Download Document 8137817

Giant steps from small thefts: bypassing constraint by trait co-­‐option by Vanya G. Rohwer A thesis submitted to the Department of Biology In conformity with the requirements for the degree of Doctor of Philosophy Queen’s University Kingston, Ontario, Canada (September 2015) Copyright © Vanya G. Rohwer, 2015 Abstract Species interactions can provide organisms with access to novel traits without having to evolve these traits themselves. Viewing such interactions as an alternative to trait evolution offers a new perspective to understanding species interactions—
organisms can bypass evolutionary constraints and gain access to adaptive traits by co-­‐opting them from other species. I explore the costs and benefits of trait co-­‐option and generate 10 hypotheses for when and where selection might favor the co-­‐option of traits as an alternative to typical trait evolution (Chapter 2). Next, I examine a possible case of trait co-­‐option in nature, involving numerous species of birds that use aromatic seed-­‐material from plants in the genus Eriocephalus to construct their nests (Chapter 3). I describe the bird-­‐Eriocephalus interaction and focus on a single species of bird, the Karoo prinia (Prinia maculosa), which uses large amounts of Eriocephalus in their nests. Although previous work has suggested that both birds and plants benefit from their interaction, prinias gathered large quantities of E. racemosus seed fluff but avoided seeds, providing limited dispersal benefits to the plant. The time and effort invested into gathering fluff suggests that this material benefits prinias. I test two hypotheses for the fitness benefits of Eriocephalus material in bird nests—that chemical compounds reduced levels of nest ectoparasites (Chapter 4) and rates of nest predation by a specialist predator, the rhombic egg-­‐eating snake (Dasypeltis scabra) (Chapter 5). Chemical compounds reduced levels of nest ectoparasites by functioning as either a deterrent to gravid female parasites, or as an ovicide. I found no evidence that chemical compounds reduced rates of nest predation from egg-­‐eating snakes. Overall, these observations ii and experiments suggest that Karoo prinias benefit from co-­‐opting Eriocephalus compounds for use in nest construction, likely at the expense of the plant. Overall, interactions similar to those involving prinias and Eriocephalus are widespread in nature, yet rarely are they viewed as an alternative to trait evolution. Viewing species interactions from the perspective of trait co-­‐option and evolution has the potential to elucidate general patterns for when and which species should interact, broadening our understanding of nature and biodiversity. iii Co-­‐authorship Chapters 2-­‐5 are co-­‐authored manuscripts either in review or in preparation for submission as follows: Chapter 2: Rohwer VG, Martin PR. Evolution by trait co-­‐option. This chapter is in preparation for submission to Trends in Ecology and Evolution. Chapter 3: Rohwer VG, Pauw A, Martin PR. Fluff-­‐thieving birds sabotage seed dispersal. This chapter is in preparation for submission to PLoS ONE. Chapter 4: Rohwer VG, Kozin I, Bonier F, Martin PR. Chemical compounds in Eriocephalus plants increase tree swallow fitness by impeding a naïve nest ectoparasite community and reducing nest predation. This chapter is currently in review at Royal Society Open Science. Chapter 5: Rohwer VG, Martin PR. No evidence that augmenting Eriocephalus compounds reduces predation on Karoo prinia nests from rhombic egg-­‐eating snakes. This chapter is in preparation for submission to Behavioral Ecology. iv Acknowledgements I’d like to thank my advisor, Paul Martin, who I refer to with admiration as “the good dr.” Working with Paul has transformed my interest in bird nests from a childhood hobby into an exciting scientific study. Paul has helped with the development of this dissertation in so many ways—expanding how I think about species interactions, pushing my writing, presentation, experimental design, critical thinking skills, and perhaps most importantly, serving as a role model for how to approach science, teaching, and the human side of academics. Paul’s love of birds, nature, and science is contagious and students immediately recognize this. My family and especially my parents, Sievert and Brigitte, have provided endless support: countless dinners, a place to sleep, and advice during the past years. My father was my first role model in science and continues to be a source of inspiration for how to approach questions in biology. Although my mother isn’t around to see the completion of this dissertation, she would have cheered with the same enthusiasm she gave when, as a kid, I scribbled on paper, hung it on the fridge, and deemed it art. She was a wonderfully positive force. My committee—Chris Eckert, Laurene Ratcliffe, and Scott Lamoureux—
helped focus my thinking about these ideas and provided helpful discussion and feedback during committee meetings and on drafts of this work. I owe many thanks to the people who helped with field work. In particular, Sara Burns and Sharon Zhang for help monitoring and dissecting nests. Tim Cook, Nick Ventor, Jurina le Roux, and Anton Pauw helped make trips to South Africa smooth, enjoyable, and productive. The Queen’s University Biological Station v provided luxurious accommodations, logistical support, and a place to live. This work would not have been possible without support from the American Ornithologists’ Union (VGR), the Society of Canadian Ornithologists (VGR), the Ontario Graduate Scholarship (VGR), the Baillie Family Chair Endowment (PRM), the Canada Foundation for Innovation (PRM), and NSERC Discovery grants (PRM). vi Table of contents Abstract……………………………………………………………………………………………….………...ii Co-­‐authorship……………………………………………………………………………………………......iv Acknowledgements………………………………………………………………………………………..v Table of Contents……………………………………………………………………………………………vii List of Figures………………………………………………………………………………………………...x List of Tables…………………………………………………………………………………………............xi Chapter 1: Introduction…………………………………………………………………………............1 Chapter 2: Evolution by trait co-­‐option…………………………………………………………...5 2.1 Abstract…………………………………………………………………………………………………5 2.2 Introduction…………………………………………………………………………………………..5 2.3 The benefits and costs of co-­‐opting traits……………………………………………......7 2.4 When should species co-­‐opt traits?.................................................................................10 2.5 Trait co-­‐option as a framework………………………………………………………………15 2.6 Conclusions…………………………………………………………………………………………...16 Chapter 3: Fluff-­‐thieving birds sabotage seed dispersal……………………………………20 3.1 Abstract…………………………………………………………………………………………………20 3.2 Introduction………………………………………………………………………………………......21 3.3 Methods………………………………………………………………………………………………...22 3.3.1 Ethical statement……………………………………………………………………………...23 3.3.2 Study site………………………………………………………………………………………….23 3.3.3 Karoo prinia……………………………………………………………………………………..24 3.3.4 Eriocephalus plants…………………………………………………………………………...25 3.3.5 Do the reproductive phenologies of Karoo prinias and Eriocephalus overlap?...........................................................................................................................................27 3.3.5.1 Breeding phenology of Karoo prinias…………………………………………...27 3.3.5.2 Phenology of Eriocephalus…………………………………………………………...28 3.3.5.3 Ease of capitula removal from Eriocephalus………………………………….29 3.3.6 Karoo prinia distance traveled and time spent gathering Eriocephalus material……………………………………………………………………………………………………30 3.3.7 How many Eriocephalus capitula are in Karoo prinia nests?........................30 vii 3.3.7.1 Distribution and amount of Eriocephalus fluff in prinia nests………...31 3.3.7.2 Factors influencing the number of capitula and amount of fluff in Karoo prinia nests ….……………………………………………………………….32 3.4 Results…………………………………………………………………………………………………..33 3.4.1 Do the reproductive phenologies of Karoo prinias and Eriocephalus overlap?…………………………………………………………………………………………………...33 3.4.2 Karoo prinia distance traveled and time spent gathering Eriocephalus material…………………………………………………………………………………………………...33 3.4.3 How many Eriocephalus capitula are in Karoo prinia nests?.......................35 3.5 Discussion.…………………………………………………………………………………………….37 3.5.1 The plant perspective……………………………………………………………………….38 3.5.2 The bird perspective………………………………………………………………………...40 3.6 Conclusions…………………………………………………………………………………………...44 Chapter 4: Chemical compounds in Eriocephalus plants increase tree swallow fitness by impeding a naïve nest ectoparasite community and reducing nest predation.……………………………………………………………………………………………………..54 4.1 Abstract……..…………………………………………………………………………………………54 4.2 Introduction………………………………………………………………………………………....55 4.3 Methods……………………………………………………………………………………………….56 4.3.1 Study site, species, and procedure……………………………………………………57 4.3.2 Do Eriocephalus chemicals reduce ectoparasite infestations?...................57 4.3.3 Do Eriocephalus chemicals increase tree swallow fitness?.........................58 4.4 Results………………………………………………………………………………………………...59 4.5 Discussion……………………………………………………………………………………………59 4.5.1 Mechanism by which Eriocephalus chemicals reduce ectoparasites…..60 4.5.2 Links between parasites and fitness………………………………………………...61 4.6 Conclusions………………………………………………………………………………………….62 Chapter 5: No evidence that augmenting Eriocephalus compounds reduces predation on Karoo prinia nests from rhombic egg-­‐eating snakes…………………..66 5.1 Abstract……………………………………………………………………………………………….66 5.2 Introduction………………………………………………………………………………………...67 viii 5.3 Methods………………………………………………………………………………….…..……….70 5.3.1 Study site………………………………………………………………………………………..70 5.3.2 Finding, monitoring, and collecting nests…………………………………………..71 5.3.3 Assigning predators to depredated nests…………………………………………..72 5.3.4 Field experiment……………………………………………………………………………...74 5.3.5 Captive experiments………………………………………………………………………...75 5.3.6 Statistical analyses—field experiments……………………………………………..77 5.3.7 Statistical analyses—captive experiments………………………………………....81 5.4 Results…………………………………………………………………………………………………..81 5.4.1 Field experiment……………………………………………………………………………....81 5.4.2 Captive experiment…………………………………………………………………………..82 5.5 Discussion……………………………………………………………………………………………..83 5.5.1 Data interpretation from field experiments……………………………………….83 5.5.2 How Eriocephalus compounds could reduce nest predation………………84 5.5.3 Nest predation and selection for earlier breeding………………………………85 5.6 Conclusions…………………………………………………………………………………………...86 Chapter 6: Summary and future research………………………………………………………..96 6.1 Trait co-­‐option and an evolutionary perspective of community ecology…...96 6.2 The predictive power of a trait co-­‐option framework………………………………97 6.3 Describing and understanding bird-­‐Eriocephalus interactions………………….99 6.4 Difficulties and opportunities of the Eriocephalus-­‐bird system………………...101 6.5 Extending back to trait-­‐co-­‐option…………………………………………………………...103 References…………………………………………………………………………………………………….105 Appendix A: Supplemental material for Chapter 3…………………………………………..115 Appendix B: Supplemental material for Chapter 4…………………………………………..128 ix List of Figures Figure 2-­‐1. Evolutionary significance of trait co-­‐option……………………………………18 Figure 2-­‐2. Shifting trade-­‐offs as a result of trait co-­‐option………………………………19 Figure 3-­‐1. Eriocephalus material in Karoo prinia nests…………………………………...45 Figure 3-­‐2. Environmental conditions at the Koeberg Nature Reserve……………...46 Figure 3-­‐3. Development of Eriocephalus seed material and the prinia breeding season……………………………..……………………………..………………………….47 Figure 3-­‐4. Number of capitula removed from Eriocephalus plants over the season……………………………..……………………………..…………………………………..48 Figure 3-­‐5. Distance traveled and time invested in gathering Eriocephalus material……………………………..……………………………..……………………………..………49 Figure 3-­‐6. Comparison of Eriocephalus capitula in bird nests from Koeberg Nature Reserve……………………………..…………………………………………….50 Figure 3-­‐7. Measures of wall thickness for Karoo prinia nests………………………….51 Figure 3-­‐8. Factors influencing the number of Eriocephalus capitula in Karoo prinia nests……………………………..……………………………..……………………....52 Figure 4-­‐1. Number and mass of parasites in tree swallow nests…………………...…64 Figure 4-­‐2. Fitness effects of Eriocephalus chemicals on nesting tree swallows……………………………..……………………………..………………………………...…..65 Figure 5-­‐1. Nest predation thoughout the Karoo prinia nesting cycle…………...….. 88 Figure 5-­‐2. Kaplan-­‐Meier survival curves of Karoo prinia nests with and without Eriocephalus compounds……………………………..…………………..…….89 Figure 5-­‐3. Summary of experiments using captive egg-­‐eating snakes………..…….90 x List of Tables Table 3-­‐1. Significant predictor variables influencing the number of Eriocephalus capitula and amount of Eriocephalus fluff in Karoo prinia nests……..……….…..53 Table 5-­‐1A and B. Best performing models and their parameter estimates for analyses testing the effect of treatment on nest survival…………………….……….91 Table 5-­‐2A and B. Best performing models and their parameter estimates for analyses testing the effect of predator type on nest survival……………….………92 Table 5-­‐3A and B. Best performing models and their parameter estimates for analyses of nests survival for nests depredated by rhombic egg-­‐eating snakes……………………………………………………………………………………………………...94 Table 5-­‐4A and B. Best performing models and their parameter estimates for analyses of nest survival for nests depredated by predators other than rhombic egg-­‐eating snakes……………………………………………………………………………………..95 xi Chapter 1: Introduction Open any textbook of community ecology and you will find chapters devoted to species interactions (Mittelbach 2012). Interactions among species are splendidly diverse—species eat one another, steal from, fight with, enslave, deceive, cultivate, and rely on other species (Stachowicz 2001, Clucas et al. 2008, Mittelbach 2012). At their core, these interactions involve two conceptually different interactions: those driven by food and other limiting resources, and those driven by access to adaptive traits. Interactions involving the exchange of adaptive traits are frequently studied from the perspective of mutualism, commensalism, and parasitism (Mittelbach 2012). These perspectives have undoubtedly improved our understanding of species interactions, but they have also revealed a dizzying array of complexities that challenge their classification (Lawton 1999, Vellend 2010). For example, variation in environmental conditions, the presences and abundance of species, and the age of interacting individuals (e.g., adult vs. juveniles) can shift the fitness outcomes of interactions between mutualism and parasitism (Cushman and Whitham 1989, Bronstein 1994, Palmer et al. 2008), leading some to question how much we gain by the grouping of ecological interactions in the first place (Lawton 1999). Perhaps more importantly, we struggle to predict the conditions favoring when, where, which species will interact and how, when we view community ecology within traditional conceptual frameworks (Lawton 1999, Martin 2014). These challenges have been ascribed to the complexities of biological communities 1 and biodiversity — an inextricable situation unique to, or more pronounced in, the field of community ecology (Simberloff 2004). These difficulties, however, might be overcome with a different framework for studying species interactions, particularly a framework grounded in evolution (Dobzhansky 1964, Vellend 2010). Evolutionary perspectives have transformed other fields of biological study, including the study of life histories (Stearns 1992, Roff 1992), physiology (Sibly and Calow 1986), and behavioral ecology (Krebs and Davies 1997). We might expect an evolutionary framework to provide similar predictive power in the study of species interactions, given that these interactions are ultimately driven and constrained by the same evolutionary forces that govern all of life. Approaching species interactions from an evolutionary perspective offers new ideas for understanding both species interactions and the evolutionary trajectories of traits involved in the interactions (Vellend 2010, Martin 2014). For example, some interactions between species can provide organisms access to adaptive traits as an alternative to evolving these traits themselves. From an evolutionary perspective, this is a tremendous advantage—species can bypass the limitations of their own genetic variation and gain access to novel, adaptive traits that they otherwise would be unlikely to evolve. We find many examples of this in nature: acacias gaining access to the mobility and aggression of biting of ants (Janzen 1966), hermit crabs obtaining stinging cells by placing anemones on their shells (Ross 1971), and, hedgehogs becoming venomous by applying the venom of toads to their spines (Brodie 1977). In all of these examples, individuals are gaining 2 access to traits that they would be unlikely to evolve themselves by co-­‐opting the fully developed traits of other species. Although trait co-­‐option is common in nature, it is rarely studied from the perspective of trait evolution (Eisner et al. 1974). Yet, incorporating an evolutionary perspective to such interactions has the potential to generate testable hypotheses for when, where, and which species should interact. Cases of mutualism, commensalism, and parasitism that involve accessing the traits of other species can all be viewed from the perspective of trait co-­‐option, and we would expect variation in the fitness outcomes of these interactions in the same way that we expect variation in trait evolution in response to different selective pressures. Thus, the perspective of trait co-­‐option has the potential to identify broad patterns and principles governing species interactions —the much sought after “rules” that community ecology seeks. In this dissertation, I explore the evolutionary significance of co-­‐opting traits from other species and report field studies that test the fitness benefits of co-­‐opted traits in nature. At the heart of trait co-­‐option lies the question: when should individuals co-­‐opt traits of other species as an alternative to traditional trait evolution? I discuss the potential costs and benefits that could impede or promote the co-­‐option of traits, then, drawing on this framework, develop hypotheses for when and where species should interact as an alternative to typical trait evolution. Next, I explore an intriguing case of possible trait co-­‐option between plants in the genus Eriocephalus and breeding birds in South Africa. Eriocephalus are aromatic plants, endemic to southern Africa (Njenga 2005), and numerous species 3 of birds use Eriocephalus material to construct their nests (Dean et al. 1990, Tarboton 2011). By using Eriocephalus material, birds gain access to a variety of novel traits (e.g., chemical compounds, insulation) that may increase their reproductive success by reducing nest ectoparasites, nest predation, or improving nest microclimate (Hansell 2000). I explore two possible fitness advantages bird may gain as a result of co-­‐opting Eriocephalus compounds: (i) reduced nest ectoparasites and (ii) lower rates of nest predation from a specialist predator, the rhombic egg-­‐eating snake (Dasypeltis scabra). 4 Chapter 2: Evolution by trait co-­‐option. 2.1 Abstract How do organisms overcome evolutionary constraints to gain swift access to novel traits? Answers to this question typically focus on evolutionary processes acting within species. As an alternative, organisms may access novel and sometimes complex traits rapidly by co-­‐opting the traits of other species. Although trait co-­‐
