Our heat wave treatment mesocosms were heated to either 1º or 5º C above ambient. The level of warming switched on a weekly basis, resulting in a mean increase of 3º C over ambient. This treatment was designed to mimic extreme heat waves which are predicted to become more frequent in the future (Coumou & Rahmstorf, 2012). Each mesocosm was wrapped with fibreglass insulation to mimic buffering that contact with soil provides from air temperatures in small ponds.

We filled mesocosms with non-chlorinated well water. Each mesocosm received a standard amount of locally gathered leaf litter (primarily Quercus douglasii and Pinus sabiniana), four sticks of rabbit chow (which also serves as a nutrient base), aquarium gravel sufficient to cover the bottom the tank, four strips of mosquito netting and frayed polypropylene rope tied to stainless steel washers for artificial structure. Finally, each mesocosm was inoculated with a standard volume of crustacean zooplankton before adding dragonfly larvae. We randomly selected and added 20 dragonfly larvae per species to each mesocosm for a total density of 80 dragonfly larvae per mesocosm (20 larvae x 4 species). This quantity falls within natural densities of ponds (Corbet, 1999). During the sorting process, an additional 20 larvae per species were randomly collected, preserved in 95% ethanol, and then photographed so that we could

88 measure head widths of larvae using ImageJ 1.43 (National Institutes of Health, Bethesda, Maryland). Dry mass was estimated from head widths using species- or genus-specific coefficients of length-mass regressions from Benke et al. (1999, p. 339), except for P. longipennis, for which there are no published data, so we used the coefficients of the closest species morphologically, Celithemis fasciata. Starting dry mass of each species varied: mean ± SD: Pl. lydia = 4.58 ± 1.73 mg, E. collocata = 3.89 ± 1.21 mg, L. luctuosa = 3.33 ± 0.96 mg, and P. longipennis = 1.29 ± 0.59 mg (Appendix, Fig. S5.1). After sorting larvae into mesocosms, we capped each mesocosm with a lid constructed with 70% shade cloth attached to a wood frame. The lids were fastened securely on the sides with ratchet straps, and small holes were made in the shade cloth to accommodate the electrical cords of the water sensors and heaters. Every lid had these holes, including the ambient treatments; therefore the mesocosms were not invulnerable to external colonization, however, it was greatly reduced (Cáceres & Soluk, 2002).

Larvae developed undisturbed within tanks for a total of 183 days. We ended the experiment at this point because we wanted to remove larvae before they began metamorphosis, and prior experiments have shown that larvae developing in similar heated conditions complete larval development around this time (McCauley et al., 2015, 2018). We removed all surviving larvae from the mesocosms on 22 through 24 April 2016. Larvae were preserved in 95% ethanol, were photographed, and their head widths were measured from these photos using ImageJ. The majority of larvae were removed on these dates, however, mesocosms were sampled in the following days to ensure that all larvae had been removed. During one of these subsequent sweeps on 29 April, two drowned adult P. longipennis were found in mesocosms 7 and 11 (warmed and increased variation treatments, respectively). No other adults were found in any of the mesocosms. To estimate zooplankton abundance in each mesocosm, we used a standard sampling protocol of sweeping a 150 µm plankton net longitudinally through the tank, starting from the top and zig-zagging in a downward motion for a total of five passes through each tank. Zooplankton were preserved in 95% ethanol. After sampling for zooplankton, all mesocosms were drained, and sediment and gravel were picked through by hand to ensure all dragonfly larvae had been removed.

We estimated larval developmental stage, we measured wing-pad length relative to abdominal length under a dissecting scope. We chose this method instead of absolute measurements to

89 control for wing/body length differences among species. Because dragonflies are hemimetabolous, their wings develop during the larval stage, with significant growth during the last few larval instars (Hawking & New, 1996). Therefore, wing-pad length provides a reliable estimate of larval instar, with larger values representing later stages of development (Hawking & New, 1996). Final dry mass of individual larvae were approximated from larval head widths using the same equations in Benke et al (1999) in order to assess larval growth rates.

We determined !13C and !15N stable isotope ratios of dragonfly larvae and zooplankton to estimate the trophic position of larvae across treatments. The preserved zooplankton and dragonfly larvae were oven-dried at 65º C for 24 h and 48 h, respectively. Dried animals were ground to a homogenous powder with a mortar and pestle. We weighed dried dragonfly larvae (0.6 – 0.8 mg) and zooplankton (~1 mg) on a microbalance accurate to ±1µg (Mettler Toledo XS3DU), then sealed the weighed samples within tin capsules. In some cases, inadequate quantities of dry mass led to pooling dragonflies (of the same species) or zooplankton specimens from a different mesocosm of the same treatment. Analyses were completed at the Trophic Ecology Laboratory at the Great Lakes Institute for Environmental Research, University of Windsor using a Delta V Advantage Mass spectrometer (Thermo) coupled to a Costech 4010 Elemental Combustion system and a ConFlo IV gas interface. Stable isotope ratios are expressed in delta notation, defined as parts per thousand (‰) deviation from lab standard material (Virtual Peedee Belemnite for d13C and atmospheric nitrogen for d15N). Because we were primarily interested in trophic position, we focus on the results of d15N, as significant increases in nitrogen isotopes occur at each successive trophic level, whereas d13C is more indicative of carbon flow pathways (DeNiro & Epstein, 1978). Temperature can increase the rate of nitrogen excretion, assimilation and tissue growth, resulting in lower 15N enrichment (Power et al., 2003). Our trophic position calculations account for these differences by comparing larval dragonflies to zooplankton in their respective mesocosm.

5.1.1 Statistical analyses

In one mesocosm (tank 10, ambient), no dragonfly larvae survived of any species. In this tank, larval survival was 1.69 standard deviations away from the mean of larval survivors in ambient conditions. For this reason, we removed this tank and all other tanks from this block (tanks 10, 11, and 12) from our analyses. It is unclear what the cause of this mortality was, however, in past

90 experiments a similar problem arose with the same tank (McCauley et al., 2015, 2018) suggesting a mechanical issue.

We performed a Poisson generalized linear mixed effects model (GLMM) with the R package lme4 (version 1.1-14) to analyze the effects of the thermal environment on the number of larval survivors. We included temperature treatment (3 levels), species (n = 4), and the interaction between treatment and species as fixed factors and the proportion of larvae surviving as the response variable. Block was included as a random effect to account for spatial heterogeneity in abiotic conditions. P-values were approximated from Type II Wald Chi-square tests using the ‘Anova’ function in the car package (version: 2.1-6) for this model and the following models.

To assess how treatment temperature affected the abundance and proportional representation of species in each community, we calculated Pielou’s evenness index, J’, which equals H’/ ln(S), where H’ is the Shannon-Wiener index = −∑ ./01./ ; ./ is the proportional abundance of species I; and S is the total number of species (Maurer & McGill, 2011). To compare evenness, (J’), across treatments, we performed a linear mixed effects model (LME; ‘lmer’ function in the lme4 R package) with J’ as the response variable. Treatment was included as a fixed effect, and block was included as a random effect.

To estimate developmental stage across treatments, we performed a Poisson GLMM and included treatment, species, and the interaction between treatment and species as fixed effects. We also calculated the growth rate (mg dry mass × day-1) of individuals across treatments using the change in dry mass divided by the length of the experiment (183 days). To analyze growth rate, we used an LME with treatment, species, and their interaction as fixed effects and block was included as a random effect.

Zooplankton samples consisted primarily of cladocerans. Ostracods were also present in the zooplankton samples, however, in relatively low abundances compared to cladocerans (< 9.0% of total zooplankton abundance per mesocosm). For this reason, we only analyzed cladoceran abundance. We used a Poisson GLMM to compare cladoceran abundance across treatments (fixed effect) with block included as a random effect.

We estimated the trophic position of dragonfly larvae relative to their zooplankton prey across treatments using the equation from Post (2002):

45 45 Trophic position = 23 6789:;<=>?3 6@A:=C + " ∆3456

$) $) Where ! Nconsumer and ! Nbase are the nitrogen isotope ratios of the dragonfly larvae and zooplankton, respectively. l is the trophic position of the zooplankton (l = 2 for primary consumers). Trophic fractionation or ∆!$)N is the enrichment in !$)N per trophic level. We used a ∆!$)N value of 2.0‰, which is lower than Post 2002 (3.4‰); however, trophic fractionation tends to be lower for consumers feeding on primarily invertebrate diets $) $) (McCutchan et al., 2003). For each ! Nconsumer value, we used the ! Nbase from the same tank to minimize variation otherwise introduced by pooling treatments across multiple tanks. Trophic position values were analyzed with an LME. We included treatment, species, and the interaction between treatment and species as fixed effects. Block was included as a random effect. For all our analyses, we used R version 3.3.2 (R Core Team, Vienna, Austria 2016). Unless otherwise mentioned, all results are presented as means ± SE.

5.2 Results

5.2.1 Temperature differences between treatments

Temperatures in the ambient and warmed treatments reflected the natural seasonal and diurnal temperature fluctuations at our field site (Fig. 5.1). Average temperatures (± SD) in the mesocosms were 8.91 ± 2.68º C in ambient, 12.02 ± 2.61º C in warmed, and 12.17 ± 3.00º C in heat wave treatments. Thermal variation, measured as SD, was significantly higher in the heat wave treatments than both the warmed and ambient treatments (ANOVAtreatment: F2,12 = 55.01, P < 0.0001; Tukey’s HSD with Bonferroni correction: P < 0.001, for both). Furthermore, our heat wave treatment met the IPCC (2012) statistical definition for extreme events, in which high temperatures during heat spikes were beyond the 90th percentile of average ambient temperatures. Maximum and minimum temperatures within the mesocosms were 20.0º and 3.5º C, respectively (Fig. 5.1).