option is common in nature, these interactions among species are rarely viewed from the perspective of trait evolution. Viewing the co-­‐option of traits from this perspective can improve our understanding of species interactions and trait evolution as a whole by allowing us to predict when and where the co-­‐option of traits should occur as an alternative to typical trait evolution, and how different species should interact in the face of varying environmental and evolutionary challenges. 2.2 Introduction The evolution of novel adaptations can be painfully slow. Limited genetic variation (Schluter 1996, Kellermann et al. 2009, Kirkpatrick 2010, Salazar-­‐Ciudad and Marín-­‐
Riera 2013), weak selection (Frankino et al. 2005, Agrawal and Stinchcombe 2009, Conner 2012), and the constraints of evolutionary history (e.g., Futuyma et al. 1995, Schluter 1996) can slow or impede the evolution of adaptive traits. The availability of genetic material for selection to act upon is further constrained by genes that control more than one trait (pleiotropy), genes that depend on other genes to create traits (epistasis), the slow rate of mutations, and the fact that most mutations are 5 deleterious (Eyre-­‐Walker and Keightley 2007, Kirkpatrick 2010). Additionally, evolving novel, complex traits can be especially difficult if they require phenotypic changes in multiple traits because mutations must be coordinated among different genes, and these genes must be linked in order to avoid disruption by recombination in subsequent generations (Leimar et al. 2012). All of these constraints can slow or even halt the speed of adaptive evolution. How can organisms overcome these constraints on evolution? Mutations at loci of large effect have the potential to break the constraints on the direction and rate of evolution (Bradshaw and Schemske 2003); however, these mutations can have dramatic, negative effects on fitness (Orr and Coyne 1992). Phenotypic plasticity, maternal effects, and other epigenetic effects can help organisms match their phenotypes to the environment, overcoming some of the constraints of genetic evolution (Mousseau and Fox 1998, Pigliucci et al. 2006, Johnson and Tricker 2010). Although these processes may promote adaptive evolution through genetic assimilation (West-­‐Eberhart 2003), they may also prevent trait evolution by optimizing phenotypes without genetic change (Price et al. 2003). Additionally, these mechanisms do not overcome the constraints on evolving completely novel traits and evolving them quickly. An alternative mechanism for overcoming the constraints on trait evolution is for organisms to co-­‐opt traits from other species. Trait co-­‐option occurs when individuals of one species use the traits of another species to their fitness advantage, rather than evolving the traits themselves. Many, perhaps most, species on earth co-­‐
opt traits of other species to improve their fitness, and these interactions are 6 commonly studied in community ecology under the topics of mutualism, commensalism, symbiosis, and parasitism (Stachowicz 2001, Mittelbach 2012). Interactions involving trait co-­‐option, however, are rarely considered from the perspective of trait evolution, despite a recent focus on functional traits in community ecology (McGill et al. 2006). Here, we outline how viewing trait co-­‐
option from the perspective of trait evolution can improve our understanding of both trait evolution and species interactions within communities by increasing our ability to predict when and where selection should favor trait co-­‐option over typical trait evolution, and when, how, and which species will interact. We first discuss the fitness consequences of trait co-­‐option from the perspective of trait evolution, and then expand these ideas to generate predictions for trait co-­‐option and species interactions in nature. 2.3 The benefits and costs of co-­‐opting traits The benefits of co-­‐opting, rather than evolving, traits can be enormous. (1) Co-­‐
opting novel traits can allow organisms to access traits that would be unlikely to ever evolve, and provide access to an almost infinite array of traits, depending on the availability of species. For example, hermit crabs co-­‐opt complex nematocysts (the stinging cells at the tips of sea anemone tentacles) — a trait unknown in crustaceans — and use these as an effective defense against predators (Ross 1971). Similarly, plants co-­‐opt the active and controlled mobility of animals for directed seed dispersal (Howe and Miriti 2004), a trait unlikely to evolve in plants. (2) Co-­‐
opting novel traits can allow organisms to jump fitness peaks without having to endure fitness costs, or "valleys", associated with evolving the traits themselves (Fig. 7 1). Such leaps—conceptually similar to mutations of large effect—could be especially effective if the organism can match co-­‐opted traits with the environmental challenges confronting them, thus eliminating the random chance involved in acquiring traits through genetic evolution (Brooks and Mariscal 1986, de Roode et al. 2013). (3) Co-­‐opting traits can allow organisms to accelerate the speed of phenotypic change, outpacing co-­‐evolving species in antagonistic evolutionary races (Brockhurst et al. 2014) and allowing species to adapt to periods of rapid environmental change. Such an advantage could be especially important when organisms face antagonistic species that evolve rapidly (e.g., parasites and diseases), periods of rapid climatic change, or newly introduced predators, pathogens, or parasites. (4) Co-­‐opted traits can allow the development and specialization of traits that were previously constrained by functional or allocative trade-­‐offs (Fig. 2). For example, plants that co-­‐opt the defensive traits of ants (e.g., directed biting and stinging) are then open to shift their resources away from other forms of defense (e.g., production of chemical or physical deterrents) and re-­‐allocate these resources to growth or reproduction (Janzen 1966, Mayer et al. 2014). This adaptive shift in the balance of functional or allocative trade-­‐offs can alter the evolutionary trajectories of other traits (Fig. 2-­‐2), providing advantages in the face of other selective pressures, such as competition. Although the benefits of co-­‐opting traits can be large, the costs of co-­‐opting, rather than evolving, traits, can also be significant. (1) Trait co-­‐option can lead to a dependence on other species (Galetti et al. 2013), particularly if it reduces the likelihood of evolving other solutions to environmental challenges. Relying on other 8 species for access to adaptive traits can result in the loss of existing traits that serve a similar purpose, and further lead to a reduction in genome size (Morris et al. 2012) and reduced potential for future adaptive evolution. Additionally, depending on other species for co-­‐opted traits can tie the fate of one species to another, such that the loss of trait-­‐donor populations (co-­‐optees) can lead to the decline or loss of populations of co-­‐opters (Colwell et al. 2012). (2) Trait co-­‐option involves interactions among species with different fitness goals, increasing the likelihood of conflict and fitness costs (Queller and Strassmann 2009). If fitness interests between co-­‐opter and co-­‐optee do not align (e.g., the co-­‐optee suffers fitness costs as a result of trait co-­‐option), then selection on co-­‐optees should favor reduced interactions with co-­‐opters, potentially favoring traits that are more difficult to access, or traits that are less effective once co-­‐opted. (3) Co-­‐opted traits may underperform traits that evolved through genetic evolution because their development and use occurs in two different organisms experiencing different selective pressures. Thus, selection may be unable to optimize the function of co-­‐
opted traits for the co-­‐opter because it must operate indirectly on the co-­‐optee. The underperformance of co-­‐opted traits should be most pronounced when the function of the trait in the co-­‐opter differs from the function in the co-­‐optee. For example, seed material used for wind dispersal by plants is co-­‐opted for thermoregulation in the nests of birds (Hansell 2000). An alternative to co-­‐opting traits that can overcome some of the costs of trait co-­‐option is to co-­‐opt genes, which many organisms do (Schönknecht et al. 2015). Co-­‐opting genes from other species avoids the dependence of one species (co-­‐opter) 9 on the other for access to traits (cost 1), removes conflicting fitness goals across species (cost 2), and can improve trait function by capturing the genes, development, and trait within the same organism (cost 3). Co-­‐opting genes to create new traits, however, requires that all genes involved in the development of the trait be co-­‐opted, or be present already in the co-­‐opting organism, and that co-­‐opted genes function well in their new genetic environment. Such requirements might be unlikely for complex traits that are co-­‐opted from phylogenetically divergent species. In summary, co-­‐opting the traits of other species can provide exceptional fitness benefits for organisms by allowing immediate access to novel, adaptive traits without having to evolve these traits themselves. The novelty of co-­‐opted traits and the speed at which they can be acquired can disrupt co-­‐evolutionary interactions with antagonistic predators, parasites, or pathogens, and can provide access to key traits when environmental conditions change too rapidly for genetic evolution to keep up. On the other hand, relying on other species for access to adaptive traits can be costly, especially when organisms depend on co-­‐opted traits for survival, or when conflict between species limits the access to, or effectiveness of, co-­‐opted traits. 2.4 When should species co-­‐opt traits? Considering the costs and benefits of trait co-­‐option relative to trait evolution, we can predict when organisms should co-­‐opt the traits of other species instead of evolving the traits themselves. Below we outline ten hypotheses for when trait co-­‐
option should be favored over traditional trait evolution. 10 (1) Strong and novel selective pressures. Trait co-­‐option should occur more often in response to strong, novel selective pressures because these favor rapid adaptive solutions that should be more difficult to obtain by genetic evolution. Strong, novel selective pressures occur more frequently during periods of rapid environmental change, such as rapid bursts of climatic change, human-­‐modification of habitats (e.g., urbanization), and the introduction of invasive species. In addition, we should expect trait co-­‐option to occur more commonly when organisms occupy novel, challenging environments relative to ancestral conditions for the species or clade. Novel, challenging environments should require new traits that are less likely to have evolved previously in recent ancestors, and are less likely to evolve by genetic evolution. (2) Coevolution and the Red Queen. Trait co-­‐option should be most common in response to antagonistic, co-­‐evolving species because these interactions often involve strong and persistent selection in the form of an evolutionary arms race (Brodie and Brodie 1999). Such strong, persistent directional selection can reduce genetic variation over time and limit the potential for future trait evolution (Hall et al. 2011), thus favoring the co-­‐option of traits. For species interactions that follow Van Valen’s Red Queen dynamics (i.e., that species must continuously evolve in the face of biotic interactions; Van Valen (1973)), co-­‐opted traits offer the potential to disrupt co-­‐evolving interactions (Brockhurst et al. 2014), providing a rest for the weary queen. 11 (3) Completely novel traits. Species should be more likely to co-­‐opt traits that are completely novel because these traits have the potential to provide the greatest fitness benefits to the co-­‐opter, and are less likely to evolve by genetic evolution. Co-­‐
opting novel traits can allow species to jump fitness "valleys" and access adaptive solutions that would be otherwise impossible, providing the equivalent of an adaptive mutation of large effect. In the case of biotic interactions, the co-­‐option of completely novel traits increases the chance of disrupting interactions among antagonistic, co-­‐evolving species (Brockhurst et al. 2014). For example, penicillin — a completely novel trait co-­‐opted from the fungi Penicillium notatum — severely altered the interactions between humans and many bacterial diseases, and created a lag in the evolution of counter adaptations in many species of bacteria (Davies and Davies 2010). (4) Divergent sources. Given that species should be more likely to co-­‐opt completely novel traits (3), trait co-­‐option should be most common among species that share little to no recent evolutionary history because divergent species are the most likely source for evolutionary novelty. In contrast, closely related species share more ecological traits, developmental pathways, and metabolic byproducts (Losos 2008), providing fewer traits for co-­‐option and reducing the potential benefits of trait co-­‐option relative to the costs. (5) Slow evolvers. Trait co-­‐option should be most common in species that evolve slowly (such as large-­‐bodied species with long generation times) because these species have reduced opportunities to generate new traits quickly. Slow evolvers 12 should be more likely to suffer in interactions with rapidly evolving parasites and diseases (2); thus we should expect slow evolving organisms to co-­‐opt traits most commonly as a response to fast evolving diseases and parasites. (6) Multiple benefits. Co-­‐opted traits may be especially common when they provide multiple benefits, allowing co-­‐opters to solve different challenges with a single trait, and increasing the likelihood of trait co-­‐option arising as a byproduct of other interactions. For example, many species receive multiple benefits from co-­‐opted traits, such as insulation and defense (Hansell 2000), and camouflage and defense (Stachowicz and Hay 1999). As the number of benefits provided by co-­‐opted traits increases, so too should the opportunities for selection to favor trait co-­‐option. (7) Behaviors facilitating co-­‐option. Behaviors that reduce the costs of acquiring and using traits should increase the likelihood of trait co-­‐option. Behaviorally co-­‐
opted traits can be acquired in response to environmental stimuli (e.g., scent of a predator, Brooks and Mariscal 1986), reducing the cost of co-­‐option to high-­‐risk periods (Fig. 2; McClintock and Janssen 1990). Additionally, behaviors that already exist in a species’ repertoire that facilitate the co-­‐option of traits should increase the likelihood of trait co-­‐option (Arnold 1992, Duckworth 2009). For example, allopreening behavior in monkeys may facilitate both the acquisition of chemical compounds from diverse sources and the application of these co-­‐opted traits to their pelage (Weldon et al. 2003, de Roode et al. 2013). (8) Matching trait to challenge. Trait co-­‐option should be most common in species with greater cognitive ability that would allow them to match co-­‐opted traits with 13 specific environmental challenges. Enhanced cognition could allow individuals to link trait function in one context with potential trait function in another context, enabling individuals to co-­‐opt specific traits for specific challenges (de Roode et al. 2013). Enhanced cognition could further increase the effectiveness or number of co-­‐
opted traits if individuals test or learn which traits perform best (Kerr 2007), or if the co-­‐option of traits is culturally inherited (Mathew and Perreault 2015). All of these characteristics of cognitively advanced organisms have the potential to change the acquisition of co-­‐opted traits from a chance event to a more directed process, increasing the potential benefits of co-­‐opting traits over typical trait evolution. (9) Alignment of fitness benefits. Trait co-­‐option should be more likely to occur and persist when both the co-­‐optee and co-­‐opter benefit from trait co-­‐option. The alignment of fitness consequences between co-­‐opter and co-­‐optee may increase cooperation and co-­‐dependency among interacting species, reducing the costs of trait co-­‐option and favoring the persistence of the interaction (Queller and Strassmann 2009). Alternatively, such mutually beneficial interactions will be open to cheating should one species benefit by reducing the fitness benefits provided to the other species (Fredrickson 2013), possibly making the alignment of fitness benefits variable over space or time. (10) Alignment of trait function. Trait co-­‐option should be more likely to occur when the co-­‐opted trait is used for the same function in both species (i.e., the co-­‐