5.2.2 Survival

There was a significant interaction between temperature treatment and species on larval survival, meaning species had differential survival depending on the temperature conditions % (GLMMtreatment x species: #' = 17.26, P = 0.008). This pattern was driven by the near extirpation of

92 all but one species in ambient conditions. Pachydiplax longipennis was the only species to survive to the end of the experiment in ambient mesocosms (40% survival compared to other species: < 1%; Fig. 5.2). Due to these large species-specific differences in survival across treatments, we repeated analyses with only warmed and heat wave treatments for the number of surviving larvae, as well as developmental stage, and trophic position (see below). All species in the heat wave treatments had significantly higher survival than in warmed treatments (79.5 ± % 3.2% and 60.5 ± 6.5% respectively; GLMMtreatment: #$ = 14.19, P = 0.0002; Fig. 5.2).

In several mesocosms (n = 8), we found larvae of the odonate colonist, Libellula saturata, which was not included in our experiment. This species readily colonizes small waterbodies, including artificial pools (McCauley et al., 2010). Interestingly, L. saturata larvae were only found in warmed and heat wave treatments and in low densities (range: 2-8 larvae per mesocosm). It remains unclear whether habitat selection by adults or post-colonization habitat filtering processes led to this nonrandom colonization pattern, as both processes can lead to superficially similar larval communities (French & McCauley, 2018). Regardless, the presence of L. saturata had no negative effect on the abundance of other odonate species in warmed and variable treatments (Wilcoxon rank sum test: W = 24, P = 1.0). For this reason, we did not include this species in our analyses except when examining trophic structure.

5.2.3 Evenness

Temperature treatment significantly impacted species evenness within the larval communities % (LME: #% = 48.05, P < 0.0001; Fig. 5.3). Communities exposed to heat waves and warming were more even in species’ abundances than ambient communities. Evenness was significantly lower in ambient conditions, given that communities were dominated by a single species, P. longipennis (Tukey’s HSD with Bonferroni correction: P < 0.0001 for both comparisons to ambient). Species evenness was not significantly different in heat waves versus warmed treatments (P = 0.32; Fig. 5.3).

5.2.4 Development

Treatment temperature had a significant effect on wing-pad development and growth rate in P. % % longipennis (wing-pad development: GLMM: #% = 6.16, P = 0.046; growth rate: LME: #% = 104.08, P < 0.0001). Specifically, P. longipennis had wing-pads 15.6 and 16.4% longer in

93 warmed and heat wave treatments, respectively, compared to ambient conditions (Tukey’s HSD: P = 0.057 and P = 0.083, respectively). Additionally, P. longipennis in warmed and heat wave treatments had growth rates 83.6 and 81.4% faster, respectively than those in ambient conditions (Tukey’s HSD: P < 0.0001, for both comparisons; Fig. 5.4). When comparing across all species, growth rates were faster for species experiencing heat waves than warming alone (LMEtreatment: % #$ = 3.92, P = 0.048), and there was a trend towards more advanced wing-pad development % between warmed and heat wave treatments (GLMMtreatment: #$ = 1.72, P = 0.19). All species had significantly different growth rates, with the fastest growing species, P. longipennis, growing over fourfold faster than the slowest growing species, E. collocata (Fig. 5.4).

5.2.5 Zooplankton

% We found a significant effect of treatment temperature on cladoceran abundance (GLMM: #% = 6144.2, P < 0.0001; Appendix, Fig. S5.2). Cladoceran abundance was highest in ambient compared to heat wave and warming treatments (eight- and three-fold higher, respectively; Tukey’s HSD: P < 0.0001, for both comparisons). Heat wave treatments had significantly lower cladoceran abundances than warmed (52.2% fewer cladocerans; Tukey’s HSD with Bonferroni correction: P < 0.0001), and ambient conditions (89.0% fewer; P < 0.0001).

5.2.6 Trophic position

We expected that the treatment with the lowest larval survival (i.e., ambient conditions) would have dragonfly larvae at higher trophic positions if higher frequencies of IGP in these conditions are driving this survival patterns. We found that trophic positions of the dragonfly larvae were significantly different across treatments (ANOVA: F2,145 = 30.98, P < 0.0001; Table 5.1). Contrary to our predictions, in heat wave treatments where survival was highest, dragonfly larvae were at the highest trophic positions relative to the zooplankton (Tukey’s HSD: P < 0.0001 for both). Surprisingly, the largest difference in mean trophic position was between the heat wave and warmed treatments (mean trophic position ± SE, increased variation: 3.22 ± 0.04; warmed: 2.79 ± 0.05). This pattern was not driven by the most abundant species, P. longipennis, which had the lowest trophic position compared to all other species (Tukey’s HSD: P < 0.001 compared to all in heat wave and warmed treatments; Table 5.1). Instead, the species occupying the highest trophic positions were L. saturata, Pl. lydia, and L. luctuosa (Table 5.1).

Table 5.1 Comparison of the calculated trophic positions of dragonfly larvae in ambient, warmed conditions, and those exposed to heat waves. Equation used to calculate trophic position is 45 45 adapted from Post (2002): Trophic position = 23 6789:;<=>?3 6@A:=C + " ; where " = 2, and a ∆3456 ∆!$)N = 2.0‰. Range indicates minimum and maximum values calculated for each species in each treatment. Treatment Species n Trophic position (mean ± SE) Range

Ambient* P. longipennis 15 2.54 ± 0.08 1.54 2.95

Warmed E. collocata 13 2.85 ± 0.10 2.34 3.57 L. luctuosa 16 2.91 ± 0.10 2.30 3.51 L. saturata 12 2.84 ± 0.11 2.44 3.50 P. longipennis 21 2.50 ± 0.08 1.87 3.05 Pl. lydia 7 3.15 ± 0.12 2.60 3.50 Heat waves E. collocata 16 3.22 ± 0.05 2.80 3.52 L. luctuosa 21 3.38 ± 0.04 2.91 3.62 L. saturata 9 3.45 ± 0.12 2.83 3.97 P. longipennis 21 2.87 ± 0.07 2.33 3.55 Pl. lydia 11 3.40 ± 0.08 2.98 3.75

* Trophic position is only shown for this species in ambient treatments, as this was the only species to survive to the end of the experiment in this treatment in sufficiently large numbers for comparison purposes.

5.3 Discussion

Along with increases in mean temperature, the frequency and severity of extreme climatic events such as heat waves, are predicted to become more common in the future (Meehl, 2004), yet to date, we lack experimental data examining how heat waves impact aquatic community structure and trophic dynamics. Using dragonfly larvae as representative members of pond communities in experimental mesocosms, we explored the community-wide consequences of periodic heat waves and warming during larval development. We found that heat waves and warming were correlated with higher larval survival, larvae developed more rapidly, and these communities were more even in species abundances. Prey abundance was lowest in communities that experienced periodic heat waves indicating strengthening top-down effects. Lastly, our stable isotope analysis revealed that larval trophic position did not appear to be related to differences in mortality, suggesting that IGP may not be driving differences in survival patterns as we initially predicted. Together, our results suggest that future warming and more frequent and prolonged heat waves may affect community composition by potentially facilitating the coexistence of several species and increasing the strength of trophic cascades. Additionally, periodic heat waves could lead to unequal developmental shifts between species, in turn altering trophic interactions amongst predators.

5.3.1 Effects of heat waves and warming on survival, evenness, and development

Based on prior theoretical and empirical work, we expected that IGP would increase with warming (Gilbert et al., 2014; Start et al., 2017; Frances & McCauley, 2018), leading to lower survival rates. Instead, our results demonstrate that simulated heat waves increased larval survival, even compared to warmed treatments in which average temperatures were equivalent (80% versus 61% survival in heat wave and warmed treatments, respectively; Fig. 5.2). This suggests that heat waves may have a positive effect on survival of this group, perhaps because larvae can maximize growth (e.g. lepidopteran larvae, Kingsolver et al., 2015) and feeding (e.g. fish, Spigarelli et al., 1982) during high temperature periods. Animals exposed to sublethal fluctuating temperatures can become more tolerant of acute high temperatures by increasing expression of heat shock proteins (Lindquist & Craig, 1988), which could potentially increase survival. However, it is poorly understood how heat shock protein expression is affected by longer periods of sustained high temperatures, such as heat waves. Regardless, higher survival

100 when exposed to heat waves suggests that over time, members of this group could increase in abundance in aquatic communities, potentially having large cascading effects on other trophic levels such as their shared prey, and in turn, affect ecosystem processes (Shurin et al., 2012).

Larval survival was lowest in the ambient conditions (Fig. 5.2). We cannot rule out low temperatures as a direct source of mortality for larvae in ambient conditions (lowest temperatures recorded in our mesocosms: 3.5º C), however, these temperatures are not abnormally low for the region (Boersma et al., 2016), and these species are likely able to tolerate temperatures below freezing (Sinclair et al., 2003). More likely, species differ in their optimal temperatures, i.e., the range of temperatures under which they can maximize growth or other measures of performance. One hypothesis is that P. longipennis has lower temperature thresholds for growth and activity compared to the other species. This difference in thermal optima could have allowed P. longipennis to become a superior competitor or IG predator at lower temperatures which allowed this species to persist while the others were extirpated from ambient conditions.