optee and co-­‐opter) because trait function can be optimized for both species by selection acting on the co-­‐optee. Co-­‐opted traits should also be more likely to 14 provide an adaptation for the same selective pressure for which it evolved, rather than a novel selective pressure. For example, secondary chemicals produced by plants to reduce fungal attack (e.g., Tewksbury et al. 2008) should be more likely to be co-­‐opted to mediate interactions with the same or other fungi, rather than for interactions with other taxonomic groups or completely different functions. This hypothesis further predicts that the co-­‐option of traits should be more likely when the interacting organisms share selective challenges, even though we expect traits to be co-­‐opted more often from evolutionarily divergent species (see 4 above). 2.5 Trait co-­‐option as a framework Species interactions are traditionally studied from the perspectives of the fitness consequences of the interactions: mutualism, commensalism, parasitism, and predation (Mittelbach 2012). These perspectives include two conceptually different types of interactions: (i) those driven by food and limited resources, and (ii) those driven by access to adaptive traits. The perspective of trait-­‐co-­‐option focuses on this latter group of interactions, and generates testable hypotheses for when, where, and how species should interact. Rarely (if ever) would a traditional view of species interactions generate any of the 10 hypotheses that we described above. And yet these hypotheses predict which species should co-­‐opt traits and from whom, which selective pressures are more likely to lead to trait co-­‐option over trait evolution, and how species interactions involving trait co-­‐option can influence, and be influenced by, typical (genetic) evolution of traits. The perspective of trait co-­‐option shifts the focus away from the fitness outcomes of interactions, towards the traits involved in the interaction to provide a 15 more complete understanding of trait evolution and species interactions over time. From the perspective of trait evolution, co-­‐opted traits are removed from the selective background in which they evolved and are introduced into the novel selective background of the species that co-­‐opted them. These dynamic selective backgrounds can create indirect patterns of selection and alter the evolutionary trajectories of co-­‐opted traits. In cases such as these, understanding trait evolution will require an understanding of species interactions and trait co-­‐option. 2.6 Conclusions Our understanding of trait evolution has progressed from the early perspectives of strictly genetic change, to a greater appreciation of phenotypic plasticity and epigenetic effects (Orr 2005, Price et al. 2003, Johnson and Tricker 2010). These views of trait evolution share a common theme: reduced emphasis on purely genetic changes within organisms. The perspective of trait co-­‐option provides yet another outward extension of these views. Indeed, trait co-­‐option highlights the ingenuity of natural selection that makes evolution such a creative force—organisms bypassing the limits of their own genetic variation by co-­‐opting the traits of other species. The perspective of trait co-­‐option not only changes how we think about overcoming constraints on trait evolution, it also changes how we approach species interactions. Viewing species interactions as a mechanism to gain access to novel, adaptive traits allows us to generate predictions for when selection might favor the co-­‐option of traits as an alternative to traditional trait evolution, and when, which, and how species should interact in nature. Our hope is that viewing trait co-­‐option as an alternative to trait evolution will enhance our understanding of both trait evolution 16 and species interactions, and further integrate evolutionary theory with community ecology with a productive outcome (Vellend 2010, Heath and Stinchcombe 2014, Martin 2014). 17 dt
r
x ai
ts
Amphipod
fitness
y-axis
Pte
ro
po
z-a
is
Amp
hipod
traits
x-ax
is
Figure 2-­‐1. To visualize the evolutionary potential of co-­‐opted traits, we draw on a permutation of Sewall Wright’s influential metaphor of the fitness landscape (Wright 1932). We illustrate the amphipod’s “predator avoidance” traits, prior to co-­‐
option, with a normal distribution along the x-­‐axis. Before co-­‐option, the amphipod’s predator avoidance trait space is confined to traits possessed only by the amphipod. When amphipods co-­‐opt the chemical compounds of pteropods, they gain access to a novel predator defense, which dramatically increases “predator avoidance” trait space. We illustrate the increase in trait space with the z-­‐axis, representing traits of the pteropod. Pteropods vary in body size and presumably chemical compounds (McClintock and Janssen 1990); we represent this variation in co-­‐opted traits with a continuous z-­‐axis. As a result of co-­‐opting the chemical defense of pteropods, amphipods experience a reduction in predation so dramatic that we visualize it as a saltatorial jump to a higher peak on the fitness landscape. Importantly, amphipods gain access to these traits without maladaptive evolution (i.e., reduced fitness resulting from the crossing of the fitness valley that separates fitness peaks). Thus, the co-­‐option of traits from pteropods results in both the expansion of available trait space, and a dramatic increase in fitness for amphipods. 18 B
120
90
60
Predator defense
effective
Fitness
Amphipod swimming
speed (mm/sec)
A
Swimming speed
fast
optimum
with
pteropods
optimum
without pteropods
30
ineffective
slow
Trait value
Figure 2-­‐2. Trait co-­‐option can change how selection acts on existing traits, with the potential to reduce the constraints that would otherwise limit trait evolution. We illustrate changes in trait function as a result of trait co-­‐option using the same amphipod-­‐pteropod example used in Figure 1. Amphipods that co-­‐opt the chemical defenses of pteropods swim slower than amphipods without pteropod defenses (Fig. 2-­‐2A) (McClintock and Janssen 1990). The size of pteropods co-­‐opted by amphipods is, on average, much smaller than the size of free-­‐swimming pteropods, suggesting that amphipods selectively abduct the smallest pteropods to reduce the hydrodynamic costs of swimming with them (McClintock and Janssen 1990). While the co-­‐option of pteropod chemical defense results in a functional trade-­‐off with amphipod swimming speed, the consequence of trait co-­‐option for other amphipod traits, including behaviors, ecological traits, and traits used in defense, remains unknown. We illustrate how trait co-­‐option can shift existing trade-­‐offs in Figure 2-­‐
2B; in this figure the x-­‐axis represents trait values of both swimming speed and predator defense, but optimal values of these traits lie at opposite ends of the axis, illustrating a trade-­‐off. The apparent costs of slower swimming speed may be offset by the benefits of increased chemical defense, allowing amphipods to spend more time foraging, occupy higher quality habitat, and move more slowly. We should expect co-­‐opted traits to shift the equilibrium values of many traits bound by functional or allocative trade-­‐offs, although such interactions between co-­‐opted and evolved traits have been rarely studied. 19 Chapter 3: Fluff-­‐thieving birds sabotage seed dispersal. 3.1 Abstract Positive interactions among species can be important for maintaining local diversity, yet we have a poor understanding of most interactions. Here we provide detailed natural history data on a complex bird-­‐plant interaction in South Africa thought to provide benefits to both birds and plants. Many birds use fluffy, aromatic Eriocephalus seed material to construct their nests, potentially dispersing seeds for the plant. We focus on the Karoo prinia (Prinia maculosa) — a small, abundant passerine bird that uses large amounts of Eriocephalus racemosus fluff in nest construction — and show that the breeding phenology of prinias overlaps the phenology of E. racemosus seed material production. Prinias invested heavily in gathering Eriocephalus seed material, spending 5 of their median 6-­‐day nest construction period adding seed material to their nests. Some individuals travelled as far as 344 meters to gather Eriocephalus material and spent over two minutes a trip plucking seed material. Yet, few Eriocephalus capitula (seed-­‐heads that contain multiple seeds) were incorporated into prinia nests because prinias actively avoided Eriocephalus capitula when collecting the surrounding fluff. The mass of fluff in the average prinia nest would have contained 579 Eriocephalus capitula while on the plant; however, prinia nests contained only 6.6 capitula per nest, on average. The avoidance of capitula by prinias suggests that prinias may provide limited or no dispersal benefits to Eriocephalus plants. In contrast, the large amounts of Eriocephalus fluff in prinia nests, and the effort that prinias invest in gathering it, suggest that prinias may benefit by constructing their nests with Eriocephalus 20 material. We outline alternative hypotheses for possible fitness benefits that Eriocephalus material could provide prinias and other birds, and await future experiments that can test among them. 3.2 Introduction Positive interactions among species can play important roles in determining the presence and abundance of species in local communities (Paine 1966, Thompson 1994, Stachowicz 2001, Bruno et al. 2003, de Roode et al. 2013). For example, plants can gain access to directed movement through their interactions with mobile animals, thus facilitating sexual reproduction, gene flow, and dispersal of their offspring to suitable sites (Howe and Miriti 2004). Similarly, many animals co-­‐opt traits from other species, including chemical compounds, insulative materials, and structures for protection against predators (Wimberger 1984, Stachowicz and Hay 1999, Hemmes et al. 2002, Dumbacher et al. 2004, Chapuisat et al. 2007, Bravo et al. 2014). In all of these cases, interactions among species provide traits or services that can positively influence the fitness of individuals and the sustainability of populations, potentially increasing local diversity (Stachowicz 2001, Galetti et al. 2013). Despite the importance of positive interactions among species in community ecology, we know little about most interactions. The goal of this paper is to describe the natural history of a bird-­‐plant interaction from a poorly known and complex system in southern Africa. In this system, many species of birds use seed material from plants in the genus Eriocephalus to construct their nests (Dean et al. 1990). Eriocephalus plants are highly aromatic (Njenga 2005) and produce many small capitula (seed-­‐heads that 21 contain several individual seeds) that are surrounded by white, cotton-­‐like fluff (Fig. 3-­‐1). By using Eriocephalus seed material in nest construction, birds gain access to traits (e.g., chemical compounds, insulative material) that may increase their reproductive success by reducing nest ectoparasitism or predation, or by improving nest microclimate. Similarly, Eriocephalus plants may benefit by having their seed material incorporated into the nests of birds through seed dispersal and increased seed survival and germination (Dean et al. 1990, Vander Wall and Longland 2004). Here we provide detailed natural history data on the interactions between Eriocephalus plants and birds in the Western Cape Province, South Africa, with a focus on phenology, bird behavior, and seed dispersal. We focus on the most abundant species of Eriocephalus at our site — E. racemosus — and the most abundant bird species breeding at our site — the Karoo prinia (Prinia maculosa). In addition to their abundance, Karoo prinias use large amounts of Eriocephalus seed material to construct their nests, and are thus expected to be one of the most important bird species interacting with E. racemosus at our site. We first describe the breeding phenologies of Karoo prinia relative to the phenology of seed material production of E. racemosus. Second, we document the time spent and distance travelled by prinias to gather Eriocephalus material for their nests. Third, we quantify the number of Eriocephalus capitula in Karoo prinia nests to assess whether nests are effective agents of dispersal for Eriocephalus seeds. Finally, we use natural history data to develop alternative hypotheses for potential fitness advantages for birds that use Eriocephalus material in nest construction. 3.3 Methods 22 3.3.1 Ethical statement Protocols for this research were approved by the Queen’s University Animal Care Committee (Martin-­‐2013-­‐013), Cape Nature (AAA005-­‐00219-­‐0028), and the Koeberg Nature Reserve (permission granted 1 March 2013). We conducted this work on private land (Eskom’s Koeberg Nature Reserve), with permission from the landowner. 3.3.2 Study site The Koeberg Nature Reserve (33° 41’ S, 18° 27’ E) is ~35km north of Cape Town, South Africa, along the south Atlantic coast. Koeberg has a Mediterranean climate with warm dry summers and cool wet winters; most precipitation falls during the winter months (Fig. 3-­‐2). Temperatures at Koeberg are moderated by the Atlantic Ocean. Winter temperatures range from 10–20°C, while in summer, cold waters of the Bengula current keep temperatures between 15–25°C (Goldblatt and Manning 2002). Winds are strong (year-­‐round average wind speed 3–4m/s), persistent, and typically from the south, further cooling temperatures during the summer (Goldblatt and Manning 2002). Two dominant types of vegetation occur at Koeberg: (i) dune thicket-­‐vegetation, a dense often impenetrable tangle of shrubs, and (ii) sand plain vegetation, which tends to be more open with isolated shrubs. Both vegetation types are generally short (< 2m), and in addition to E. racemosus, dominant plants include: Passerina vulgaris, Chrysanthemoides incana, Rhus leavigata, Euclea racemosa, Trichocephalus stipularis, Nylandtia spinosa, and multiple species of Restios (Manning 2007). Koeberg supports a diversity of small passerine birds, many of which use Eriocephalus material to construct their nests 23 (Tarboton 2011); in addition to Karoo prinias, common breeding birds include: grey-­‐backed cisticola Cisticola subruficapilla, southern double-­‐collared sunbird Cinnyris chalybeus, chestnut-­‐vented tit-­‐babbler Sylvia subcaeruleum, Cape white-­‐eye Zosterops capensis, Cape bulbul Pycnonotus capensis, Cape weavers Ploceus capensis, white-­‐backed mousebird Colius colius, yellow canary Crithagra flaviventris, and common waxbill Estrilda astrild. Most of these species nest in low shrubs, share the challenges of maintaining consistent nest temperatures, and have similar nest predators. Nest predators at Koeberg include a diversity of birds (pied crow Corvus albus, fiscal shrike Lanius collaris, bokmakierie Telophorus zeylonus), mammals (Cape grey mongoose Galerella pulverulenta, Southern African vlei rat Otomys irroratus, four-­‐striped grass mouse Rhabdomys pumilio), and snakes (rhombic egg-­‐
eater Dasypeltis scabra, boomslang Dispholidus typus, mole snake Pseudaspis cana, Cape cobra Naja nivea). Field observations of depredated nests suggest that rhombic egg-­‐eating snakes and pied crows are the dominant nest predators of small passerines (Lloyd et al. 2001, Nawalga et al. 2004, Martin et al. 2011, Bates and Little 2013, VGR pers. obs.). Brood parasites (e.g., Klaas’s cuckoo Chrysococcyx klaas, pin-­‐tailed whydah Vidua macroura) are present but uncommon, and we found no brood parasite eggs in over 200 nests of common small passerines at Koeberg. 3.3.3 Karoo prinia Karoo prinias are small (~10g), non-­‐migratory passerines (Passeriformes: Cisticolidae) native to southern Africa and have life-­‐histories similar to many tropical passerines with high adult survival and low nesting success (Peach et al. 2001, Hockey et al. 2005, Lloyd et al. 2014). They are sexually monochromatic, have 24 long-­‐term pair bonds, and maintain year round territories that are aggressively defended during the breeding season (Rowan and Broekhuysen 1962). Karoo prinias are common breeders throughout Koeberg, often placing their nests low to the ground (<1 meter)(Nalwanga et al. 2004) in Passerina shrubs and Restios. Females are the primary builders of the nest’s grass frame, and both males and females help line the nest with downy plant material, often bringing Eriocephalus fluff when it is available (Rowan and Broekhuysen 1962, VGR pers. obs.). Females lay 1 egg per day, but may delay egg deposition during periods of cold weather. Only females incubate and males make few feeding visits to incubating females (Chalfoun and Martin 2007). Both males and females brood newly hatched young (usually until nestling day 7), feed young, and provide post-­‐fledging care for 2-­‐3 weeks after young leave the nest (Rowan and Broekhuysen 1962). Pairs are socially monogamous. Extra-­‐pair copulations are probably rare because the seminal vesicles of Karoo prinias are small (VGR pers. obs.), like those found in species with low extra-­‐pair paternity and sperm competition (Birkhead et al. 1993). Karoo prinias face several challenges to reproduction. Nest predation, extreme weather, brood parasites, and infestations of nest ectoparasites can cause partial or entire brood loss, or reduce nestling condition (Rowan and Broekhuysen 1962). At Koeberg, nest predation is the primary cause of nest failure and Karoo prinias face some of the highest daily nest predation rates recorded for passerine birds (7.5% daily nest predation; Nalwanga et al. 2004, Martin et al. 2006). 3.3.4 Eriocephalus plants 25 Eriocephalus (Family: Asteraceae) are highly aromatic, perennial shrubs native to southern Africa (Njenga 2005). In the most recent taxonomic treatment of this group, Müller et al. (2001) recognized 32 species within the genus Eriocephalus based on morphology. Two species of Eriocephalus occur at our study site: E. racemosus and E. africanus. E. racemosus is far more common than E. africanus and our analyses of Eriocephalus phenology, material use in prinia nests, and prinia-­‐