Several studies have shown that differences in thermal tolerance may lead to community reorganization with future warming (Rosset et al., 2010; Sunday et al., 2012; Stuart-Smith et al., 2015). For instance, in a guild of aphid species, shifts in competitive dominance between species was facilitated by the direct effect of extreme high temperatures on species-specific demographic rates (Ma et al., 2015). Similarly, we expected that temperature-related differences in performance among species would shift the community composition towards species that become dominant IG predators in those conditions. Instead, our results suggest that warming and heat waves associated with climate change actually facilitates the maintenance of species diversity for this group, potentially due to species-specific differences in performance at low temperatures, as opposed to warmer temperatures.

Our results demonstrate that larvae grow and develop more rapidly in warmer conditions (Fig. 5.4). These findings complement survey data demonstrating earlier adult emergence over the last several decades (Hassall et al., 2007) and experimental evidence that warming advances adult emergence phenology (McCauley et al., 2015, 2018). Interestingly, the rate of development and growth was unequal among species (Fig. 5.4). In warmed and heatwave conditions, the species with the most advanced wing-pad development and the fastest growth rate was P. longipennis. In fact, during sampling, two adults of this species were found in warmed and increased variation

101 tanks, indicating that in those conditions, P. longipennis was nearing the end of larval development. Plathemis lydia and E. collocata, on the other hand, were the slowest growing species when exposed to heat waves and warming, with growth rates comparable to P. longipennis in ambient treatments (Fig. 5.4). Differences in development and growth rate could be attributed to species differences in foraging mode. In lizards, sit-and-wait predators conserve more energy than active foragers; however, the energetic gains they miss out on are 1 to 2 times greater (Huey & Pianka, 1981). Our results indicate that warmer and more variable temperatures can lead to large developmental asynchronies among species, however, it does not appear that this caused higher frequencies of IGP. Instead, species may be able to coexist in these conditions potentially by resolving life-history trade-offs differently (e.g., fast versus slow life-history strategies; McCauley, 2008; Lancaster et al., 2017). Future studies should investigate the possibility that species with different life history strategies may be differentially affected by climate change in ways that decrease their niche overlap, and hence their interactions with potential competitors, potentially facilitating coexistence in warmer and more variable conditions (Lancaster et al., 2017).

5.3.2 Stable isotopes and food web structure

Along with being top-predators in fishless ponds, dragonfly larvae are also IG predators, which can further enrich 15N in their tissues (Ponsard & Arditi, 2000) and hence, their trophic positions within their food webs. Because survival was high in warmed and heat wave conditions relative to ambient conditions (Fig. 5.2), our results suggest that IGP was infrequent between dragonflies in these conditions. Our stable isotope analysis, however, did not support this hypothesis; P. longipennis, the only successful species in ambient conditions, was at a lower trophic position in ambient compared to heat wave treatments (Table 5.1). A limitation of stable isotope analysis is that an animal’s tissue does not immediately reflect its diet. It is possible that IGP occurred close to the time we sampled; in this case, the isotopic signature of IGP would not be as clear due to rate of tissue isotopic turnover, e.g., isotopic half-life (Vander Zanden et al., 2015). In our insects, the isotopic half-life is short, at approximately 16 days, meaning, that the diet of a larva can be detected in its tissues starting at around 16 days (calculated from: Vander Zanden et al., 2015). Alternatively, if IGP occurred early in larval development, the isotopic signature of IGP would also be less clear. While it is unknown if there is a particular developmental stage in which IGP is most likely to occur (Anholt, 1994; Hopper et al., 1996), there is a strong

102 relationship between body size differences among individuals and frequency of IGP (Wissinger, 1988). At the time we added the early instar larvae to the experiment, the relative variation in body mass among species was 38% (coefficient of variation). However, the smallest species initially, P. longipennis, was the only species in ambient conditions to survive to the end of the experiment, which suggests that the initial variability in size did not lead to higher frequencies of IG predation at that time. Therefore, the differential mortality in these treatments is unlikely to have arisen from IGP. It is more likely that the interaction between species’ thermal tolerances and competitive abilities at different temperatures led to the observed patterns in larval survival, zooplankton abundance, and trophic level across treatments.

Interestingly, trophic positions of dragonfly larvae were highest in the heat wave treatments, even though survival was highest and species evenness was similar in both heat wave and warmed treatments (Figs. 5.2 & 5.3). There could be an interaction between perceived IG predation risk and the physiological effects of fluctuating temperatures on prey switching behaviour. Warming can induce diet switches in grasshoppers, which prefer to eat diets high in protein when predation risk is high (Schmitz et al., 2016). Similarly, a heat wave may increase the likelihood of IGP between dragonfly larvae, while periods of lower temperatures may cause larvae to switch to zooplankton, which are less nutritious but also a less risky option. Warming caused grasshoppers to switch from grasses to less nutritious forbs Alternatively, another way that 15N is enriched is when muscle tissue is catabolized during periods of low food availability (reviewed by Hertz et al., 2015). Given that zooplankton abundance was significantly lower in heat wave treatments at the time we sampled, it is possible that increased competition for prey led to stoichiometric differences in !15N between superior and inferior competitors.

5.4 Conclusion

Warmer mean temperatures, as well as more frequent and severe heat waves, will affect organisms across diverse landscapes, including freshwater systems. Given that the majority of species in these systems are ectothermic and are therefore highly sensitive to changes in their thermal environment, it is crucial that we have a better understanding of how their communities will be affected by future climate change. Long-term monitoring of these communities will be essential to document temporal patterns of compositional change, while modelling approaches can provide insight into future shifts. Our study experimentally demonstrates the drastic effects

103 of warming and heat waves on community composition. Specifically, we provide evidence to support predicted compositional shifts towards increased species coexistence in warmer temperate environments. Contrary to predicted effects of warming on predation rate, our results suggest that IGP will not increase to frequencies that would have re-structuring effects on communities, at least for warming at these levels. However, other trophic groups such as primary consumers/herbivores will likely experience strengthening top-down control with increasing frequency of heat waves. Our work emphasizes the potentially large effects that heat waves can have on the dynamics and composition of aquatic communities, and future studies should investigate the how these conditions affect the long-term persistence of these communities.

Acknowledgements

We would like to thank S. Waddell at Quail Ridge Reserve for infrastructure and logistical support, the Napa Land Trust for access to source ponds, C. Fitzpatrick and R.A. Rehman for field and laboratory assistance, B. McMeans for providing advice about stable isotope analyses, and M. Johnson for providing comments on this manuscript.

5.5 References Abrahams M V, Mangel M, Hedges K (2007) Predator-prey interactions and changing environments: who benefits? Philosophical Transactions of the Royal Society B: Biological Sciences, 362, 2095–2104. Allen AP, Brown JH, Gillooly JF (2002) Global biodiversity, biochemical kinetics, and the energetic- equivalence rule. Science, 297, 1545–1548. Anholt BR (1994) Cannibalism and early instar survival in a larval damselfly. Oecologia, 99, 60–65. Ball-Damerow JE, M’Gonigle LK, Resh VH (2014) Changes in occurrence, richness, and biological traits of dragonflies and damselflies (Odonata) in California and Nevada over the past century. Biodiversity and Conservation, 23, 2107–2126. Barton BT, Schmitz OJ (2009) Experimental warming transforms multiple predator effects in a grassland food web. Ecology Letters, 12, 1317–1325. Barton BT, Beckerman AP, Schmitz OJ (2009) Climate warming strengthens indirect interactions in an old-field food web. Ecology, 90, 2346–2351. Benke, A. C., Huyrn, A. D., Smock, L. A., & Wallace, J. B. (1999). Length-mass relationships for freshwater macroinvertebrates in North America with particular reference to the southeastern United States. Journal of the North American Benthological Society, 18(3), 308–343.

104

Boersma KS, Nickerson A, Francis CD, Siepielski AM (2016) Climate extremes are associated with invertebrate taxonomic and functional composition in mountain lakes. Ecology and Evolution, 6, 8094–8106. Brown JH, Gillooly JF, Allen AP, Savage VM, West GB (2004) Toward a metabolic theory of ecology. Ecology, 85, 1771–1789. Burgmer T, Hillebrand H, Pfenninger M (2007) Effects of climate-driven temperature changes on the diversity of freshwater macroinvertebrates. Oecologia, 151, 93–103. Cáceres CE, Soluk DA (2002) Blowing in the wind: a field test of overland dispersal and colonization by aquatic invertebrates. Oecologia, 131, 402–408. Calosi P, Bilton DT, Spicer JI (2008) Thermal tolerance, acclimatory capacity and vulnerability to global climate change. Biology Letters, 4, 99–102. Cayan DR, Maurer EP, Dettinger MD, Tyree M, Hayhoe K (2008) Climate change scenarios for the California region. Climatic Change, 87, S21–S42. Corbet PS (1999) Dragonflies: Behaviour and Ecology of Odonata, Revised Ed edn. Harley Books, Colchester, Essex. Coumou D, Rahmstorf S (2012) A decade of weather extremes. Nature Climate Change, 2, 491– 496. Daufresne M, Boet P (2007) Climate change impacts on structure and diversity of fish communities in rivers. Global Change Biology, 13, 2467–2478. Daufresne M, Lengfellner K, Sommer U (2009) Global warming benefits the small in aquatic ecosystems. Proceedings of the National Academy of Sciences, 106, 12788–93. Dell AI, Pawar S, Savage VM (2011) Systematic variation in the temperature dependence of physiological and ecological traits. Proceedings of the National Academy of Sciences 108, 10591–10596. Dell, A. I., Pawar, S., & Savage, V. M. (2014). Temperature dependence of trophic interactions are driven by asymmetry of species responses and foraging strategy. Journal of Animal Ecology, 83(1), 70–84. DeNiro MJ, Epstein S (1978) Influence of diet on the distribution of carbon isotopes in animals. Geochimica et Cosmochimica Acta, 42, 495–506. Deutsch CA, Tewksbury JJ, Huey RB, Sheldon KS, Ghalambor CK, Haak DC, Martin PR (2008) Impacts of climate warming on terrestrial ectotherms across latitude. Proceedings of the National Academy of Sciences, 105, 6668–6672. Easterling DR, Meehl GA, Parmesan C, Changnon SA, Karl, Thomas R, Mearns LO (2000) Climate extremes: observations, modeling, and impacts. Science, 289, 2068–2074. Frances DN, McCauley SJ (2018) Warming drives higher rates of prey consumption and increases rates of intraguild predation. Oecologia. 10.1007/s00442-018-4146-y Frances DN, Moon J, McCauley SJ (2017) Effects of environmental warming during early life- history on libellulid odonates. Canadian Journal of Zoology, 95, 373–382. French SK, McCauley SJ (2018) Canopy cover affects habitat selection by adult dragonflies. Hydrobiologia, 1–15.