Eriocephalus interactions, refer to E. racemosus exclusively. E. racemosus is common in open sand plain habitats (often on drier, sandy ridges) at low elevations (<100m above sea level) primarily along the southern coast of South Africa from Lambert’s Bay to Port Elisabeth (Müller et al. 2001). Plants reach ~2 meters height at maturity and often have irregular growth patterns and flimsy branches. Flowering occurs in spring (August–September at Koeberg). Typical of Asteraceae, several flowers are grouped together into a capitulum surrounded by involucral bracts. The capitulum (seed head) is only ~2 mm in diameter and contains only ~3 white ray florets and ~7 purple disc florets. Unusually for Asteraceae, the seeds are retained inside the capitulum, which is dispersed as a single unit. After flowering, very long silky trichomes grow from the involucral bracts surrounding the capitulum (Fig. 3-­‐1). The pappus, which normally aids in wind dispersal in the Asteraceae (e.g., dandelions), plays no role in seed dispersal. Mature E. racemosus plants produce hundreds to thousands of capitula, giving plants an overall pale appearance (Fig. 3-­‐1). Capitula are thought to be dispersed primarily by wind, with their fluff functioning as a dispersal aid (Ausberger 1986). Several aspects of Eriocephalus life history remain unknown, such as plant life span and age 26 at first flowering. Observations of short (~50 cm), sparsely vegetated plants with ~25-­‐50 capitula suggests that some plants may flower after 1-­‐2 years of growth. These plants had between one and three main stems of brown-­‐grey bark, indicating that this wood was at least 1 year old (Müller et al. 2001). For our study we distinguish two parts of Eriocephalus seed material: the capitulum (seed head) and the fluffy exterior surrounding the capitulum (Fig 3-­‐1). We refer to the fluffy exterior as “fluff” and to both the capitulum and fluff as “seed material”. We distinguish between these components of a capitulum because later observations of prinia nests suggested that prinias gathered primarily fluff. While a single capitulum contains several individual seeds (like many plants in the family Asteraceae) (Müller et al 2001), we counted capitula only because this is the part of the plant with which birds interact. 3.3.5 Do the reproductive phenologies of Karoo prinias and Eriocephalus overlap? To compare the reproductive phenologies of Karoo prinias and Eriocephalus, we monitored breeding activity of prinias and the development of Eriocephalus seed material during the austral spring (August-­‐November) of 2013. 3.3.5.1 Breeding phenology of Karoo prinias We found Karoo prinia nests by watching adults carry material to their nests and by searching suitable habitat. Upon finding a nest, we marked its location with a handheld GPS receiver (GPSmap 60, Garmin International Inc., Olathe, Kansas) and checked it every three days (Martin and Roper 1988, Martin and Geupel 1993, 27 Martin et al. 2000). We recorded or estimated first egg date for all nests using the following methods: nests found during the building stage were monitored until the first egg was laid; for nests found during the laying period, we back-­‐counted one egg per day to the first egg (Karoo prinias typically lay one egg per day until the clutch is complete and begin incubation on the day that the last egg is laid; Rowan and Broekhuysen 1962); for nests found during incubation, we either back-­‐counted from the hatch date (by 14 days for incubation plus the appropriate number of days, based on clutch size, assuming that females laid one egg per day; Rowan and Broekhuysen 1962) or, for nests that did not hatch, we took the midpoint between the least and most advanced possible stage of incubation (Martin et al. 2000). 3.3.5.2 Phenology of Eriocephalus We surveyed three sites with Eriocephalus plants in the Koeberg Nature Reserve to monitor the development of fluff; sites were separated by 500–1500 meters, and each site contained ~500 plants. We scored phenology for over 100 plants at each site using eight categories that ranged from least to most developed: bud, early flower, peak flower, late flower, sparse fluff, early fluff, medium fluff, and thick fluff (see Table 1 in Appendix A for descriptions of categories). We examined all of the flowers/capitula on each plant and then categorized the plant into the phenological category that characterized >50% of its flowers/capitula. We visited sites once every 10±1 days for 86 days (starting 24 August 2013, ending 18 November 2013), and stopped when all plants in a patch had thick fluff and had lost roughly half of their capitula. 28 3.3.5.3 Ease of capitula removal from Eriocephalus We estimated the ease by which Eriocephalus capitula could be removed from plants as a function of season by pulling on the fluff of capitula with forceps at three different dates (26 August; 1 October; 3 November), corresponding to early, middle, and late breeding times for Karoo prinias in 2013. We simulated fluff-­‐picking behavior of prinias using forceps that had a surface area similar to the beaks of prinia and other small passerine birds. During each assessment, we plucked at the fluff of 100 Eriocephalus capitula from 10 different plants (1000 capitula in total) and counted the number of plucks that resulted in the removal of a capitulum. We marked all plants with aluminum tags and returned to each plant for all subsequent picking trials. We targeted capitula from the entirety of each plant (rather than select branches), in case capitula on some branches matured faster than others. We tested for differences in the number of capitula removed from plants between each trial using a linear mixed effect model in R (R Core Team 2014). The number of capitula removed was our dependent variable, date of each capitula-­‐
removal trial was our predictor variable, and individual plant id was a grouping variable to account for multiple measures originating from the same plant. Prior to analysis, we transformed the number of capitula removed using [natural logarithm(number of seeds removed+1)]. We checked that model residuals did not deviate from normality using a Shapiro-­‐Wilk test, and that residual model variance did not differ significantly between time intervals using Bartlett’s test, following Zuur et al. (2009). 29 3.3.6 Karoo prinia distance traveled and time spent gathering Eriocephalus material We measured the distance prinias travelled to gather Eriocephalus material by watching focal birds gather material and then return to their nests. Using a GPS receiver, we measured and compared the straight-­‐line distance (meters) from the nest to the plant from which prinias gathered material versus the distance from the nest to the closest Eriocephalus plant that was at a similar phenological stage. For all nests (including those where building was not observed), we measured the proximity (meters) to the closest Eriocephalus bush to estimate the minimum distance required to gather Eriocephalus material. We assessed the time spent gathering Eriocephalus material in two ways. We watched focal birds and measured the time spent perched in Eriocephalus bushes during a single bout of material acquisition. We used a stopwatch to the nearest second and defined the time spent gathering material as the interval between the first and last picks at Eriocephalus fluff. Our second method estimated the time (in days) allocated to nest construction and lining the nest with Eriocephalus material, as the number of days between the start of nest construction (the appearance of the first green strands of grass) until the first egg date. We measured the time allocated to lining the nest with Eriocephalus fluff as the number of days between the completion of the grass frame and the appearance of the first egg. Although some Karoo prinias begin lining the nest prior to completion of the grass frame (VGR pers. observ.), this time interval encompasses the majority of the nest-­‐lining process. 3.3.7 How many Eriocephalus capitula are in Karoo prinia nests? 30 To assess the effectiveness of Karoo prinias as seed dispersers, we counted the number of Eriocephalus capitula and weighed the total amount of Eriocephalus fluff in each nest. All measurements were made from nests collected once they were inactive (i.e., monitored nests that were recently depredated, abandoned or fledged young). The condition of inactive nests was variable—some were completely destroyed with material scattered throughout the nest bush, while others were perfectly intact. Nests that were destroyed or missing their lining were excluded from this analysis. Many other bird species breeding at Koeberg use Eriocephalus capitula and fluff in nest construction, and could disperse Eriocephalus seeds as well. Thus, we compared the number of Eriocephalus capitula found in Karoo prinias nests with 20 nests from seven other species at our site: southern double-­‐collared sunbird (n=3), grey-­‐backed cisticola (n=9), chestnut-­‐vented tit-­‐babbler (n=1), Cape white-­‐eye (n=1), Cape bulbul (n=3), yellow canary (n=2), and bokmakierie (n=1). We tested for differences in the number of Eriocephalus capitula between nests of Karoo prinias and all seven other species combined using a Wilcoxon rank sum test. 3.3.7.1 Distribution and amount of Eriocephalus fluff in prinia nests We quantified the distribution of Eriocephalus material in Karoo prinia nests by cutting nests in half and measuring nest-­‐wall thickness at 10 evenly spaced locations around the nest (see Fig. 3-­‐7). We measured wall thickness using a single half of the nest, and summarized these data using boxplots. We quantified the amount of Eriocephalus fluff in four ways: (i) measuring the depth of fluff in the base of nests that were cut in half; (ii) photographing the 31 interiors of all cut-­‐in-­‐half nests and quantifying the proportion of interior nest area covered in pale-­‐colored materials (mostly, but not exclusively, Eriocephalus fluff) using ImageJ (Rasband 2014); (iii) examining the amount of Eriocephalus material on the exterior of the nests, and (iv) weighing the total amount of Eriocephalus material in each nest (see Appendix A for more details). 3.3.7.2 Factors influencing the number of capitula and amount of fluff in Karoo prinia nests We measured six factors that may influence the number of Eriocephalus capitula and the amount of fluff in prinia nests: distance to the closest Eriocephalus plant, first egg date, number of days the nest remained active (i.e., interval between first egg date and the date the nest was depredated, fledged, or abandoned), minimum ambient temperature on first egg date (temperature data from a weather station at Koeberg Nature Reserve), nest height, and bush species in which nests were placed. We examined the number of Eriocephalus capitula and the amount of fluff in Karoo prinia nests separately, using five different statistical models. We corrected for multiple comparisons for the four analyses of Eriocephalus fluff following Pike (2011). For three analyses (number of capitula in nests, depth of fluff, and mass of fluff), the distributions of our data showed two peaks, one at zero and one at an integer value greater than zero. For these analyses, we used hurdle models with a zero-­‐adjusted Poisson distribution in R with package pscl (Jackman 2014), following Zuur et al. (2009). For analyses of Eriocephalus fluff on the interior and exterior of nests, data were more evenly distributed and thus we used generalized linear models. For all analyses, we checked the assumptions and fit of models following 32 Zuur et al. (2009). We selected the best fit models based on Akaike’s Information Criterion corrected for small samples sizes (AICc), using the dredge function in package MuMIn (Bartoń 2014). See Appendix A for detailed statistical procedures and data transformations. 3.4 Results 3.4.1 Do the reproductive phenologies of Karoo prinas and Eriocephalus overlap? The phenology of Eriocephalus racemosus seed fluff production overlapped with the breeding season of Karoo prinias at the Koeberg Nature Reserve (Fig. 3-­‐3). Fluff of E. racemosus plants was sufficiently developed for prinias to start using it by 14 September 2013, which coincided with first egg dates from early nests. The peak in first egg dates for Karoo prinia was approximately 20 October 2013, corresponding to the point at which nearly all Eriocephalus plants we surveyed had reach the “thick fluff” stage (Fig. 3-­‐3). Capitula loss associated with our plucking trials was strongly influenced by date (comparing glmm models with and without date: χ2 = 53.0, p < 0.0001); capitula were difficult to remove early in the season, but became easier to remove as the season progressed (glmm; intercepts for each picking date differed from each other; t = 8.0, p < 0.001, Fig. 3-­‐4). 3.4.2 Karoo prinia distance traveled and time spent gathering Eriocephalus material 33 Some focal prinias that we watched gather Eriocephalus fluff and return to their nests travelled as far as 209 meters (straight-­‐line distance between bush and nests). Data from dissected nests further suggest that some prinias traveled at least 344 meters to gather Eriocephalus fluff, based on the minimum distance from their nests to the nearest Eriocephalus plants (Fig. 3-­‐5; although prinias could also have salvaged material from previously constructed nests). Prinias that gathered Eriocephalus material from plants >100 meters from their nests typically flew to the closest available plant, whereas birds that gathered material from plants <80 meters from their nests, did not go to the closest available plant (Fig. 3-­‐5). These distances (80–100 meters) correspond roughly to the average diameter of Karoo prinia territories (~85.3 ± 18.6 meters, n = 19; Rowan and Broekhuysen 1962, VGR unpublished data), suggesting that prinias breeding on territories lacking Eriocephalus plants gather material from the closest available plant on neighboring territories. On four occasions we witnessed aggressive interactions between prinias when gathering Eriocephalus material; all four occasions involved individuals trespassing onto another's territory to gather fluff. On all occasions, the intruder was chased out of the defender's territory. These observations occurred during the early breeding season (before 15 September 2013) when Eriocephalus fluff was scarce (Fig. 3-­‐3). Karoo prinias spent a median of 6 days constructing their nests and 5 days lining their nests with Eriocephalus fluff (Fig. 3-­‐5). Eriocephalus plants have flexible branches that bend easily, and prinias frequently made hovering flights to prevent 34 falling while gathering fluff in strong winds. Prinias remained perched in the tops of moving branches for as long as two and a half minutes during a single bout of material acquisition, during which time they made as many as 102 picks at Eriocephalus capitula to gather a single bill full of fluff. When we saw prinias arrive at Eriocephalus bushes, they gathered fluff for an average of one minute, and made an average of 57 picks (Fig. 3-­‐5), before returning to their nest with a load of fluff. We observed prinias gathering fluff from only one bush per trip before returning directly to their nests. 3.4.3 How many Eriocephalus capitula are in Karoo prinia nests? Prinia nests (n = 104) contained an average of 6.6 ± 8.0sd Eriocephalus capitula (range: 0–43). In contrast, the 20 nests of the other seven species contained an average of 81.0 ± 83.7sd capitula (range: 0–262; Wilcoxon rank sum test, W = 1768, p < 0.00001, Fig. 3-­‐6). Karoo prinias nests constructed with Eriocephalus fluff contained on average 1.39g ± 0.71sd of Eriocephalus fluff. The average mass of fluff on the plant associated with a single capitulum was 0.0024g (± 0.0019sd, n = 20); thus Karoo prinia nests contained, on average, an amount of fluff equal to that associated with 579 Eriocephalus capitula. Two behaviors suggest that Karoo prinias actively avoided gathering Eriocephalus capitula. First, we watched many prinias (n > 10) gathering fluff from capitula that were held between their toes, which may help hold capitula in place so they are not removed with the fluff. Second, we witnessed four different prinias drop Eriocephalus capitula picked while gathering fluff, instead of transporting 35 those capitula to their nests. Dropping capitula caused these individuals to lose their bill-­‐full of fluff, and forced them to gather a new load of fluff before returning to their nest. Prinias placed the majority of Eriocephalus fluff in the nest interior, at the base of the nest cup (Fig. 3-­‐7). Nest wall thickness varied within nests and most of this variation was caused by the differential placement of Eriocephalus fluff; nest-­‐
wall measures taken from the top section of nests (i.e., measures 8–10 in Fig. 3-­‐7) were significantly thinner than those taken from the bottom section of nests (i.e., measures 3–5 in Fig. 3-­‐7) (Wilcoxon rank sum test, W = 1019.5, p < 0.00001, n = 124). We summarize results for the factors influencing the number of Eriocephalus capitula and the amount of Eriocephalus fluff in Karoo prinia nests in Table 3-­‐1 and Figure 3-­‐8 (for capitula only), and provide detailed statistical summaries of each analysis in Appendix A (Tables S2-­‐S5). Overall, our analyses revealed three significant predictors of the amount of Eriocephalus capitula and fluff in nests: proximity of Eriocephalus plants, first egg date, and the number of days a nest remained active. Nests that were closer to Eriocephalus plants contained more fluff and capitula, later nests contained less fluff but more capitula, and nests that remained active for longer contained more Eriocephalus fluff in their interior, consistent with observations of Karoo prinias adding material to the nest throughout the nesting cycle (Tarboton 2011). Predictor variables of nest height, bush species, and ambient temperature (after controlling for first egg date) were not significant in any analyses (Table 3-­‐1). 36 3.5 Discussion We provide some of the first detailed natural history data regarding a complex interaction between Karoo prinias and Eriocephalus plants in southern Africa. The reproductive phenologies of Karoo prinia and E. racemosus plants corresponded well, with seed fluff availability coinciding with nest building in prinias. Indeed, all but one of the earliest first egg dates occurred after Eriocephalus fluff was sufficiently developed for use in nest construction (Fig. 3-­‐3). Karoo prinias invested considerable time and effort gathering Eriocephalus fluff for their nests. Some prinias travelled at least 209 meters, and perhaps as far as 344 meters, to gather fluff, and sometimes left their territories and fought with other prinias in order to access Eriocephalus plants. Prinias spent, on average, 1 minute perched in Eriocephalus branches picking fluff during each material acquisition trip. Karoo prinias actively avoided gathering Eriocephalus capitula with the fluff by holding capitula in place with their toes and pulling only fluff, and by dropping capitula when they were removed from plants instead of bringing them to their nest. The fluff in the average prinia nest at our study site would have contained 579 capitula on the plant, yet nests contained an average of only 6.6 capitula. The few capitula in prinia nests contrasted with other species of birds that incorporated many more Eriocephalus capitula into their nests (Fig. 3-­‐6). These results suggest that Karoo prinias are poor dispersers of Eriocephalus seeds in general, and relative to some other species of birds on our site. Why Karoo prinias avoided Eriocephalus capitula, while other species did not, remains unknown. Incorporating Eriocephalus capitula into nests might attract 37 seed predators or parasites (e.g., ants, fungi, bacteria) that reduce prinia nesting success. Alternatively, capitula may be bulky and difficult for prinias to manipulate during nest building, or create an uneven nest lining for developing young. Gathering and transporting Eriocephalus capitula with the fluff could also increase the energetic costs of nest building; however, the capitula are light weight (average dry mass (g): 0.0061 ± 0.0019 sd, n = 20), and prinias appeared to expend significant time and energy avoiding them during fluff gathering, increasing the costs of nest building. 3.5.1 The plant perspective Plants whose seeds are dispersed by animals face an evolutionary dilemma: they must attract good seed dispersers while avoiding poor seed dispersers and seed predators. Karoo prinias used large amounts of seed fluff material — material believed to be produced by the plant to enhance seed dispersal (Augsperger 1986) — but actively avoided dispersing seeds, suggesting that they are poor seed dispersers. Prinias, however, did disperse some seeds, especially later in the breeding season (Fig. 3-­‐8) when capitula are more easily removed from plants and perhaps more mature (Fig. 3-­‐4). How well these seeds germinate and grow relative to seeds not incorporated into prinia nests or relative to seeds incorporated into the nests of other birds remains unknown. Dean et al. (1990) found low germination rates (2.2% of n = 724 seeds) of Eriocephalus capitula removed from the nests of several species of birds and planted in a nursery setting; however, they did not compare germination rates with capitula that were not incorporated into bird nests. 38 Seeds incorporated into bird nests may gain several advantages compared to seeds that fall directly to the ground. Dispersal away from parental plants may reduce density dependent seed mortality and improve germination success if seeds are transported to locations that promote growth (Howe and Miriti 2004). Bird nests also have the potential to provide seeds with added nutrients from organic material, especially when nests successfully fledge young, as these nests are often ringed with guano and contain added nitrogen from the feather-­‐sheathing of nestlings (Murphy and King 1986). Additionally, seeds incorporated into bird nests may escape from, or have reduced exposure to, terrestrial seed predators (e.g., harvester ants) (Dean et al. 1990, Kerley 1991), or reduced mortality caused by fungi or bacteria (Janzen 1977, Tewksbury et al. 2008). All of these factors have the potential to increase seed germination and early seedling growth, especially in areas of poor soil quality and high risk of seed predation or infection. Although the Karoo prinia breeding season corresponded well with the production of Eriocephalus fluff (Fig. 3-­‐3), we do not know if the production of fluff is representative of seed maturity. Unfortunately, we found no information about how morphology of the seed head corresponds with seed maturation or if seed maturation occurs on or off Eriocephalus plants. Over half the Eriocephalus plants we surveyed had reached the most advanced phenological stage (“thick-­‐fluff”) by 5 October (Fig. 3-­‐3), but capitula on these plants were possibly still developing as their stems were green and well fastened to plants (Jalink et al. 1999). Presumably, seeds removed from plants prior to maturation have reduced germination success and/or survival relative to mature seeds (Hong and Ellis 1990). 39 Other birds that harvest Eriocephalus seed material for their nests treated capitula differently. On three occasions we witnessed yellow canaries gathering Eriocephalus seed material, then crushing capitula in their bills. In the two yellow canary nests we examined, 53 of 118 Eriocephalus capitula were broken in half or in thirds, consistent with the seed crushing that we observed in canaries gathering Eriocephalus material. By contrast, nests of Cape white-­‐eyes, Cape bulbuls, bokmakierie, and chestnut-­‐vented tit-­‐babblers contained many Eriocephalus capitula (mean: 73, range 0–199), none of which appeared to be damaged. These observations suggest that interactions with some bird species (e.g., yellow canary, Karoo prinia) may be more costly to Eriocephalus plants than interactions with other birds (Stanton and Palmer 2011). 3.5.2 The bird perspective Life history theory predicts that investment in each reproductive attempt should decrease as predation risk increases because many attempts are often required to achieve reproductive success (Ghalambor and Martin 2001). Karoo prinias face extremely high rates of nest predation (Nalwanga et al. 2004, Martin et al. 2006), yet they spent a median of 6 days constructing nests and traveled at least as far as 209 meters, up to 3 territory diameters, to gather Eriocephalus fluff (Fig. 3-­‐5). The large amount of time and effort prinias spent gathering Eriocephalus fluff suggests that it provides fitness benefits to prinias that build their nests with it. Below, we outline six alternative hypotheses for potential fitness benefits Karoo prinias, and other bird species, might gain by constructing their nests with Eriocephalus material. 40 Climate. Eriocephalus seed material is soft and downy and may help maintain optimal temperatures and humidities inside the nest (Webb 1987, Hilton et al. 2004). During the early breeding season, and especially at night, ambient temperatures can drop to 8–10° C (Fig. 3-­‐2), challenging birds to maintain warm nest temperatures optimal for embryo and nestling development (~36–40˚C; Webb 1987). Later in the breeding season, and especially in interior valleys away from the coast, temperatures can exceed 35° C, challenging birds to prevent overheating of eggs and nestlings (Manning 2007). Breeding sites along the Atlantic coast of South Africa also receive heavy rains during the start of the nesting cycle, when the winter rainy season has not yet ended. Thus, Eriocephalus material could help keep nests warm during cool periods and cool during hot periods, maintain stable nest humidities, or help protect eggs, young, and attending females from precipitation by repelling water. Predation. Eriocephalus material added to bird nests could help reduce the risk of nest predation. Karoo prinias suffer exceptionally high rates of nest predation, with daily nest failure rates of at least 7.5% (Nalwanga et al. 2004, VGR unpublished data). Nest predators at Koeberg are diverse (e.g., snakes, mammals, rodents, and birds) and are thought to use visual and olfactory cues to locate nests. The amount of Eriocephalus material, its color, or its chemical compounds could help reduce nest predation by functioning as a physical barrier that impedes entry of predators to the nest, by concealing nests through crypsis or disruptive patterning (Stevens and 41 Merilaita 2009), by masking olfactory cues associated with nests (Clucas et al. 2008), or by providing an odor that is repulsive to predators (Swennen 1968). Parasites. Chemical compounds in Eriocephalus material could reduce ectoparasite numbers, development, and/or diversity. Eriocephalus compounds have been shown to reduce the growth of microbes and fungi cultured in laboratory settings (Njenga et al. 2005) and to reduce the numbers of nest ectoparasites in species that do not naturally encounter Eriocephalus compounds (Rohwer dissertation, Chapter 4). Karoo prinias continue to add Eriocephalus fluff to the interior of their nests during the breeding period, as we would expect if they were trying to maintain levels of aromatic compounds that dissipate over time (Petit et al. 2002). Alternatively, the pale color of Eriocephalus material in the nest lining could help parents recognize deleterious foreign objects in the nest, such as ectoparasites or eggs from brood parasites. However, the large amounts of Eriocephalus material placed in the base of Karoo prinia nests (Fig. 3-­‐7) suggests that fluff does not function to aid in the recognition and exclusion of foreign objects because smaller amounts of Eriocephalus lining could achieve the same function. Social signal. The amount of Eriocephalus material brought to the nest by the male or female could signal their quality or condition to their mate (Tomas et al. 2013). Female Karoo prinias are the primary builders of the nest’s grass frame, but both males and females help line the nests. Greater amounts of Eriocephalus material brought to the nest could signal a high quality individual, and promote increased investment in breeding by their mates, such as increased clutch size (Soler et al. 42 2001), nestling provisioning rates (English and Montgomerie 2011), or more vigorous nest defense against predators (Tomas et al. 2013). Time constraints. An abundance of Eriocephalus material in prinia nests could benefit prinias by facilitating rapid renesting after nest failure, if prinias reuse Eriocephalus material in successive nesting attempts. Birds often salvage nesting material from unsuccessful nests and, in some cases, use empty nests of other species for additional breeding attempts later in the breeding season (McKeller 2010). These strategies likely reduce the time and costs of nest construction, especially when nest predation is high, perhaps allowing additional breeding attempts that would not otherwise be possible. Adult condition. Chemical compounds of Eriocephalus material could benefit attending females by reducing feather degrading bacteria, ectoparasite loads, or by boosting their immune system. While we know of no studies that link chemical compounds in nests to adult female condition, this hypothesis seems plausible given that female Karoo prinias (and females of many species) spend as many as 21 days incubating and brooding young in the nest (Rowan and Broekhuysen 1962). Thus, nests constructed with Eriocephalus material may create an environment that optimizes or improves female condition during and/or after the breeding season. All of these hypotheses could improve the survival of nests, the development of eggs and young, and/or reduce physiological, immunological, and thermoregulatory demands on attending parents. These hypotheses are not 43 mutually exclusive; Eriocephalus might provide birds with multiple benefits, perhaps explaining its use by so many species from diverse avian lineages. 3.6 Conclusions Karoo prinias actively avoided capitula when gathering Eriocephalus fluff, and dispersed very few capitula relative to the amount of fluff used in their nests. These observations, coupled with comparisons with other birds that use Eriocephalus fluff in nest construction, suggest that Karoo prinias are poor dispersers of Eriocephalus seeds. On the other hand, the distance traveled and time spent gathering Eriocephalus fluff suggest that prinias may benefit from incorporating Eriocephalus material into their nests. These fitness benefits could be related to nest microclimate, reducing predation or parasitism, or providing a signal of individual condition. 44 Figure 3-­‐1. Eriocephalus material in Karoo prinia nests. Karoo prinia nests constructed without (A), and with an abundance of (B), Eriocephalus material. Eriocephalus material may help conceal nests—photo (C) is taken from the ground looking up at the base of a Karoo prinia nest (nest is in center of photo). (D) Eriocephalus bush. (E) Eriocephalus capitulum with fluff (left), half of fluff missing (plucked by authors) (middle), and fluff removed from capitulum (right). (F) Five Karoo prinia nests cut in half (nest entrance is in the upper right in each photo), with a range of Eriocephalus material on the nest interior. The left four nests all have Eriocephalus material; the right most nest has none and is lined primarily with Trichocephalus stipularis fluff and Helichrysum spp. leaves (all photographs: V. G. Rohwer). 45 Figure 3-­‐2. Environmental conditions at the Koeberg Nature Reserve. Grey band spans the range of average maximum and minimum temperatures (from years 2002–2012), and black line shows average ± 1SE rainfall (from years 1980–2013) throughout the year. Black box and dotted line in lower right of plot shows breeding activity of Karoo prinias, with most breeding activity indicated by the black box and less frequent breeding activity indicated by the dotted line (Tarboton, 2011). 46 Eriocephalus phenology
A
Number of nests
B
313
377 406
393 420
405
387
389
24
Aug
4
Sep
14
Sep
5
25
Sep Oct
15
Oct
26
Oct
6
Nov
24
Aug
4
Sep
14
Sep
25
Sep
5
Oct
15
Oct
26
Oct
6
16
Nov Nov
8
1
30
20
10
0
26
Nov
Figure 3-­‐3. Development of Eriocephalus seed material and the prinia breeding season. (A) Progression of Eriocephalus seed material for three patches of plants at the Koeberg Nature Reserve; y-­‐axis ranks plant development from least (1) to most (8) developed (see Table S1 in Appendix A for descriptions of phenological categories). Grey dotted line represents the earliest phenological stage of plants from which we observed prinias gathering fluff. Numbers above boxplots represent the number of plants surveyed, and boxplots show medians (thick lines), 25th and 75th percentiles (boxes), 1.5 times the interquartile range (whiskers), and outliers (points outside 1.5 times the interquartile range). (B) Histogram of Karoo prinia first egg dates using 139 nests found during the 2013 breeding season at the Koeberg Nature Reserve. 47 MAKE FONT SIZES 20!
Number of Eriocephalus
capitula removed
100
75
50
25
0
26
Aug
1
Oct
3
Nov
Figure 3-­‐4. Average number of capitula removed from Eriocephalus plants during the early (26 Aug), middle (1 Oct), and late (3 Nov) stages of the Karoo prinia breeding season. Boxplots show medians (thick lines), 25th and 75th percentiles (boxes), and 1.5 times the interquartile range (whiskers), which include all outliers. For each plant (n = 10), we plucked at the fluff of 100 capitula, and returned to the same plants for each plucking event. The number of capitula removed with plucking increased with date, and all plucking events differed from each other in the number of capitula removed (glmm; p<0.001 for all intercepts). 48 Figure 3-­‐5. Distance traveled and time invested in gathering Eriocephalus material. (A) Summary of 83 observations from 33 different Karoo prinia nests showing the distance prinias traveled to gather Eriocephalus fluff regressed on the distance to the closest Eriocephalus plant that was in the same phenological stage as the plant from which the focal bird gathered material; multiple observations for a single nest are shown with standard deviations. (B) Boxplot of the distance to the closest Eriocephalus bush for 100 prinia nests that contained Eriocephalus material. (C) Picture of Karoo prinia gathering Eriocephalus fluff from capitula held between toes. (D) Number of days that prinias allocated to nest building and lining the nests with Eriocephalus material. (E) The number of picks that prinias made during a single bout of material gathering. (F) The time spent perched in Eriocephalus bushes while gathering fluff. Boxplots show medians (thick lines), 25th and 75th percentiles (boxes), 1.5 times the interquartile range (whiskers), and outliers (points outside 1.5 times the interquartile range). 49 Cinnyris chalybeus
Prinia maculosa
n=3
n = 104
Cisticola subruficapilla
n=9
Sylvia subcaeruleum
n=1
Zosterops capensis
n=1
Pycnonotus capensis
n=3
Crithagra flaviventris
n=2
Telophorus zeylonus
n=1
0
100
200
Number of
of Eriocephalus
Eriocephalus
Number
capitula
nests
seeds ininnests
Figure 3-­‐6. Comparison of Eriocephalus capitula in bird nests from Koeberg Nature Reserve. Boxplots show medians (thick lines), 25th and 75th percentiles (boxes), 1.5 times the interquartile range (whiskers), and outliers (points outside 1.5 times the interquartile range); numbers in plot show sample sizes of nests for each species. Nests of other bird species contained, on average, more Eriocephalus capitula compared to nests of Karoo prinias (Prinia maculosa) (Wilcoxon sign rank test, p<0.0001). 50 Figure 3-­‐7. Measures of wall thickness for Karoo prinia nests. Nest-­‐wall thickness for 10 measures evenly distributed around the perimeter of Karoo prinia nests that were cut in half (n = 124). Measures include the grass frame and nest lining, which was primarily Eriocephalus fluff. Boxplots show medians (thick lines), 25th and 75th percentiles (boxes), 1.5 times the interquartile range (whiskers), and outliers (points outside 1.5 times the interquartile range). Photo illustrates the location of the 10 measures and indicates the nest entrance in the upper left hand corner; white arrow points to a line spanning the opening of the nest entrance. 51 Number of
Eriocephalus capitula
A
40
B
30
20
10
0
0
100
200
300
Closest Eriocephalus
bush (m)
17
Sep
7
27
16
Oct
Oct Nov
First egg date
Figure 3-­‐8. Factors influencing the number of Eriocephalus capitula in Karoo prinia nests. (A) Prinia nests that are closer to Eriocephalus plants and (B) nests constructed later in the breeding season contained more Eriocephalus capitula. Number of nests examined for each plot (n = 99). 52 Table 3-­‐1. Significant predictor variables influencing the number of Eriocephalus capitula and amount of Eriocephalus fluff in Karoo prinia nests. Summaries of analyses for: Eriocephalus capitula in nests, mass of Eriocephalus fluff, average depth of Eriocephalus fluff in the base of nests, and the proportion of Eriocephalus fluff in the interior and exterior of nests. Analyses of capitula, mass, and depth used hurdle models that examine the variation between nests constructed with and without Eriocephalus material (ne), and the variation in nests with Eriocephalus material (we), separately. Analyses of the proportion of Eriocephalus fluff in the interior and exterior of Karoo prinia nests used GLMs. Table shows p-­‐values and if predictor variables were positively (+) or negatively (-­‐) correlated with Eriocephalus material in nests. No analyses recognized nest height, bush species, or ambient temperature (after controlling for first egg date) as important predictors of the amount of Eriocephalus capitula or fluff in nests. Numbers of nests included in each analysis are presented in each column (range 99–105). Proximity of nest to bush Capitula n = 99 0.0009(we-­‐) First egg date Number of days active 0.0010(ne+) Fluff mass n = 104 0.0023(we-­‐) 0.0160(ne-­‐) 0.0001(we-­‐) 53 Average depth n = 104 0.0001(we-­‐) 0.0020(ne-­‐) 0.0170(we-­‐) Interior n = 105 0.0001(-­‐) Exterior n = 105 0.0001(-­‐) 0.0001(+) Chapter 4: Chemical compounds in Eriocephalus plants increase tree swallow fitness by impeding a naïve nest ectoparasite community and reducing nest predation. 4.1 Abstract Evolutionary constraints can limit the evolution of novel, complex traits, but organisms can bypass these constraints by co-­‐opting adaptive traits from other species. Diverse species of birds in southern Africa co-­‐opt aromatic material from plants in the genus Eriocephalus to build their nests. Secondary chemicals in Eriocephalus plants may increase the reproductive success of birds by reducing both infestation of nest ectoparasites and/or nest predation. We experimentally tested this hypothesis by adding secondary chemicals of Eriocephalus plants to tree swallow (Tachycineta bicolor) nests, a North American bird species with no evolutionary history of using aromatic plant material in nest construction. We supplemented nests with either Eriocephalus secondary chemicals suspended in mineral oil (experimental) or mineral oil without secondary chemicals (control), and measured ectoparasite infestation and fledging success. Nests supplemented with Eriocephalus oils had, on average, ~50 per cent fewer ectoparasites and fledged, on average, two young more than nests that received only mineral oil. The greater fledging success of nests with Eriocephalus oils was caused by reduced nest predation. Our results support the hypothesis that birds co-­‐opting secondary chemicals of Eriocephalus material for use in nest construction can gain fitness benefits by reducing the number of ectoparasites and nest predation. 