105

Gershunov A, Cayan DR, Iacobellis SF (2009) The great 2006 heat wave over California and Nevada : Signal of an increasing trend. Journal of Climate, 22, 6181–6203. Gilbert B, Tunney TD, Mccann KS et al. (2014) A bioenergetic framework for the temperature dependence of trophic interactions. Ecology Letters, 17, 902–914. Hassall C, Thompson DJ (2008) The effects of environmental warming on Odonata: a review. International Journal of Odonatology, 11, 131–153. Hassall C, Thompson DJ, French GC, Harvey IF (2007) Historical changes in the phenology of British Odonata are related to climate. Global Change Biology, 13, 933–941. Hawking JH, New TR (1996) The development of dragonfly larvae (Odonata: Anisoptera) from two streams in north-eastern Victoria, Australia. Hydrobiologia, 317, 13–30. Hertz E, Trudel M, Cox MK, Mazumder A (2015) Effects of fasting and nutritional restriction on the isotopic ratios of nitrogen and carbon: A meta-analysis. Ecology and Evolution, 5, 4829–4839. Hopper KR, Crowley PH, Kielman D (1996) Density dependence, hatching synchrony, and within-cohort cannibalism in young dragonfly larvae. Ecology, 77, 191–200. Huey RB, Pianka ER (1981) Ecological consequences of foraging mode. Ecology, 62, 991–999. Jentsch A, Kreyling J, Beierkuhnlein C (2007) A new generation of climate-change experiments: events, not trends. Frontiers in Ecology and the Environment, 5, 365–374. Kingsolver JG, Diamond SE, Buckley LB (2013) Heat stress and the fitness consequences of climate change for terrestrial ectotherms. Functional Ecology, 27, 1415–1423. Kingsolver JG, Higgins JK, Augustine KE (2015) Fluctuating temperatures and ectotherm growth: distinguishing non-linear and time-dependent effects. Journal of Experimental Biology, 218, 2218–2225. Klein Tank AMG, Können GP (2003) Trends in Indices of daily temperature and precipitation extremes in Europe, 1946-99. Journal of Climate, 16, 3665–3680. Kratina P, Greig HS, Thompson PL, Carvalho-Pereira TSA, Shurin JB (2012) Warming modifies trophic cascades and eutrophication in experimental freshwater communities. Ecology, 93, 1421–1430. Krishnaraj R, Pritchard G (1995) The influence of larval size, temperature, and components of the functional response to prey density on growth rates of the dragonflies Lestes disjunctus and Coenagrion resolutum (Insecta: Odonata). Canadian Journal of Zoology, 73, 1672–1680. Lancaster LT, Morrison G, Fitt R (2017) Life history trade-offs, the intensity of competition, and coexistence in novel and evolving communities under climate change. Philosophical Transactions of the Royal Society B, 372, 20160046. Lemoine NP, Burkepile DE (2012) Temperature-induced mismatches between consumption and metabolism reduce consumer fitness. Ecology, 93, 2483–2489. Lindquist S, Craig EA (1988) The heat-shock proteins. Annual Review of Genetics, 22, 631–77. Lord JP, Barry JP, Graves D (2017) Impact of climate change on direct and indirect species interactions. Marine Ecology Progress Series, 571, 1–11.

106

Ma G, Rudolf VHW, Ma CS (2015) Extreme temperature events alter demographic rates, relative fitness, and community structure. Global Change Biology, 21, 1794–1808. Magurran AE (2007) Species abundance distributions over time. Ecology Letters, 10, 347–354. Manolis T (2003) Dragonflies and Damselflies of California, v. 72 edn. University of California Press, Berkeley and Los Angeles, California. Maurer BA, McGill BJ (2011) Measurement of species diversity. In: Biological Diversity: Frontiers in Measurement and Assessment (eds Magurran AE, McGill BJ), p. 345. Oxford University Press. McCauley SJ (2008) Slow, fast and in between: habitat distribution and behaviour of larvae in nine species of libellulid dragonfly. Freshwater Biology, 53, 253–263. McCauley SJ, Brodin T, Hammond J (2010) Foraging rates of larval dragonfly colonists are positively related to habitat isolation: results from a landscape-level experiment. The American Naturalist, 175, E66–E73. McCauley SJ, Hammond JI, Frances DN, Mabry KE (2015) Effects of experimental warming on survival, phenology, and morphology of an aquatic insect (Odonata). Ecological Entomology, 40, 211–220. McCauley SJ, Hammond JI, Mabry KE (2018) Simulated climate change increases larval mortality, alters phenology, and affects flight morphology of a dragonfly. Ecosphere, 9, e02151. McCutchan JH, Lewis WM, Kendall C, McGrath CC (2003) Variation in trophic shift for stable isotope ratios of carbon, nitrogen, and sulfur. Oikos, 102, 378–390. Meehl GA (2004) More intense, more frequent, and longer lasting heat waves in the 21st century. Science, 305, 994–997. Moritz C, Patton JL, Conroy CJ, Parra JL, White GC, Beissinger SR (2008) Impact of a century of climate change on small-mammal communities in Yosemite National Park, USA. Science, 322, 261–264. O’Connor MI (2009) Warming strengthens an herbivore: plant interaction. Ecology, 90, 388– 398. Ockendon N, Baker DJ, Carr JA et al. (2014) Mechanisms underpinning climatic impacts on natural populations: Altered species interactions are more important than direct effects. Global Change Biology, 20, 2221–2229. Peters, M. K., Hemp, A., Appelhans et al. (2016) Predictors of elevational biodiversity gradients change from single taxa to the multi-taxa community level. Nature Communications, 7(13736). Polis GA, Myers CA, Holt RD (1989) The ecology and evolution of intraguild predation: potential competitors that eat each other. Annual Review of Ecology and Systematics, 20, 297–330. Ponsard S, Arditi R (2000) What can stable isotopes (d15 N and d13 C) tell us about the food web of soil macro-invertebrates? Ecology, 81, 852–864. Post DM (2002) Using stable isotopes to estimate trophic position : Models, methods, and assumptions. Ecology, 83, 703–718.

107

Power M, Guiguer KRRA, Barton DR (2003) Effects of temperature on isotopic enrichment in Daphnia magna : implications for aquatic food-web studies. Rapid Communications in Mass Spectrometry, 17, 1619–1625. Pritchard G, Harder LD, Mutch R a, Robert a (1996) Development of aquatic insect eggs in relation to temperature and strategies for dealing with different thermal environments. Biological Journal of the Linnean Society, 58, 221–244. Pritchard G, Harder LD, Kortello A, Krishnaraj R (2000) The response of larval growth rate to temperature in three species of coenagrionid dragonflies with some comments on Lestes disjunctus (Odonata: Coenagrionidae, Lestidae). International Journal of Odonatology, 3, 105–110. Ragland GJ, Kingsolver JG (2008) Evolution of thermotolerance in seasonal environments: The effects of annual temperature variation and life-history timing in Wyeomyia smithii. Evolution, 62, 1345–1357. Rapoport E (1975) Areografía: estrategias geográficas de las especies. Fondo de Cultura Econémica, Mexico. Rosset V, Lehmann A, Oertli B (2010) Warmer and richer? Predicting the impact of climate warming on species richness in small temperate waterbodies. Global Change Biology, 16, 2376–2387. Ruhland K, Paterson AM, Smol JP (2008) Hemispheric-scale patterns of climate-related shifts in planktonic diatoms from North American and European lakes. Global Change Biology, 14, 2740–2754. Schmitz OJ, Rosenblatt AE, Smylie M (2016) Temperature dependence of predation stress and the nutritional ecology of a generalist herbivore. Ecology, 97, 3119–3130. Schneider K, Mauser W (1996) Processing and accuracy of Landsat Thematic Mapper data for lake surface temperature measurement. International Journal of Remote Sensing, 17, 2027–2041. Shurin JB, Clasen JL, Greig HS, Kratina P, Thompson PL (2012) Warming shifts top-down and bottom-up control of pond food web structure and function. Philosophical Transactions of the Royal Society B: Biological Sciences, 367, 3008–3017. Sinclair BJ, Addo-Bediako A, Chown SL (2003) Climatic variability and the evolution of insect freeze tolerance. Biological Reviews, 78, 181–195. Sinclair, BJ, Marshall, KE, Huey, RB, et al., (2016). Can we predict ectotherm responses to climate change using thermal performance curves and body temperatures? Ecology Letters, 1372–1385. Spigarelli SA, Thommes MM, Prepejchal W (1982) Feeding, growth, and fat deposition by brown trout in constant and fluctuating temperatures. Transactions of the American Fisheries Society, 111, 199–209. Start D, Kirk D, Shea D, Gilbert B (2017) Cannibalism by damselflies increases with rising temperature. Biology Letters, 13. Stevens GC (1989) The latitudinal gradient in geographical range: how so many species coexist in the tropics. The American Naturalist, 133, 240–256.

108

Stuart-Smith RD, Edgar GJ, Barrett NS, Kininmonth SJ, Bates AE (2015) Thermal biases and vulnerability to warming in the world’s marine fauna. Nature, 528, 88–92. Suhling F, Suhling I, Richter O (2015) Temperature response of growth of larval dragonflies – an overview. International Journal of Odonatology, 18, 15–30. Sunday JM, Bates AE, Dulvy NK (2012) Thermal tolerance and the global redistribution of animals. Nature Climate Change, 2, 686–690. Thompson RM, Beardall J, Beringer J, Grace M, Sardina P (2013) Means and extremes: Building variability into community-level climate change experiments. Ecology Letters, 16, 799–806. Vasseur DA, DeLong JP, Gilbert B et al. (2014) Increased temperature variation poses a greater risk to species than climate warming. Proceedings of the Royal Society B: Biological Sciences, 281, 20132612–20132612. Visser ME, Both C (2005) Shifts in phenology due to global climate change: the need for a yardstick Shifts in phenology due to global climate change : the need for a yardstick. Proceedings of the Royal Society. Biological Sciences, 272, 2561–2569. Walker MD, Wahren CH, Hollister RD et al. (2006) Plant community responses to experimental warming across the tundra biome. Proceedings of the National Academy of Sciences, 103, 1342–6. Walther GR, Post E, Convey P et al. (2002) Ecological responses to recent climate change. Nature, 416, 389–395. Williams SE, Shoo LP, Isaac JL, Hoffmann AA, Langham G (2008) Towards an integrated framework for assessing the vulnerability of species to climate change. PLoS Biology, 6, e325. Wissinger SA (1988) Effects of food availability on larval development and inter-instar predation among larvae of Libellula lydia and Libellula luctuosa (Odonata: Anisoptera). Canadian Journal of Zoology, 66, 543–549. Wissinger SA (1992) Niche overlap and the potential for competition and intraguild predation between size- structured populations. Ecology, 73, 1431–1444. Yvon-Durocher G, Allen AP, Cellamare M et al. (2015) Five years of experimental warming increases the biodiversity and productivity of phytoplankton. PLoS Biology, 13, 1–22. Vander Zanden MJ, Clayton MK, Moody EK, Solomon CT, Weidel BC (2015) Stable isotope turnover and half-life in animal tissues: A literature synthesis. PLoS ONE, 10, 1–16.

109

General Discussion 6 Discussion

The central goal of my thesis was to assess how differences in species' responses to warming affect community interactions to better understand if particular traits can predict the "winner" and "loser" species in the context of climate change and gain insights into how species interactions may change in future, warmer environments. Specifically, I examined the effects of warming on growth, development, and behaviour at the individual level using common garden and behavioural assays. I then used these responses to make informed predictions and empirically test how warming affects community interactions, structure, and composition with laboratory and mesocosm experiments.

To understand how warming affects developmental rates and the synchrony in the timing of hatching, I reared eggs of four taxonomically related and ecologically similar dragonfly species in common gardens over a gradient of temperatures and measured interspecific differences in the timing of hatching phenology (chapter 2). I also measured the performance of hatchling dragonfly larvae at a range of constant and fluctuating temperatures to understand how warming and increased thermal variation affected body size at hatching, as well as growth and survival during the early stages of development (chapter 2). To establish if behavioural responses to temperature influence the outcomes of predaceous interactions between dragonfly species, I first performed behavioural assays on a suite of dragonfly species over a range of temperatures. Next, using data from the behavioural assays, I selected three species that differed the most in their behavioural responses to warming and conducted experiments measuring intraguild predation (IGP) between these three species This allowed me to assess if differences in activity expression at warmer temperatures predict who became the dominant IG predator (chapter 3). To evaluate whether warming affects behavioural plasticity exhibited in response to predation risk, I measured activity, foraging, and other behaviours of damselfly prey exposed to warming, predation risk, and the combination of these factors. I also assessed how temperature affects their risk of being preyed upon (chapter 4). Lastly, I combined four species of dragonfly larvae in pond mesocosms to determine the impact of warming and heat waves on community composition, and trophic dynamics (chapter 5).

110

Results and conclusions of each of my data chapters have been summarized in their respective chapters. In this section, I discuss the general conclusions of my thesis, how my work has advanced our understanding of the effects of warming on ecological communities, and how future work should address remaining knowledge gaps.

Species’ responses to warming

Idiosyncratic species’ responses to temperature complicate efforts to accurately forecast the future effects of climate change on communities (Walther et al., 2002). This is because differential responses to temperature can modify how and when species interact (Gilman et al., 2010). In addition, interaction strength changes with temperature, leading to higher consumption rates and predation rates when organisms experience warmer temperatures (Dell et al., 2014). The importance of the effects of temperature on species interactions are so significant that recent reviews have demonstrated they are more likely responsible for driving patterns of extirpation with recent climate change than temperature acting as a proximate cause of mortality (Cahill et al., 2012; Ockendon et al., 2014). To make reliable predictions about how future climate change will shape communities, we need information about how species respond to temperature as well as how those responses affect the outcome of their interactions with others in the community.

6.1.1 Inter- and intraspecific responses to warming

The relative order in which species arrive to an assembling community is important in determining community structure, with the first species to arrive generally being conferred the greatest benefit in terms of establishing territory or accruing resources when competition is low (i.e., priority effects; Connell & Slatyer, 1977). Also, the interactions between species in a community can be strongly influenced by the relative timing of arrival, wherein early arriving species tend to have size advantages over later occurring species (size-mediated priority effects, Rasmussen et al., 2014). However, recent work has shown that climate change can alter the relative timing in which species normally arrive (Stuble et al., 2017) and can thus have large consequences on competitive or predatory dynamics occurring within a patch. In communities that are strongly shaped by size-structured interactions such as intraguild predation, temperature- induced changes in the relative arrival of species could significantly impact which species are conferred the benefit of size-mediated priority effects (Rasmussen et al., 2014).

111

I explored the possibility of temperature induced size-mediated priority effects occurring in several odonate species with overlapping habitat distributions. A lack of interspecific variation in the effect of temperature on egg development rate suggests that rank order of hatching remains synchronous, and thus, priority effects are likely to remain consistent in future climates (chapter 2, Frances et al., 2017). However, in chapter 5 where I explored the community-level effects of warming and heat waves, I found that the rate of larval development appeared to be different among species. Even though eggs of all four species were laid and presumably hatched around the same time, there were still differences in growth and developmental rates among species. This suggests that size-mediated priority effects as the result of warmer temperatures may not be as important at the egg stage, however differences in the temperature dependence of growth among species during the larval stage may be more important in shaping communities.

Interestingly, intraspecific variation in egg development rate attributed to the timing of breeding of the mothers was much higher than any interspecific variation in response to temperature (chapter 2, Frances et al., 2017). In other words, early-breeding mothers laid eggs that developed more slowly than later-breeding mothers. If this variation in egg development time between early and later-breeding mothers is heritable, my results suggest that rising temperatures due to climate change could more strongly affect the structure of individual populations compared to interspecific dynamics and that variation within populations may actually confer long-term benefits to these populations and their persistence with future climate change. This work highlights the importance of understanding inter- and intraspecific variation in response to warming, especially at different ontogenetic stages. Future work should examine if the plasticity of responses within a species is adaptive and if eco-evolutionary feedbacks can affect the persistence of these species in warmer climates.

6.1.2 The temperature-size rule in dragonflies

While my thesis did not explicitly explore the TSR in adult dragonflies, several observations in my studies highlight interesting patterns involving body size in odonates and how temperature affects this critical trait. For instance, in chapter 2, I found that larval dragonflies were significantly smaller at the time of egg-hatching when raised in warmer temperatures (Frances et al., 2017), similar to results of a study with damselflies (Van Doorslaer & Stoks, 2005). Interestingly, previous experimental work has shown that adult dragonflies that developed in

112 warmer temperatures were not smaller than those that had developed in ambient conditions, yet they completed larval development significantly earlier (McCauley et al., 2015, 2018). This suggests, that even though larvae are smaller at hatching and grow faster when raised in warmer conditions, both parts of the predictions underlying the TSR, adult dragonflies are not smaller when they emerge.

Oxygen has been suggested as an important driver of the TSR, especially for aquatic organisms because oxygen availability decreases with increasing temperature, and thus the selective pressure to have smaller cells is stronger in warmer conditions (Forster et al., 2012; Hoefnagel & Verberk, 2015). Rollinson and Rowe (2018) recently demonstrated that mode of oxygen uptake is likely important in determining which species are most affected by warmer temperatures. Specifically, animals that respire aerially (lung-breathers), as opposed to strictly gill- or cutaneous-breathers, do not conform to the TSR, potentially because oxygen limitation in water is not as crucial of a selection pressure for animals that aerially breathe as it is for animals dependent on the passive diffusion of oxygen across their membranes (Rollinson & Rowe, 2018). Hypothetically, because odonate development occurs strictly in aquatic habitats, and because they acquire oxygen via gills (Corbet, 1999), they should then be expected to conform to these rules, yet evidence for this is mixed (Cothran & Thorp 1082, De Block & Stoks 2003, McCauley et al. 2015, 2018).