54 4.2 Introduction Evolving novel adaptive traits can be difficult, particularly when the evolution of these traits would require new developmental or metabolic pathways (Futuyma 2010, Kirkpatrick 2010). Organisms can bypass the challenges of trait evolution by evolving behaviors that enable them to co-­‐opt adaptive traits from other species. We might expect organisms to co-­‐opt traits when: (i) generating novel traits requires completely new and complex pathways that are unlikely to evolve, and (ii) when challenges facing organisms require rapid evolution, such as vertebrate hosts facing more rapidly evolving parasites. Both of these conditions confront several species of birds that breed in southern Africa. These species use material from plants in the genus Eriocephalus to construct their nests (Dean et al. 1990). Eriocephalus plants are native to southern Africa and are highly aromatic (Njenga 2005), and their secondary chemicals represent complex traits that birds cannot produce themselves. Thus, birds constructing their nests with Eriocephalus material gain access to a novel trait (chemical compounds) that may provide fitness benefits by reducing nest predation and/or the number of arthropod ectoparasites that live in the nest and feed on nestling birds. Nest ectoparasites and predators are important selective agents on bird nests—they can influence nest size, placement, composition, and the likelihood of re-­‐using nests (Hansell 2000). Several observations suggest that the secondary chemicals in Eriocephalus material may benefit birds: (i) laboratory experiments show that Eriocephalus chemicals reduce bacterial growth (Njenga et al. 2005), (ii) birds invest considerable time and effort gathering Eriocephalus material and often 55 travel outside of their breeding territories to collect it (Rohwer dissertation, Chapter 3), (iii) compounds in Eriocephalus plants are hypothesized to repel snakes (Ferko 2006) that are known to depredate nests, and (iv) some species of birds continue to bring fresh Eriocephalus material to their nests during the incubation and nestling periods (Hockey et al. 2005), possibly to maintain levels of volatile secondary compounds that dissipate over time (Petit et al. 2002). Overall, observations to date suggest that birds using Eriocephalus material for nest construction gain access to complex traits (chemical compounds) that may increase reproductive success. We tested the hypothesis that the secondary chemicals in Eriocephalus plants increase reproductive success in bird nests by presenting Eriocephalus chemicals to a naïve ectoparasite and predator community and a bird species (tree swallow Tachycineta bicolor) that is not known to use aromatic plants in nest construction (Winkler et al. 2011). Ectoparasites infesting tree swallow nests are related to parasites infesting nests of birds breeding in southern Africa (Order: Diptera, Subsection: Calyptratae (Systema Dipterorum, http://www.diptera.org/)), but have no evolutionary history of interacting with Eriocephalus plants. Similarly, the most important predator of tree swallow nests at our site are gray ratsnakes (Pantherophis spiloides) that have no history of interacting with Eriocephalus. Thus, our experiment tests the potential of Eriocephalus chemicals to influence ectoparasite and predator species that have lacked the opportunity to evolve resistance to Eriocephalus compounds. 4.3 Methods 56 4.3.1 Study site, species, and procedure We used a box-­‐nesting population of tree swallows breeding at the Queen's University Biological Station in southeastern Ontario, Canada, which suffers high rates of ectoparasitism by blowflies (most common at our study site: Protocalliphora sialia, Diptera: Calliphoridae) (Rogers et al. 1991). We tested the hypothesis that secondary chemicals of Eriocephalus plants benefit birds by adding ~0.062mL (four small drops) of either diluted Eriocephalus secondary chemicals suspended in mineral oil (experimental) or mineral oil without secondary chemicals (control), to cotton balls placed inside tree swallow nests. To manipulate only levels of secondary chemical compounds, we used Eriocephalus africanus essential oil from New Directions Aromatics Inc., Mississauga, Ontario. We diluted essential oils by half, thus the amount of E. africanus essential oil added to experimental nests was ~0.031mL; these levels of secondary chemical compounds added to nests approximated levels found in nature (see Appendix B). We started oil additions with the first egg and continued until the nest was depredated or nestlings reached 15±1 days of age, when any further disturbance to the nest risked force fledging young. Nestlings typically fledged from the nest when they were 21 days old. We applied oils every three days to prevent complete dissipation of chemical compounds and to simulate the addition of new aromatic material to the nests (Petit et al. 2002, Hockey et al. 2005). 4.3.2 Do Eriocephalus chemicals reduce ectoparasite infestation? We counted the number and diversity of ectoparasite species and measured total parasite mass in control and experimental nests. We collected all nests as soon as 57 they were inactive and froze them for one week before dissection to stop parasite development at the stage when larval parasites lost their avian hosts. We tested if Eriocephalus chemicals reduce ectoparasites in nests using linear mixed-­‐effects models in the nlme package (Pinheiro et al. 2014) in R (R Core Team 2014) with either parasite number or parasite mass as our response variable and a treatment*exposure and a treatment*first egg date interaction as our predictor variables; we included breeding location (nest grid) as a random effect to control for spatial variation in parasite abundance. Parasite number and mass were square-­‐
root transformed, and first egg date was transformed using a Johnson Su transformation in JMP (SAS 2007) to fit a normal distribution. In our analyses, we included only nests where young hatched because parasites cannot develop without nestlings (Bennett and Whitworth 1991). First larval instars in nests that survived to nestling day 15 were excluded because we had evidence that they originated from eggs laid after the treatment had ended, but before nestlings fledged (see Appendix B). We used Akaike Information Criterion, corrected for small sample size (AICc), to select the best performing model and checked model assumptions following Zuur et al. (2009). 4.3.3 Do Eriocephalus chemicals increase tree swallow fitness? We estimated fitness by measuring the total number of young fledged from control and experimental nests and recorded the cause of any nest failure (see Appendix B). We paired control and experimental nests by location and first egg date to control for variation in site quality, seasonal changes in parasite abundance and reproductive investment, and used a Wilcoxon sign-­‐rank test for all nest pairs. We 58 also measured the stress hormone corticosterone in adult females and nestlings, and nestling growth rates from control and experimental nests (see Appendix B). 4.4 Results Nests supplemented with Eriocephalus chemicals had, on average, lower parasite mass (treatment estimate: -­‐0.58, z = 2.04, p = 0.041), and fewer parasites (treatment estimate: -­‐2.98, z = 2.52, p = 0.012, Fig. 4-­‐1) compared to nests supplemented with mineral oil. Nests supplemented with Eriocephalus chemicals fledged, on average, two young more than control nests (Wilcoxon sign-­‐rank test, v = 2.5, p = 0.012, n = 14), and this difference was caused by higher predation in control nests (Fisher’s exact test p = 0.05; see Appendix B). Nests with and without Eriocephalus compounds had similar clutch sizes, and hatched similar numbers of eggs (Wilcoxon sign-­‐rank tests p > 0.59, Fig. 4-­‐2). We found no differences in corticosterone levels in adult females or nestlings, and no differences in nestling growth rate between experimental and control nests (see Appendix B). 4.5 Discussion Nests supplemented with secondary chemicals of Eriocephalus plants had reduced ectoparasite loads and fledged more young compared to nests that did not receive Eriocephalus compounds. By manipulating only levels of secondary chemicals found in Eriocephalus plants, we isolate one mechanism by which Eriocephalus material can increase the reproductive success in birds. Many studies have investigated the functions of aromatic plant material in bird nests (reviewed in Dubiec et al. (2013), 59 and Tomás et al. (2013)), however, this study is the first, to our knowledge, to independently manipulate levels of secondary chemicals. This allowed us to control for variation in the chemical content of plant materials, and exclude alternative mechanisms whereby aromatic plant material may reduce parasite loads (e.g., by altering humidity levels, egg-­‐laying substrates, or the structural complexity and composition of bird nests, independent of secondary chemicals). 4.5.1 Mechanism by which Eriocephalus chemicals reduce ectoparasites Several compounds in Eriocephalus essential oils (eucalyptol, p-­‐cymene, camphor, sabinene) act as ovicides, larvacides, and/or repellents in Diptera (Musca domestica, and mosquitoes, Family: Culicidae) when applied directly as a liquid or indirectly as a fumigant (Rice and Coats 1994, Amer and Mehlhorn 2006, Kumar et al. 2014). The mechanisms by which secondary chemicals in bird nests reduce ectoparasites likely vary with different compounds, and diverse compounds likely target multiple life-­‐
stages in ectoparasites. Four possible patterns of naïve P. sialia larva across different developmental stages help identify potential mechanisms for how Eriocephalus compounds reduce ectoparasite infestation. (1) If nests receiving Eriocephalus compounds have consistently fewer parasites across different development stages compared to nests receiving only mineral oil, then Eriocephalus compounds may function as a deterrent to gravid females or as an ovicide; without data for numbers of P. sialia eggs, deterrent and ovicide hypotheses are indistinguishable. (2) If nests receiving Eriocephalus compounds have progressively fewer numbers of larvae with each developmental stage, relative to nests receiving only mineral oil, then Eriocephalus 60 compounds may function as a general larvacide, targeting all developmental stages. (3) Alternatively, Eriocephalus compounds may function as a larvacide but target only certain developmental stages; if so numbers of larva should be similar between treatments, until the targeted stage when numbers of larva in nests receiving Eriocephalus compounds should decline. (4) Finally, if numbers of P. sialia accumulated at a certain developmental stage in nests receiving Eriocephalus compounds relative to nests receiving only mineral oil, then compounds may function to inhibit larval development beyond certain instars. Our data are consistent with hypotheses that Eriocephalus compounds function as a deterrent to females or an ovicide against naïve P. sialia blowflies. Nests receiving Eriocephalus compounds had consistently fewer numbers of blowflies across different developmental stages, compared to nests that received only mineral oil (Appendix B; Fig. S5), suggesting that Eriocephalus compounds deterred gravid females or targeted P. sialia eggs. While we found no difference in the number of nests parasitized by blowflies (Appendix B; Fig. S6; Chi squared test χ2 = 0.0, df = 1, p = 1), Eriocephalus chemicals may repel some but not all flies, consistent with individual flies varying in their sensitivity to chemical compounds. The overall similar pattern in parasite numbers between treatments and across developmental stages suggests little role of Eriocephalus compounds as a growth inhibitor, at least for the developmental stages we examined. 4.5.2 Links between parasites and fitness Nest ectoparasites may reduce fitness of their avian host in a diversity of ways. Nest ectoparasites may reduce nestling survival by consuming too much blood, 61 introducing diseases, or weakening the immune system (Whitworth and Bennett 1992). They may also harm adults causing reduced feeding rates, nest abandonment, or adult mortality; however nestling and adult mortality caused by nest ectoparasites seems unlikely to occur as nest ectoparasites require live birds for their growth and development. Finally, nest ectoparasites may interact with predation risk. Increased ectoparasite loads may increase offspring metabolic demand (due to blood loss), and cause parents to increase feeding rates to nestlings. Increased parental activity around the nest could then increase predation risk by cueing predators to active nests. Any of these mechanisms may result in the loss of entire broods or partial brood reduction. Our results suggest that Eriocephalus compounds increased fledging success by reducing entire brood loss rather than partial brood reduction (Appendix B; Fig. S7). Comparing the number of nests with and without Eriocephalus compounds that fledged young to those that failed to hatch, were depredated, or contained dead nestlings, suggests that nests receiving Eriocephalus compounds suffered lower nest predation compared to nests receiving only mineral oil (depredated nests: Fisher’s exact test, p = 0.050; nests with eggs that failed to hatch or with dead nestlings: Fisher’s exact test, p > 0.23). Eriocephalus compounds may lower the risk of predation by reducing parental activity around the nests, reducing the ability of predators to locate nests, or by physically repelling predators. 4.6 Conclusions Secondary chemicals likely serve multiple functions in plants (Berenbaum 1995), but perhaps the most widely recognized function is defending against herbivores. 62 Birds that co-­‐opt secondary chemicals of plants may gain access to an effective defense — against insects and possibly nest predators — that evolved in a completely different ecological context. Our results support this hypothesis, and suggest that secondary chemicals of Eriocephalus plants can reduce both ectoparasite loads and rates of nest predation for birds that co-­‐opt them. By co-­‐
opting these chemical defenses and using them in a novel way, birds bypass the constraints that impede the evolution of novel defenses against natural enemies. 63 1.5
0.5
70
parasite numbers
parasite mass (g)
2.5
50
30
10
Figure 4-­‐1. Average (±1SE) mass and number of P. sialia parasites in tree swallow nests supplemented with mineral oil (control, black, n=16) versus nests supplemented with mineral oil and secondary chemicals from Eriocephalus plants (experimental, gray, n=15). 64 clutch
size
nestlings
at hatch
young
fledged
count
6
4
2
0
1.5
0.5
parasite numbers
parasite mass (g)
Figure 4-­‐2. Average (±1SE) clutch size, number of eggs hatched, and number of young fledged in nests supplemented with Eriocephalus compounds suspended in mineral oil (experimental, gray, n=14) versus nests supplemented with mineral oil without compounds (control, black, n=14). 2.5
70
50
30
10
65 Chapter 5: No evidence that augmenting Eriocephalus compounds reduces predation on Karoo prinia nests from rhombic egg-­‐eating snakes. 5.1 Abstract Organisms can overcome challenges in their environment by co-­‐opting the traits of other species. Many species of birds that breed in southern Africa use aromatic material from plants in the genus Eriocephalus to construct their nests. Chemical compounds in Eriocephalus are complex traits that birds cannot produce themselves, but these traits may benefit birds by reducing nest ectoparasitism or predation. We tested the hypothesis that Eriocephalus compounds reduce nest predation by a specialist predator—the rhombic egg-­‐eating snake (Dasypeltis scabra), using both field and captive experiments. For our field experiments, we supplemented active Karoo prinia (Prinia maculosa) nests with either Eriocephalus compounds suspended in mineral oil to create a nest-­‐chemical-­‐phenotype that exceeded the levels found in natural nests (experimentals), or mineral oil without Eriocephalus compounds (controls), and measured nest survival throughout the egg-­‐laying and incubation periods. For our captive experiments, we presented six Dasypeltis snakes with the option of (i) consuming eggs in nests with or without Eriocephalus compounds, and (ii) resting under hides with or without Eriocephalus compounds. Both field and captive experiments suggest that Eriocephalus compounds do not reduce predation from egg-­‐eating snakes, nor do they deter snakes from resting under hides. Our results suggest that if Eriocephalus compounds reduce nest predation, they likely do not function as a repellent of predators but 66 instead may function as a chemical camouflage that is sensitive to chemical profiles of local Eriocephalus plants. 5.2 Introduction Evolutionary constraints limit the genetic diversity and developmental pathways that are available to organisms, and can restrict or prevent the evolution of novel, complex traits (Moczek 2008). As an alternative, organisms can gain access to novel traits simply by co-­‐opting them from other species. For example, when organisms co-­‐opt chemical compounds from plants or insects to combat bacteria, parasites, or predators, these individuals bypass evolving the developmental pathways required to produce the chemical compounds themselves, yet gain access to novel, adaptive traits (Eisner et al. 1974). Co-­‐opted traits are often used against parasites and predators (Eisner et al. 1974, Wimberger 1984, Stachowicz and Hay 1999, Hemmes et al. 2002, Chapuisat et al. 2007, Bravo et al. 2014). In these cases, one species co-­‐opts traits from a second species to mediate negative interactions with a third species. The use of co-­‐opted traits to mediate interactions with other species might be especially common among tightly-­‐linked, co-­‐evolving species like predators and their prey, or parasites and their hosts, because co-­‐evolutionary races can provide persistent directional selection that exhausts genetic diversity, limiting the potential for adaptive trait evolution (Dawkins and Krebs 1979). By co-­‐opting traits, species gain access to novel traits that may function as an effective defense because they lie outside the evolutionary history of their antagonistic partner(s) (Brockhurst et al. 2014). 67 We examined a potential case of trait co-­‐option in South Africa involving breeding birds, aromatic plants in the genus Eriocephalus, and a specialist predator of bird eggs—the rhombic egg-­‐eating snake (Dasypeltis scabra). We focus on a single species of bird, the Karoo prinia (Prinia maculosa), which uses an abundance of Eriocephalus seed material in nest construction (Rohwer dissertation, Chapter 3). While Eriocephalus material in Karoo prinia nests may have multiple functions (e.g., insulation, reduce nest ectoparasitism, signal individual quality; Rohwer dissertation, Chapter 3), we focus here on testing the hypothesis that chemical compounds in Eriocephalus plants reduce egg predation from rhombic egg-­‐eating snakes. Karoo prinias face some of the highest rates of nest predation known in birds (7.5% daily nest predation rate; Nalwanga et al. 2004, Martin et al. 2006), with the dominant nest predator thought to be the nocturnal, rhombic egg-­‐eating snake (Lloyd et al. 2001, Nawalga et al. 2004, Martin et al. 2011, Bates and Little 2013, VGR pers. obs.). Natural history observations suggest egg-­‐eating snakes rely primarily on olfactory cues to find bird nests. Visual cues like parental activity at the nest site are unavailable to these snakes because adults birds are inactive at night, when snakes are searching for nests. Additionally, egg-­‐eating snakes are not known to have sensitive thermal receptors like pits of pit-­‐vipers or labial pits of pythons (de Cock Buning 1983, Breidenbach 1990), suggesting that they do not use the body heat of birds to locate nests. These observations suggest egg-­‐eating snakes rely on olfactory cues to find nests, although they may also use visual cues of the nest itself. 68 Rhombic egg-­‐eating snakes eat only bird eggs, and thus feeding is restricted to the avian breeding season (Gartner and Greene 2008). Long periods of fasting between avian breeding seasons likely require periods of intense foraging. Mid-­‐
sized Dasypeltis snakes held in captivity eat as many as six Japanese quail (Coturnix coturnix) eggs during a single feeding (Jones 1996), suggesting that one snake could eat up to 60 Karoo prinia eggs, or ~15 prinia clutches, in a single feeding bout (1 quail egg ~10g, 1 prinia egg ~1g, average prinia clutch size is 4 eggs; Rowan and Broekhuysen 1962). Thus, selection on prinias to reduce predation by Dasypeltis snakes is likely strong. How might Karoo prinias reduce nest predation by egg-­‐eating snakes? Karoo prinias use an abundance of aromatic Eriocephalus material to construct their nests (Rohwer dissertation, Chapter 3) and several observations suggest that this material may reduce predation from rhombic egg-­‐eating snakes, either by concealing nests (e.g., Clucas et al. 2008, Bailey et al. 2015) or by repelling snakes (e.g., Swennen 1968). First, Karoo prinias traveled as far as 344 meters to gather Eriocephalus material and remained perched in the tops of Eriocephalus branches for up to two and a half minutes during a single bout of material acquisition (Rohwer dissertation, Chapter 3). The time and effort invested in gathering Eriocephalus material suggests that it is beneficial to breeding prinias. This investment is especially remarkable because the high nest predation rates experienced by prinias should favor reduced investment in a single nesting attempt, because multiple attempts are usually required before young are successfully fledged (Martin 1995). Second, Karoo prinias bring fresh Eriocephalus material to the nest site throughout the breeding cycle — a 69 pattern that we expect if birds are maintaining levels of chemical compounds that dissipate over time (Clark and Mason 1985, Petit et al. 2002). Third, at least two compounds in Eriocephalus plants, camphor and eucalyptol, are thought to repel snakes (Harper and Brothers Publishers 1874, Ferko 2006). Thus, Eriocephalus material in Karoo prinia nests might reduce predation from rhombic egg-­‐eating snakes by masking olfactory cues associated with nests and/or by actively deterring snakes. We tested the hypothesis that the chemical compounds in Eriocephalus plants reduce predation on Karoo prinia nests using both field and captive experiments. For our field experiment, we supplemented active Karoo prinia nests with either Eriocephalus compounds suspended in mineral oil to create a nest-­‐
chemical-­‐phenotype that exceeded levels found in the original nests (experimentals), or mineral oil without Eriocephalus compounds (controls). We measured nest survival and predicted reduced predation on experimental nests relative to controls. Next, we used captive Dasypeltis snakes in two choice experiments where snakes had the option of (i) eating eggs from nests with or without Eriocephalus compounds, and (ii) resting under hides (dark refuges where snakes spent daylight hours) with or without Eriocephalus compounds. We predicted that nests and hides with Eriocephalus compounds would have reduced predation and occupancy, respectively. 5.3 Methods 5.3.1 Study site 70 We conducted field experiments at the Koeberg Nature Reserve, Western Cape Province, South Africa, ~35km north of Cape Town along the Atlantic coast. Koeberg has cool wet winters and warm dry summers (Goldblatt and Manning 2002). Vegetation at Koeberg is a mix of (i) dune thicket scrub, a dense tangle of coastal succulents and woody shrubs, and (ii) sand plain vegetation, which is more open with isolated shrubs; both vegetation types are short (<2 meters) (Manning 2007). Two species of Eriocephalus occur at Koeberg: E. racemosus and E. africanus. E racemosus is far more common than E. africanus, and both species are unevenly distributed throughout the landscape. Karoo prinias are one of the most abundant breeding birds at Koeberg, use an abundance of Eriocephalus material in nest construction, and suffer high rates of nest predation (Nalwanga et al. 2004). A diversity of nest predators occur at Koeberg, including birds (pied crow Corvus albus, fiscal shrike Lanius collaris, bokmakierie Telophorus zeylonus), mammals (Cape grey mongoose Galerella pulverulenta, Southern African vlei rat Otomys irroratus, four-­‐striped grass mouse Rhabdomys pumilio), and snakes (rhombic egg-­‐
eating snake Dasypeltis scabra, boomslang Dispholidus typus, mole snake Pseudaspis cana, Cape cobra Naja nivea), but the dominant nest predator is thought to be the rhombic egg-­‐eating snake (Lloyd et al. 2001, Nawalga et al. 2004, Martin et al. 2011, Bates and Little 2013, VGR pers. obs.). See Rohwer dissertation (Chapter 3) for more details about the study site. 5.3.2 Finding, monitoring, and collecting nests We found Karoo prinia nests by watching adults carry material to their nests and by searching suitable habitat. Upon finding a nest, we marked its location using a GPS 71 receiver (GPSmap 60, Garmin International Inc., Olathe, Kansas). We recorded or estimated first egg date for all nests. For nests found in the laying stage, we assumed that females laid one egg per day and back counted until the first egg date (Rowan and Broekhuysen 1962). For nests found during incubation, we estimated first egg dates either by back counting from the hatch date (by 14 days for incubation plus the appropriate number of days for clutch size, assuming females laid 1 egg per day Rowan and Broekhuysen 1962), or by taking the midpoint between the least and most advanced possible stage for depredated nests (Martin et al. 2000). We checked nests every three days. Once nests became inactive (e.g., depredated, abandoned, or fledged young), we recorded nest height and the bush species in which nests were placed. 5.3.3 Assigning predators to depredated nests For depredated nests, we estimated the identity of predators using several clues. Rhombic egg-­‐eating snakes regurgitate egg shells in a distinctive form (laterally compressed with a cut along the length of the egg on one side), usually at the base of nests that they depredate (Gartner and Greene 2008). Any nest with these distinctive egg shells below it (within 2m of the nest), coincident with the depredation of the nest, was considered to be depredated by rhombic egg-­‐eating snakes. However, rhombic egg-­‐eating snakes do not always regurgitate egg shells below the nests that they depredate. Three of six video-­‐taped predation events by egg-­‐eating snakes, and six of 11 depredated nests with temperature data that strongly suggested predation by egg-­‐eating snakes (e.g., rapid cooling of the nest at night, providing a nest-­‐temperature profile similar to females that were flushed 72 from the nest during predation events by egg-­‐eating snakes; VGR, video data) lacked egg shells below the nest. Thus, relying solely on the presence of egg shells likely underestimates predation by these snakes (Bates and Little 2013). Egg-­‐eating snakes cause little to no damage to nests during predation events (based on 33 nests where rhombic egg-­‐eating snakes were the known predator because of regurgitated egg shells beneath nests). Thus, when we found no regurgitated egg shells below nests that were depredated during the laying or incubation stage, but the nest was left in excellent condition (without rips or tears in nest material), we assigned these nests as depredated by egg-­‐eating snakes. If depredated nests were found: (i) with large holes in the roof, sides, or back of the nest, often with material scattered throughout the nest bush and surrounding area, (ii) with egg-­‐yolk in the nest lining, or (iii) depredated during the nestling stage — all signs inconsistent with depredation by egg-­‐eating snakes — then we assigned "non egg-­‐eating snake predator" (e.g., pied crows, mongoose, rodents, or other snake predators) to the nest predation event. Our classifications of suspected predators based on nest condition and stage follow other studies (Martin and Roper 1988, Bates and Little 2013) but undoubtedly lead to some error because we did not witness predation events. Additionally, nests could have been depredated by an egg-­‐eating snake, and subsequently altered by another animal, leading to a misclassification. However, we suspect that the secondary alteration of nests is rare, particularly during the short window of time (three days) in between our nest checks. In addition, the distinctive differences between nests depredated by egg-­‐eating snakes versus other predators 73 at our site (e.g., distinctive egg shells, damage to nests) gave us confidence in our nest predator assignments. 5.3.4 Field experiment We tested if augmenting levels of Eriocephalus chemical compounds reduced nest predation by adding 0.05mL of either diluted Eriocephalus compounds suspended in mineral oil (experimental) or mineral oil without Eriocephalus compounds (control), to active Karoo prinia nests. Because most prinia nests in our study (101 of 104 dissected nests) included some Eriocephalus material, this experimental design compared the survival of nests with nest-­‐chemical-­‐phenotypes that typically exceeded naturally occurring levels of Eriocephalus compounds (experimentals) to nests with naturally occurring levels of Eriocephalus compounds (controls). To augment levels of Eriocephalus compounds, we used Eriocephalus africanus essential oil from New Directions Aromatics Inc., Mississauga, Ontario. We diluted essential oils by half. Thus, the amount of E. africanus essential oil added to experimental nests was 0.025mL; these levels of chemical compounds added to nests approximated levels found in nature (Rohwer dissertation, Appendix B). We added treatments to nests by inserting a blunt syringe ~1–2cm into the base of the nest material, which was primarily Eriocephalus seed fluff. Adding treatments to this section of nests concentrated compounds where prinias concentrate Eriocephalus material and also ensured that oil treatments did not contact eggs. We added treatments every three days because Eriocephalus chemicals are highly volatile and dissipate quickly (Rohwer dissertation, Appendix B). We began oil addition with the appearance of the first egg, or immediately upon finding nests that already 74 contained eggs, and continued until the nest was depredated or until nestlings reached 10±1 days of age; continuing treatment beyond this date risked premature fledging of young. 5.3.5 Captive experiments Because our field experiments could not compare nest survival between nests with and without Eriocephalus compounds, we performed two additional experiments involving the presence versus absence of Eriocephalus compounds. For these experiments we presented 6 captive Dasypeltis snakes (which included three species: D. scabra (n=2), D. inornata (n=1), and D. gansi (n=3)) with the choice of consuming eggs from nests with or without Eriocephalus compounds, or resting under hides with or without Eriocephalus compounds. For each experiment, we allowed snakes to acclimate to terrariums for 24 hours, waited at least five days between trials that used the same snake, and cleaned cages and cage substrates between trials using different snakes. The first experiment tested if Eriocephalus compounds reduced nest predation. We presented snakes with the option of eating one Japanese quail egg from nests supplemented with either 0.05mL of Eriocephalus compounds suspended in mineral oil (experimental) or 0.05mL of mineral oil without Eriocephalus compounds (control). We used cup-­‐shaped nests that were constructed without any Eriocephalus material and added treatments around the exterior of nests, which prevented contact between treatments and eggs. We placed nests 1.2 meters apart from each other in a 1.5 meter long terrarium with the snake’s hide directly between control and experimental nests (terrarium dimensions: length x width x 75 height in meters = 1.5 x 0.5 x 0.65). Both nests were placed 0.45m above the ground and held in place by a stiff piece of wire that extended up from the base of the terrarium. At the base of each nest and surrounding the wire, we placed a tangle of dry, leafless branches to obscure the shape of nests and to allow snakes easy access to nests. We video recorded each trial using a Sony Handycam DCR-­‐SR85 video camera (Sony Corporation, Tokyo, Japan), equipped with infrared light. We then scored (i) which nest each snake visited first, and (ii) from which nest each snake ate eggs. Low video resolution during night-­‐time filming prevented us from counting tongue flicks made by snakes. We considered a snake to have visited a nest if its head came within 3 cm of the nest. We tested each snake twice, switching the position of control and experimental nests between trials to control for nest-­‐
position effects within the terrarium. The second experiment examined if Eriocephalus compounds repel snakes — a hypothesis that may not have been tested in experiment 1 if hunger overrode their discomfort. For this experiment, we provided snakes with the option of resting under hides with and without Eriocephalus compounds. Our addition of Eriocephalus compounds for control versus experimental treatments followed experiment 1, except that we used two identical hides instead of nests, and the terrarium was slightly smaller (length x width x height in meters = 1.2 x 0.5 x 0.65). We added treatments to a small strip of paper towel (length x width in cm = 1.5 x 3.5) fastened to the roof inside of each hide. We placed hides with their entrances facing each other, 0.8 meters apart, with a single light source placed above the terrarium, directly between the hides. This lighting arrangement controlled for the 76 amount of light entering each hide. Dasypeltis snakes are primarily nocturnal, so snakes had the option of choosing which hide to rest under each night. The following day, we recorded the location of the snake. Once we had recorded the location of the snake, we temporarily removed the snake in between trials. A hide choice experiment lasted three nights and consisted of three trials. Night one presented snakes with a control and experimental hide, as described above. Night two switched the position of control versus experimental hides, to control for potential positional effects within the terrarium. Night three did not switch the position of the hides from night two, but disturbed the snake, allowing us to test if snakes switched hides when they were disturbed. 5.3.6 Statistical analyses—field experiment We tested if prinia nests with augmented amounts of Eriocephalus compounds survived longer than nests without augmented compounds using Cox proportional hazard regressions. These models are frequently used in the biological sciences (McKinnon et al. 2010, Best et al. 2012, Rivers et al. 2012) because they can assess the relative importance of time-­‐dependent covariates on survival without making assumptions about the underling hazard function (Nur et al. 2004). In our case, Cox proportional hazard regressions modeled the time interval between the start of our oil application treatment until nests became inactive, either because they were depredated or fledged young. We ran four analyses with slightly different sets of predictor variables. For all analyses, however, our response variable was the same—a survival function that contained the number of exposure days (total number of days from the date of first 77 oil application until the date nest was inactive) and nest fate (depredated, successful). When exact dates of failure/fledging were unknown, we assigned the midpoint between the previous nest check (when the nest was active) and the final nest check (when the nest was inactive) as the date of failure or fledging, following previous studies (Martin and Roper 1988, Martin and Geupel 1993, Martin et al. 2000); nests were checked every 3 days. The first analysis tested for treatment effects on nest survival. For this analysis, our predictor variables were: treatment, first egg date, maximum temperature at night, residual mass of Eriocephalus material (controlling for the number of days a nest remained active), nest height, bush species, density of active Karoo prinia nests (within a 350m radius circle of each focal nest, and including all nests that were active within seven days prior to failure or fledging of the focal nest) and interaction terms between treatment and all predictor variables except bush species (a categorical variable with six levels). The second analysis tested for differences in survival probabilities as a result of suspected nest predators—rhombic egg-­‐eating snakes versus non egg-­‐eating snake predators. For this analysis our predictor variables were: predator type (i.e., egg-­‐eating snake vs. non egg-­‐eating snake), treatment, first egg date, maximum temperature at night, residual mass of Eriocephalus material, nest height, bush species, density of active Karoo prinia nests, and interaction terms between predator type and first egg date, predator type and maximum temperate at night, predator type and nest height, predator type and residual Eriocephalus mass, and predator type and density of active nests. 