Results from my thesis suggest a reason why exceptions have occurred. Increased predation on smaller individuals may be driving the community size structure towards larger-bodied individuals (Wellborn et al., 1996). In chapter 3, I demonstrated that the frequency of predation on smaller individuals increases with rising temperature (increased size-biased predation, Frances & McCauley, 2018). If the recruitment of smaller individuals cannot keep up with increased size-biased predation at higher temperatures, the average body size of IG predators will appear to increase due to the prevalence of the larger bodied, dominant IG predators and the loss of smaller individuals. Selection will ultimately favour individuals that grow quickly and become superior IG predators in warmer conditions. Community composition may also shift as larger, more dominant species are favoured under these conditions. Future work should examine how altered predation at warmer temperatures affects not only the size distribution but also the species composition in these communities, to determine if biotic interactions are masking any detectable effects of the TSR.

113

6.1.3 Climate change experiments: warming regimes, scale, and trade-offs

Experiments are necessary to test theories and uncover mechanisms underlying patterns occurring in nature. However, ecologists have long realized that the choice of the experimental design comes with trade-offs between realism, precision, and generality (Hairston, 1989; Morin, 1998). Laboratory experiments generally have a high degree of control, allowing for precise measurements, however, they are also criticized for being over simplistic with low generalizability. Field experiments, on the other hand, have a high degree of realism, yet they can often be confounded by natural spatiotemporal processes and can suffer from non- independence among experimental units (Chalcraft et al., 2005). Mesocosm studies are often thought to be an excellent middle ground between field and laboratory studies (Morin, 1998). Another issue with ecological experiments is that the choice of the experimental protocol can affect the interpretation of results (Skelly, 2002) as well as the size of the effects (Wolkovich et al., 2012).

In my thesis, I performed a variety of experiments to gain a thorough understanding of how future warming will affect aquatic invertebrate communities. In chapter 3, I directly measured the impact of warming on the frequency of IGP in the laboratory. This experiment was performed over a relatively short time scale (days), and I was able to track changes in the frequency of IGP for individual species. In pond mesocosms, I examined the effects of warming on the survival of IG predators over a longer temporal scale (months). In the laboratory, warming increased the frequency of IGP twofold over a 72-h period. In the mesocosms, while I could not track IGP directly (another limitation of using this type of experiment), there was little evidence to suggest that warmer conditions increased IGP because dragonfly larvae had relatively high survival at the end of the experiment (183 days). At first glance, these outcomes appear to be conflicting, however, fundamentally they were demonstrating two very different things. In the laboratory, experimental temperatures represented contemporary average summer temperatures and future, warmer summer scenarios. The warmed conditions in the laboratory (28-32º C) were higher than dragonfly larvae experienced in the warmed mesocosms (12º C), as this experiment spanned larval development in the winter months. The importance of seasonality in driving patterns of change within communities may be as critical or even exceed the significant role of spatial variation (McMeans et al., 2015). Warmer than average winters over the last few decades have facilitated winter survival of some species (e.g. lepidopterans Crozier, 2004; Battisti et al.,

114

2005). Warmer winters may generally result in relaxed environmental filtering, decreasing overwinter mortality and increase the performance of species which otherwise are approaching their lower temperature tolerance limits. Warmer summer temperatures, on the other hand, are more likely to approach ectotherm optimal thermal tolerance (Kingsolver et al., 2013), and are more likely to increase encounters with predators (Dell et al., 2014), both factors that are likely to increase mortality and decrease the ability to persist in these warmer environments.

Warming may also indirectly affect species interaction strengths. Aside from size ratios, one of the major predictors of IGP is habitat complexity. Habitats with more structure reduce the likelihood of IGP because it creates more refugia for IG prey, thereby reducing the likelihood of predatory encounters between individuals (Janssen et al., 2007). Many aquatic organisms, including, odonate larvae use algae as structure, and experiments that remove this structure observe a substantial increase in mortality. In chapter 5, I found that survival was highest in the warmest conditions in which zooplankton abundance was strongly depressed. It is possible that the release of grazing pressure by zooplankton allowed periphyton to increase and potentially create more structure for larvae in these communities. While we did not measure phytoplankton or periphyton in our mesocosms, other work has found that higher temperatures in aquatic systems increase phytoplankton biomass (Yvon-Durocher et al., 2011). Odonate larvae could therefore be engaging in niche construction where their increased consumption rates on grazers positively feeds back to increase their refugia from potential IG predators (Odling-Smee et al., 2003). To test this hypothesis, mesocosm experiments with algal removal/additions crossed with warming treatments could be performed to test if 1) higher temperatures have cascading effects on the density of primary producers, 2) more/less algae changes the likelihood of IGP, and 3) if this form of environmental engineering significantly modifies selection pressures acting on individuals. This experiment could also help determine if warmer temperatures indirectly facilitate coexistence through this mechanism.

6.1.4 Mechanisms driving community patterns

Understanding the mechanism surrounding patterns of change is challenging due to a complex interplay of direct and indirect interactions in ecological networks (Zarnetske et al., 2012). My thesis explored several possibilities for how changing species interactions in warmer temperatures can affect communities. Laboratory experiments can help determine the

115 mechanisms driving patterns of change because they are easily controlled, can be replicated, and can be complex in design, e.g. test multiple factors, including interactive effects (Cadotte et al., 2005), yet they can often oversimplify complex systems, hence potentially under- or over- estimating effect sizes. Mesocosm or field studies, on the other hand, are more realistic in terms of replicating natural systems, yet they can also be logistically challenging, expensive, and often cannot be used to decipher mechanisms driving patterns of change. While mesocosm studies are powerful in that they offer realism and precision, they can also be criticized concerning their applicability and generalization to natural communities (Morin, 1998).

How then, can we better design experiments so that we can make more realistic projections of how complex natural systems behave, yet also mechanistically understand what causes these changes? Comprehensive studies that incorporate ecological responses in multiple dimensions can give a complete picture of how future climate change will shape species and communities. For instance, Ma et al. (2015) performed a comprehensive study that was able to relate changes in species’ thermal tolerances to their competitive abilities at extreme high temperatures. By including data from long-term monitoring projects, they were also able to use historical trends to inform models of how communities would be affected by extreme temperature events. These studies, while incredibly informative and thorough, tend to be rare for a reason; they often require excellent data on the thermal physiology of the organism, and also good quality long- term data that can identify patterns over longer temporal scales. Paired with realistic mesocosm studies, they can uncover mechanisms surrounding patterns of community change to be able to make well-informed predictions for how future climate change will shape communities.

Conclusions

My thesis examined the effects of warming on odonate species and their aquatic communities. By conducting experiments that tested predictions of the community effects of future climate change, I was able to highlight the importance that biotic interactions play in overall community structure and dynamics. A change in the strength of interactions, both between guild members (IGP) and across trophic groups (predator-prey) can have a large effect on the structure and composition of freshwater communities. The potential for feedbacks occurring within communities due to the indirect effects of warming on species is an emerging field of interest (e.g., Gilman et al., 2010; Cahill et al., 2012; Ohlberger, 2013; Ockendon et al., 2014) and work

116 from my thesis strengthens the argument that more work should be done to understand how complex direct and indirect effects of warming interact to affect ecological communities.

References Battisti A, Stastny M, Netherer S, Robinet C, Schopf A, Roques A, Larsson S (2005) Expansion of geographic range in the pine processionary moth caused by increased winter temperatures. Ecological Applications, 15, 2084–2096. Cadotte MW, Drake JA, Fukami T (2005) Constructing nature: laboratory models as necessary tools for investigating complex ecological communities. Advances in Ecological Research, 37, 333–353. Cahill AE, Aiello-lammens ME, Fisher-Reid MC et al. (2012) How does climate change cause extinction? Proceedings of the Royal Society B: Biological Sciences, 280, 20121890– 20121890. Chalcraft DR, Binckley CA, Resetarits WJ (2005) Experimental venue and estimation of interaction strength: comment. Ecology, 86, 1061–1067. Connell JH, Slatyer RO (1977) Mechanisms of succession in natural communities and their role in community stability and organization. The American Naturalist, 111, 1119–1144. Cothran, M. L., & Thorp, J. H. (1982). Emergence patterns and size variation of Odonata in a thermal reservoir. Freshwater Invertebrate Biology, 1(4), 30–39. Corbet PS (1999) Dragonflies: Behaviour and Ecology of Odonata, Revised Ed edn. Harley Books, Colchester, Essex. Crozier LG (2004) Warmer winters drive butterfly range expansion by increasing survivorship. Ecology, 85, 231–241. De Block, M., & Stoks, R. (2003). Adaptive sex-specific life history plasticity to temperature and photoperiod in a damselfly. Journal of Evolutionary Biology, 16(5), 986–995. Dell AI, Pawar S, Savage VM (2014) Temperature dependence of trophic interactions are driven by asymmetry of species responses and foraging strategy. Journal of Animal Ecology, 83, 70–84. Van Doorslaer W, Stoks R (2005) Thermal reaction norms in two Coenagrion damselfly species: Contrasting embryonic and larval life-history traits. Freshwater Biology, 50, 1982–1990. Forster J, Hirst AG, Atkinson D (2012) Warming-induced reductions in body size are greater in aquatic than terrestrial species. Proceedings of the National Academy of Sciences of the United States of America, 109, 19310–4. Frances DN, McCauley SJ (2018) Warming drives higher rates of prey consumption and increases rates of intraguild predation. Oecologia. Gilman SE, Urban MC, Tewksbury J, Gilchrist GW, Holt RD (2010) A framework for community interactions under climate change. Trends in Ecology and Evolution, 25, 325–331. Hairston NG (1989) Ecological Experiments: Purpose, Design, and Execution. Cambridge University Press, 370 pp.

117

Hoefnagel KN, Verberk WCEP (2015) Is the temperature-size rule mediated by oxygen in aquatic ectotherms? Journal of Thermal Biology, 54, 56–65. Janssen A, Sabelis MW, Magalhães S, Montserrat M, Van Der Hammen T (2007) Habitat structure affects intraguild predation. Ecology, 88, 2713–2719. Kingsolver JG, Diamond SE, Buckley LB (2013) Heat stress and the fitness consequences of climate change for terrestrial ectotherms. Functional Ecology, 27, 1415–1423. McCauley SJ, Hammond JI, Frances DN, Mabry KE (2015) Effects of experimental warming on survival, phenology, and morphology of an aquatic insect (Odonata). Ecological Entomology, 40, 211–220. McCauley SJ, Hammond JI, Mabry KE (2018) Simulated climate change increases larval mortality, alters phenology, and affects flight morphology of a dragonfly. Ecosphere, 9, e02151. McMeans BC, McCann KS, Humphries M, Rooney N, Fisk AT (2015) Food web structure in temporally-forced ecosystems. Trends in Ecology and Evolution, 30, 662–672. Morin PJ (1998) Realism, precision, and generality in experimental ecology. In: Experimental Ecology: Issues and Perspectives (Eds Resetarits WJ, Bernardo J), pp. 50–56. Oxford University Press. Ockendon N, Baker DJ, Carr JA et al. (2014) Mechanisms underpinning climatic impacts on natural populations: Altered species interactions are more important than direct effects. Global Change Biology, 20, 2221–2229. Odling-Smee FJ, Laland KN, Feldman MW (2003) Niche Construction: The Neglected Process in Evolution. Princeton University Press, 472 pp. Ohlberger J (2013) Climate warming and ectotherm body size - from individual physiology to community ecology. Functional Ecology, 27, 991–1001. Rasmussen NL, Van Allen BG, Rudolf VHW (2014) Linking phenological shifts to species interactions through size-mediated priority effects. Journal of Animal Ecology, 83, 1206– 1215. Rollinson N, Rowe L (2018) Temperature-dependent oxygen limitation and the rise of Bergmann’s rule in species with aquatic respiration. Evolution. Skelly DK (2002) Experimental venue and estimation of interaction strength. Ecology, 83, 2097– 2101. Stuble KL, Zefferman EP, Wolf KM, Vaughn KJ, Young TP (2017) Outside the envelope: rare events disrupt the relationship between climate factors and species interactions. Ecology, 98, 1623–1630. Walther GR, Post E, Convey P et al. (2002) Ecological responses to recent climate change. Nature, 416, 389–395. Wellborn, G. A., Skelly, D. K., & Werner, E. E. (1996). Mechanisms creating community structure across a freshwater habitat gradient. Annual Review of Ecology and Systematics, 27(1996), 337–363. Wolkovich EM, Cook BI, Allen JM et al. (2012) Warming experiments underpredict plant phenological responses to climate change. Nature, 485, 494–497.

118

Yvon-Durocher G, Montoya JM, Trimmer M, Woodward G (2011) Warming alters the size spectrum and shifts the distribution of biomass in freshwater ecosystems. Global Change Biology, 17, 1681–1694. Zarnetske PL, Skelly DK, Urban MC (2012) Biotic multipliers of climate change. Science, 336, 1516–1518.

119

Appendices Appendix I

Chapter 2 Supporting information

Fig. S2.1. Visual of experimental design to examine the effects of temperature on egg development rates in four dragonfly species. Eggs were collected from adult females and sorted into eggs cups filled with water. Eggs cups were then placed in water baths heated to one of four treatment conditions: 22.2º, 24.9º, 27.0º, and 30.5º C

120

Fig. S2.2. Visual of experimental design to measure the effects of temperature and thermal variation on growth rates and survival for L. intacta. After hatching, larvae were placed individually in plastic larva cups and floated in a water bath heated to three different constant temperatures (23.4º, 27.5º, 30.4º C) and one variable condition: (27.4º, alternating between 23º and 30º C weekly)

121

Alternative data analyses: We used a linear mixed effects model (LME) with the function ‘lmer’ from the R package lme4, version 1.1-9) to assess the effects of treatment temperature, species identity, and the interaction between treatment temperature and species on egg development rate (1/days to hatching). We included treatment tank, maternal identity (female), and oviposition date as random effects. Starting with a fully parameterized model, we evaluated the significance of each random effect using log-likelihood ratio tests (logLRT) and eliminated factors with P > 0.05. Next, to assess the significance of each fixed effect in the model, we used the same step-wise elimination method above to obtain their contribution towards the final model.

For our second experiment we used a LME model to analyze the effect of treatment temperature on mean egg development rate for the species Le. intacta. Temperature was included as a categorical factor, because we were not just interested in the mean temperature conditions, but also the influence of temperature fluctuations. Oviposition date and female were included as random effects. Model selection procedure followed the same steps as above, in which we sequentially removed random effects and performed log-likelihood ratio tests to assess their importance in the final model. P-values for fixed effects were determined using log-likelihood ratio tests after sequentially removing the effect and comparing to the full model. This same procedure was used to analyze the effect of treatment temperature on hatchling head width. Female, the interaction between treatment and female, and oviposition date were included as random effects. Additionally, we used an LME with the same model selection procedure to analyze the effect of treatment temperature on larval growth rate. To account for spatial heterogeneity across tanks during the growth experiment, we included tank as a random effect as well as oviposition date, and the interaction between treatment and female. To determine the effects of temperature treatment on survival, we used a survival analysis using the package “survival” (version 2.38) and “coxme” (version 2.2-5). All analyses were performed in R version 3.2.3 (R Development Core Team 2016).

122

Table S2.1. Results from the log-likelihood ratio tests on the effects of fixed and random factors on egg development rate, hatchling head width, and egg development rate based on a linear mixed-effects (LME) model.

Response variable Factor Predictor df (2 P Egg development rate Fixed Treatment 1 76.81 < 0.001 Species 3 35.48 < 0.001 Treatment × Species 3 2.60 0.46 Random Female 1 0 1 Oviposition date 1 4.28 0.39 Tank 3 16.69 < 0.001

Le. intacta egg Fixed Treatment 3 59.07 < 0.001 development rate Random Female 1 0.01 0.90 Oviposition date 1 4.32 0.04 Hatchling head width Fixed Treatment 3 25.84 < 0.001 Random Female 1 4.96 0.03 Treatment × female 1 6.26 0.01 Oviposition date 1 0.30 0.58 Growth rate Fixed Treatment 3 16.72 < 0.001 Random Female 1 0.19 0.67 Treatment × female 1 2.43 0.12 Tank 1 0.48 0.49 Oviposition date 1 0.83 0.36

123

Table S2.2. Pair-wise Tukey contrasts for each response variable, with Bonferroni correction for multiple comparisons. Bold-faced font denotes significant differences (α = 0.05)

Egg development time (days) Hatchling head width (mm) Growth Rate (mm/day) Survival comparison estimate z value p-value estimate z value p-value estimate t value p-value estimate z value p-value Low – high -0.05 -20.60 <0.001 0.06 11.20 <0.001 -0.008 -3.54 0.003 -1.88 -6.55 <0.001 Med – high -0.02 -6.69 <0.001 0.04 8.39 <0.001 -0.011 -4.63 <0.001 -0.84 -3.46 0.003 Var – high -0.04 -17.42 <0.001 0.04 7.45 <0.001 -0.009 -3.86 0.001 -0.76 -3.07 0.01 Med – low 0.04 14.42 <0.001 -0.02 -2.91 0.02 -0.003 -1.12 1.0 -1.04 6.55 0.001 Var – low 0.01 3.30 0.01 -0.02 -3.88 <0.001 -0.001 -0.41 1.0 1.11 3.93 <0.001 Var – med -0.03 -11.13 <0.001 -0.005 -0.97 1.0 0.001 0.69 1.0 0.08 0.34 1.0

124

Appendix II

Chapter 3 Supporting information

Sampling dates of species

Larvae were collected from ponds based on their natural phenologies, meaning that we collected larvae when each species was abundant and also of a large enough size to be used in experiments. We collected E. canis from ponds at the University of Toronto’s Koffler Scientific Reserve (KSR) in early June 2015. We collected L. intacta and S. vicinum from KSR in early July 2015. We collected L. pulchella, E. simplicicollis, and P. lydia in early September 2015 from KSR and ponds at the University of Toronto Mississauga (UTM). Due to low sample sizes of E. simplicicollis in 2015, we also collected larvae of this species in September 2016 at UTM.

Methods and results of the prey behavioural responses to temperature

To understand how the prey of dragonfly larvae respond to increasing temperatures, we conducted an experiment measuring the activity level of mosquito larvae at 24º, 28º and 32º C. A single mosquito was placed in a tank (same setup as above), allowed to acclimate for 16 h, and then movement of the mosquito was observed every 5 minutes for 2 h 30 m. To account for behavioral responses of mosquitoes to the presence of a dragonfly larvae predator, we included caged (nonlethal) dragonflies in each tank with the mosquito being observed so that chemical cues of the dragonfly could be detected by the focal mosquito. Assays at each temperature were replicated 15 times. Mosquitoes were only used once during the trials.

Using a GLM with a Poisson error distribution, we analyzed the effect of temperature on the activity (number of moves) of the mosquito larvae. We found that mosquito larvae did not increase activity in response to temperature (F1,149 = 1.50, P = 0.22).

125

Chapter 3 supplemental figures

Fig. S3.1. Head widths (mm) of dragonfly species used in the behavioural assays. Same letters above error bars indicate species that are not significantly different. Error bars are ± 1 SD. Species codes are the first two letters of the genus and species. Sample size varied by species (Ersi: n = 20, Plly: n = 22, Lipu: n = 32, Epca: n = 20, Syvi: n = 45, Lein: n = 32)

126

Fig. S3.2. The relationship between temperature and dragonfly larva microhabitat preference. Microhabitat preference was measured as the proportion of time a larva spent on structure during the observation period. Points represent means ± 1 SE. Sample size varied by species (Ersi: n = 20, Plly: n = 22, Lipu: n = 32, Epca: n = 20, Syvi: n = 45, and Lein: n = 32)

127

Fig. S3.3. The relationship between maximum realized temperature and species’ activity responses to temperature. Activity response was measured as the change in activity (number of moves) from 28º to 32º C. Maximum realized temperatures are median maximum temperatures of the warmest month extracted from point locations across species’ ranges. Species codes are the first two letters of the genus and species

128

Molts 2

0 23 28 32 o Temperature ( C)

Fig. S3.4. The relationship between the total number of dragonfly larvae molts at the end of the 72-hour IGP trial and temperature shown with 95% confidence intervals

129

Chapter 3 supplemental table

o o o o Species Northern range Southern Tlow ( C) Tav ( C) Thigh ( C) Tbreadth ( C) limit (o) range limit (o) E. simplicicollis 44.2 26.54 -3 14.7 32.2 35.2 P. lydia 46.25 29.99 -5.3 12.6 30.3 35.6 L. pulchella 47.8 30.76 -8.4 9.9 28.8 37.2 S. vicinum 48.66 31.23 -8.8 9.4 27.9 36.7 L. intacta 50.57 39.12 -10.9 7.5 27 37.9 E. canis 53.07 38.07 -16.8 5.3 25.1 41.9

Table S3.1. The latitudinal and temperature ranges of dragonfly species used in the study. Latitudes were extracted from the Global Biodiversity Information Facility (http://www.gbif.org). Thermal data were estimated from point locations with Worldclim (Fick & Hijmans, 2017). Northern and southern range boundaries are the top and bottom 5% of latitudinal range extent for each species.

References:

Fick SE, Hijmans RJ (2017) Worldclim 2: New 1-km spatial resolution climate surfaces for global land areas. In: Int. J. Climatol.

129 130

Appendix III

Chapter 4 Supporting information

How does predation risk by Anax increase with temperature?

We collected Anax junius (hereafter Anax), the dragonfly predators of damselflies (McPeek, 1990), from the same fishless pond as the damselflies at KSR on 20 September 2017. Anax were housed individually in cups in our laboratory and fed one damselfly once per day. There was some variance in size between Anax individuals, however, they were all within at least one instar from each other (mean head width ± SD: 0.54 ± 0.04 cm; Corbet, 2002).

We quantified behavioral responses of Anax to test the prediction that these predators increase activity and feeding rates at higher temperatures. We did not feed Anax on the day before the assays to standardize hunger levels between individuals. Behavioral assays were performed on 21 and 25 September 2017. Assay tanks were identical as those described in the damselfly assays. Anax were randomly assigned to tanks and treatments. Twelve behavioral assays were performed on each day and temperature treatment was randomized between tanks (control, 21.92 ± 0.26º C, and heated, 27.31 ± 1.02º C). We acclimated Anax individually, in their respective tanks 17 h before the start of the trial. At the start of the assay, 10 damselfly larvae were added to each tank as prey for the Anax. After 5 minutes, a single observer (DNF) measured the position of the Anax in the tank in relation to the grid (measured as the square the individual’s head was found in), if it was moving, and if it was eating. This was repeated every five-minutes for 130 minutes for every individual. Anax were allowed to continue feeding for an additional 130 minutes. After Anax were removed from tanks, we swept plankton nets through each tank several times to remove any remaining damselfly larvae. Each Anax experienced both temperature conditions and the order in which they experienced the treatment was randomized. On the second day of assays, we had to use 5 damselfly larvae per predator as opposed to the initial 10, due to low numbers of Enallagma boreale. This had no effect on the feeding rate of Anax, and prior work has shown that damselfly mortality by Anax is not density dependent (McPeek, 1998).

We tested the prediction that Anax increases activity level and feeding rate with increasing temperature. Feeding rate was calculated as the number of damselflies eaten divided by the time

131 spent within the tank with their prey. We performed a linear mixed effects model for repeated- measures with the function “lme” using the package nmle (version 3.1.131) in R (version 3.3.2) to determine if feeding rate increases with temperature (fixed effect, 2 levels). Individual Anax were the repeated subjects (coded as: random = ~1 | Date/Anax_number). We chose this method to allow for the slope and intercept to vary between individuals (Maindonald & Braun, 2010). Anax activities related to locomotion were correlated, so we first used Principal Components Analysis (PCA) using the function “princomp” (base stats package in R version 3.3.2) to collapse collinear behaviors into principle components. This method also reduces the likelihood of spurious results caused by collinearity among variables (Graham 2003). The function “scale” (base package) was first used transform variables prior to the PCA. This function centers (subtracts the column mean from the corresponding columns) and scales values (divides centered values by their standard deviations and root mean square). Principal component 1 (PC1) explained 71.7% of the variance in locomotor behavior. Number of moves, total distance moved, and distance per move all loaded negatively on this axis (loadings < -0.48; Fig. S4.1). PC1 was used as the dependent variable in a linear mixed effects model for repeated-measures. Similar to above, temperature was included as a fixed effect and Anax individuals nested within date were included as random effect to account for repeated measures on the same individuals.

The order that Anax experienced the temperature treatments had no effect on either feeding rate

(ANOVA: F1,9 = 0.106, p = 0.752), or movement (F1,9 = 1.98, p = 0.19). One Anax (number 6) was dropped from all analyses because it died in the second trial. Anax feeding rate marginally increased in warmer temperatures (feeding rate in warmed treatments: 0.82 ± 0.34 prey/hour versus control: 0.62 ± 0.22 prey/hour; !2 = 2.71, df = 1, p = 0.10; Fig. S4.2). Anax locomotor activity, measured as PC1, was not affected by temperature (!2 = 0.002, df = 1, p = 0.97).

132

Groups LH

1 Distance_per_move

0 Total_distance

Comp.2 (26.81%) Comp.2 Num_moves -1

-2

-3 0 3 Comp.1 (71.71%)

Fig. S4.1. Ordination plot of Anax locomotor activities: total number of moves, total distance moved, and distance per move. Colour of points represent different temperature treatments (L = ambient, purple; H = heated, orange) and shaded ellipses are 95% confidence intervals. PC1 explained 71.71% of the total variation in locomotor activity with distance per move, total number of moves, and total distance loading negatively on this axis.

133

0.9

0.8

0.7

0.6 Feeding rate, prey x hour -1 hour x prey rate, Feeding

L H Treatment

Fig. S4.2 Anax feeding responses to higher temperatures. Feeding rate was measured as the number of prey eaten per hour. Points are means ± SE. Sample size for Anax: n = 11. Treatment temperatures were 23 and 27ºC. References Corbet PS (2002) Stadia and growth ratios of Odonata: A review. International Journal of Odonatology, 5, 45–73. Graham, M. H. (2003). Confronting multicollinearity in ecological multiple regression. Ecology, 84(11), 2809–2815. https://doi.org/10.1890/02-3114 Maindonald, J., & Braun, W. (2010). Multi-level models and repeated measures. In Data Analysis and Graphics Using R: An Example-Based Approach (pp. 303–350). Cambridge University Press. https://doi.org/doi:10.1017/CBO9781139194648.013 McPeek, M. A. (1990). Behavioral differences between Enallagma species (Odonata) influencing differential vulnerability to predators. Ecology, 71(5), 1714–1726. https://doi.org/10.2307/1937580 McPeek MA, Peckarsky BL (1998) Life histories and the strength of species interactions: combining mortality, growth, and fecundity effects. Ecology, 79, 867–879.

134

Appendix IV

Chapter 5 Supporting information

Fig. S5.1. Dragonfly larvae initial size at the beginning of the experiment. Head widths are measurements from n = 20 larvae per species. Dry mass (M) was estimated from head widths (HW) using species- or genus-specific coefficients in Benke et al. (1999) using the allometric scaling equation: M = a HWb. Data for Pachydiplax longipennis was unavailable, so for this species we used coefficients of the closest species morphologically, Celithemis fasciata.

135

1 log(abundance Daphnia) log(abundance

0 AWV Treatment

Fig. S5.2. Boxplots of the log-transformed abundance of the zooplankton prey of dragonfly larvae in the mesocosms. Abundances were measured at the end of the experiment. A = ambient conditions; W = warmed (+3º C above ambient); V = heat waves (+1 and + 5º C above ambient, fluctuating weekly).