78 The third and fourth analyses examined survival of nests suspected to have been depredated by egg-­‐eating snakes or by predators other than egg-­‐eating snakes, respectively. We analyzed these two data sets separately because our second analysis revealed predator type interactions between first egg date and maximum temperature at night. For both analyses, our predictor variables were: treatment, first egg date, maximum temperature at night, residual mass of Eriocephalus material, nest height, density of prinia nests, bush species, and interaction terms between treatment and first egg date, treatment and temperature, treatment and residual Eriocephalus mass, treatment and nest height, and treatment and nest density. We transformed or redefined three predictor variables prior to analyses: nest bush species, Eriocephalus mass, and density of active Karoo prinia nests. Nests were found in 19 different bush species, and thus we reduced the number of factor levels for this variable by grouping bush species into taxonomic families, and then by grouping rare families (those with fewer than five nests) into a single category “other”. This reduced factor levels for this variable from 19 to six. For Eriocephalus mass, Karoo prinias continue to add Eriocephalus material to their nests throughout the nesting cycle, and thus mass of Eriocephalus material in prinia nests co-­‐varies positively with the number of days a nest remains active (p<0.0001, r2=0.21). To address this issue, we regressed Eriocephalus mass on the number of days a nest remained active and used the residuals of this regression as a predictor variable describing the relative amount of Eriocephalus in nests in all analyses. For density of active Karoo prinia nests, we used a single measure of nest density that included all 79 active nests within a 350m radius circle and that were active (e.g., contained eggs or nestlings) for at least one day within seven days from when the focal nest was depredated or fledged. We could find no data about home-­‐range size, movement patterns, or spatial memory of rhombic egg-­‐eating snakes (or any other Dasypeltis snake). We chose to examine a seven day time interval and 350m radius area for our calculation of nest densities because Dasypeltis diet and metabolic rate suggest that snakes will wait up to one week between meals (Greene et al. 2013), and because other predator species have well-­‐developed spatial and temporal memory (Sonerud and Fjeld 1987, Pravosudov and Roth 2013). Additionally, we made several observations of rhombic egg-­‐eating snakes returning to nests that they depredated the night before, suggesting that they remember the locations of individual nests within their home range. For all Cox regressions, we identified the best performing models using Akaike’s Information Criterion corrected for small sample size (AICc), with the dredge command in package MuMIn (Bartoń 2015) in R (R Core Team 2014). Our analyses examined only nests that contained eggs because egg-­‐eating snakes are not thought to eat nestlings (Greene et al. 2013). For two variables — residual mass of Eriocephalus material and nest height — we did not have complete data because many nests were destroyed when they were depredated. Missing values for these variables limited our sample size; thus, if these variables were not included in the top models (i.e., models with delta AICc<2), then we removed them and reran the analysis with a larger sample of nests. For all Cox regressions, we tested the assumption of proportional hazards graphically following Fox (2002), for full 80 models prior to model selection and for the best performing models after model selection. We plotted scaled Schoenfeld residuals of our best models against our predictor variables to ensure that these relationships showed no trends or patterns. We tested for influential data points by examining the contribution of each observation to the regression coefficient using the dfbeta function. Finally, we tested for non-­‐linearity of covariates by plotting Martingale residuals of our final model and component-­‐plus-­‐residuals against all continuous variables. In these plots, we looked for non-­‐linear trends of Martingale residuals with covariates and found none, indicating that covariates met the assumptions of linearity. 5.3.7 Statistical analyses—captive experiments For both captive snake experiments, we tested if Eriocephalus compounds reduced visitation to nests or predation of eggs, and if compounds repel snakes from resting under hides, using binomial tests in R (R Core Team 2014). 5.4 Results 5.4.1 Field experiment Of 131 Karoo prinia nests that we monitored, 122 were depredated before the eggs had hatched, 4 were depredated during the nestling period, 4 fledged young, and one nest was abandoned after the eggs failed to hatch (after 22 days of incubation). Nest predation was highest during the laying and early incubation stage (Fig. 5-­‐1). Nests that received augmented levels of Eriocephalus compounds suspended in mineral oil did not have significantly higher survival probabilities compared to nests that received only minerial oil (Table 5-­‐1, Fig. 5-­‐2). Although treatment was 81 included in 1 of 4 best performing models (ΔAICc<2), the best performing model did not include treatment and was 2.21 times more likely than the closest model that included treatment. All best performing models included maximum temperature at night and nest height as important predictor variables and both of these predictors were significant (p<0.05; Table 5-­‐1); risk of predation increased as temperature and nest height increased. Predator type had a significant effect on nest survival—nests suspected to have been depredated by rhombic egg-­‐eating snakes remained active, on average, for less time compared to nests suspected to have been depredated by predators other than egg-­‐eating snakes (Table 5-­‐2). First egg date and interactions between predator type and first egg date, and predator type and maximum temperature at night were also significant predictors of nest survival (Table 5-­‐2). For analyses of nests suspected to have been depredated by egg-­‐eating snakes, best performing models (those with ΔAICc<2) included 6 different predictor variables, but only maximum temperature at night was statistically significant (Table 5-­‐3); survival probability decreased as temperatures increased. For analyses of nests depredated by predators other than egg-­‐eating snakes, best performing models included 4 different predictor variables, but only first egg date was statistically significant (Table 5-­‐4); survival probability decreased for later-­‐nesting prinias. 5.4.2 Captive experiments 82 Eriocephalus compounds did not reduce visitation to nests (p = 1) or predation of eggs (p = 0.77), nor did they repel snakes from resting under hides (p = 0.87; Fig. 5-­‐
3). 5.5 Discussion Karoo prinia nests suffered high rates of predation, most of which occurred during the early laying and incubation period (Fig. 5-­‐1). We tested the hypothesis that the chemical compounds of Eriocephalus material in prinia nests reduce predation, especially from rhombic egg-­‐eating snakes. However, our field experiments detected no evidence that augmented levels of Eriocephalus compounds reduced nest predation (Fig. 5-­‐2). Rather, the most important predictor variables of prinia nest survival were first egg date, maximum ambient temperature, and nest height. Our trials with six captive egg-­‐eating snakes further suggested that Eriocephalus compounds did not reduce snake visitation to nests, their consumption of eggs, or function to repel snakes from resting under hides with Eriocephalus compounds (Fig. 5-­‐3). 5.5.1 Data interpretation from field experiments The experimental addition of Eriocephalus compounds resulted in no apparent reduction in nest predation, suggesting that Eriocephalus compounds do not reduce nest predation rates by egg-­‐eating snakes. Alternatively, Eriocephalus compounds may reduce nest predation rates, but our study design prevented us from detecting this effect. Four possible factors may have prevented us from measuring a reduction in nest predation rates. First, our sample size may have been too small given the 83 high rates of nest predation that caused high variance in our survival analyses. Second, field experiments occurred during one breeding season that had extremely high rates of nest predation (cf. Nalwanga et al. 2004). These high rates of nest predation could have masked the effect of Eriocephalus compounds on nest predation rates. Third, chemical profiles of E. africanus oils that we added to nests may not have matched the chemical profiles of E. racemosus plants at our study site, perhaps negating the effects of our treatment on nest predation rates. Finally, our experiment increased Eriocephalus compounds above natural levels in each nest, potentially elevating levels of secondary compounds above the ideal or threshold levels for reducing nest predation rates. Consistent with the hypothesis that Eriocephalus compounds reduce nest predation, treatment was retained in one of four top models (ΔAICc<2) (Table 5-­‐1A), and nests receiving Eriocephalus compounds had, on average, higher survival relative to nests receiving odorless mineral oils (Fig. 5-­‐2). 5.5.2 How Eriocephalus compounds could reduce nest predation Eriocephalus compounds could reduce nest predation either by functioning as a chemical camouflage that conceals nests, or by repelling egg-­‐eating snakes. If compounds function as an olfactory camouflage, then our experiments may not have detected a reduction in predation for two reasons. First, Karoo prinias may optimize the effect of Eriocephalus compounds with the amount of Eriocephalus material used in nest construction such that augmenting levels of compounds (as we did in our experiment) does not increase nesting success. Second, Eriocephalus plants show individual-­‐, population-­‐, and species-­‐level variation in their chemical profiles 84 (Njenga 2005), and our addition of E. africanus oils distilled from plants outside of Koeberg likely differed in their chemical profiles from local E. racemosus and E. africanus plants. Both of these factors suggest that our experimental design could have been inappropriate for testing the chemical camouflage hypothesis. If chemical compounds functioned as a predator repellent, however, then augmented levels of Eriocephalus compounds should have increased their effectiveness and reduced nest predation rates. Our results of similar rates of nest predation between nests with and without augmented Eriocephalus compounds suggest that Eriocephalus chemicals do not repel nest predators. Similarly, our choice experiments with captive snakes suggest that Eriocephalus compounds do not repel egg-­‐eating snakes from resting under hides or consuming eggs from nests with Eriocephalus compounds (Fig. 5-­‐3). 5.5.3 Nest predation and selection for earlier breeding Karoo prinias suffered exceptionally high rates of nest predation at the Koeberg Nature Reserve. Mayfield estimates of daily nest predation for prinias during the laying and incubation stage are 15.9% per day (759 exposure days, 121 nest failures), and only slightly lower when considered over the entire nesting cycle (14.9% per day). For comparison, Mayfield estimates for temperate and tropical breeding species typically range from 3–9% (Martin et al. 2006). High rates of nest predation are characteristic of most bird species breeding at Koeberg (Martin et al. 2006). In an earlier study of Karoo prinia nest success, Nalwanga et al. (2004) reported daily nest predation rates of 7.5% (presumably using Mayfield estimates over the entire nesting cycle), nearly half as high as we found, despite the fact that 85 both Nalwanga et al.’s study and ours occurred at Koeberg Nature Reserve. The cause of this difference is unknown. In both studies, the frequency of nest checks were similar (if anything, nest checks were slightly more frequent in Nalwanga et al. 2004), suggesting that changes or variation in local or regional land use, climate, breeding bird density, or predator abundance may have caused changes in nest predation rates for prinias. Our survival analyses suggest that predation on Karoo prinia nests increases with warmer ambient temperatures, later nesting attempts, and when nesting higher above the ground. Decreased survival with both warmer temperatures and nesting later in the season should favor early breeding in Karoo prinias, presumably because predators, especially snakes, are less active in cold temperatures. Yet prinias apparently have not shifted their breeding phenology over time (Percy FitzPatrick Institute for African Ornithology, nest-­‐record card repository), suggesting other constraints. One hypothesis for why prinias do not breed earlier in the season is that food availability and cold ambient temperatures prevent early breeding attempts (Martin 1987). 5.6 Conclusions Overall, our field experiments suggest that augmented levels of Eriocephalus compounds do not reduce predation on Karoo prinia nests. Experiments using six captive Dasypeltis snakes further suggest that Eriocephalus compounds do not reduce visitation to nests, consumption of eggs, or function to deter snakes. Although we cannot rule out the hypothesis that Eriocephalus compounds reduce nest predation, our results suggest that Eriocephalus compounds are unlikely to 86 repel nest predators. If Eriocephalus compounds do reduce nest predation, they may be most effective as a locally-­‐specific chemical camouflage. 87 40
All nests
n = 131
Number of Karoo prinia nests depredated
20
0
30
20
Egg-eating snake
n = 88
10
0
15
10
Non egg-eating snake
predators n = 43
5
0
0
8
16
24
Day in nesting cycle
Figure 5-­‐1. Histograms showing when in the nesting cycle Karoo prinia nests were depredated, for all nests (top), nests suspected to have been depredated by rhombic egg-­‐eating snakes (middle), and nests suspected to have been depredated by non egg-­‐eating snake predators (bottom). Histograms are binned in two day intervals and breeding stage is divided into three color-­‐coded stages: light gray bars indicate the average four day egg laying stage, blue bars indicate the average 14-­‐day incubation stage, and brown bars indicate the nestling stage (normally ~14 days) (Rowan and Broekhuysen 1962); dark grey bars in middle histogram show 33 nests beneath which we found regurgitated egg shells, confirming predation by egg-­‐eating snakes. Nestling stage is restricted to 10 days in this figure because no predation events occurred after day eight of the nestling period. Nest predation is highest during the laying and early incubation stages. 88 A
Rhombic egg-eating snakes
Survival probability
1.0
Exp n=44
Con n=46
0.5
0.0
0
5
10
15
Exposure days
B
Non rhombic egg-eating snakes
Suvival probability
1.0
Exp n=26
Con n=19
0.5
0.0
0
5
10
15
Exposure days
Figure 5-­‐2. Kaplan-­‐Meier survival curves for nests suspected to have been depredated by rhombic egg-­‐eating snakes (A) and predators other than rhombic egg-­‐eating snakes (B). Total sample sizes are control n = 62 and experimental n = 66. Augmenting levels of Eriocephalus compounds did not reduce predation rates on Karoo prinia nests (treatment p>0.61 for both predator groups). 89 3
0
Con Exp
Number of
nights hide chosen
Number of
eggs consumed
Number of first visits
8
6
4
0
Con Exp
20
10
0
Con Exp
Figure 5-­‐3. Plots showing the number of times captive Dasypeltis snakes (i) first visited nests, (ii) ate eggs from nests, or (iii) rested under hides, supplemented with either Eriocephalus compounds diluted in mineral oil (Exp) or mineral oil without Eriocephalus compounds (Con). We found no evidence that Eriocephalus compounds reduced the frequency that Dasypeltis snakes first visited nests, the number of eggs that they consumed, or their selection of hides (Binomial tests, p>0.77 for all tests). Sample size: n = 6 snakes, with 2-­‐3 trials per snake (see Methods for details). 90 Table 5-­‐1A. Best performing models (ΔAICc < 2) testing the influence of treatment (Eriocephalus compounds augmented (exp) vs. not augmented (con)) on the survival of Karoo prinia nests during the laying and incubation periods. Models also included the following other predictor variables: maximum temperature at night, nest height, first egg date, and nest bush species. We dropped Eriocephalus mass from our model because this factor was not retained in any top model (ΔAICc < 2) during initial model selection, and because dropping this term allowed us to include nests without Eriocephalus mass data. Sample sizes are: control, n = 53; experimental, n = 61. Models df Log likelihood AICc ΔAICc Weight Temperature + nest height 2 -­‐403.52 811.07 0.00 0.42 Temperature + nest height + nest density 3 -­‐403.14 812.33 1.25 0.23 Temperature + nest height + treatment 3 -­‐403.31 812.68 1.61 0.19 Temperature + nest height + first egg date 3 -­‐403.48 813.01 1.94 0.16 Table 5-­‐1B. Parameter estimates averaged across all best performing models (ΔAICc < 2) in Table 5-­‐1A for factors influencing the survival of Karoo prinia nests during the laying and incubation periods. Treatment estimates are given for nests receiving augmented Eriocephalus compounds (exp) relative to nests receiving no Eriocephalus compounds (con). Sample sizes are: control, n = 53; experimental, n = 61. Parameters Estimate SE z p Temperature 0.168 0.034 4.90 <0.0001 Nest height 0.014 0.006 2.31 0.021 Nest density -­‐0.007 0.021 0.33 0.74 Treatment -­‐0.024 0.098 0.24 0.81 First egg date 0.0003 0.003 0.12 0.91 91 Table 5-­‐2A. Best performing models (ΔAICc < 2) testing the influence of predator type (rhombic egg-­‐eating snake vs. non-­‐
snake predators) on the survival of Karoo prinia nests during the laying and incubation period. Models also included the following other predictor variables: treatment (Eriocephalus compounds augmented (exp) vs. not augmented (con)), maximum temperature at night, nest height, first egg date, residual Eriocephalus mass (controlling for the number of days the nest was active), and nest bush species. Analysis includes only depredated nests and excludes nests with missing values for nest height and Eriocephalus mass because both of these variables were identified as important predictors during initial model selection using AICc. Sample size: control, n = 45; experimental, n = 52. Models df Log likelihood AICc ΔAICc Weight Predator type + first egg date + temperature + nest height + predator type:first egg + predator type:temperature 6 -­‐335.58 683.40 0.00 0.28 Predator type + first egg date + temperature + nest height + nest density + predator type:first egg + predator type:temperature 7 -­‐334.86 684.02 0.63 0.21 Predator type + first egg date + temperature + predator type:first egg + predator type:temperature 5 -­‐337.22 684.60 1.20 0.16 Predator type + first egg date + temperature + nest height + Eriocephalus mass + predator type:first egg + predator type:temperature 7 -­‐335.37 685.04 1.65 0.12 Predator type + first egg date + temperature + nest height + treatment + predator type:first egg + predator type:temperature 7 -­‐335.41 685.12 1.73 0.12 Predator type + first egg date + temperature + nest density + predator type:first egg + predator type:temperature 6 -­‐336.53 685.28 1.88 0.11 Table 5-­‐2B. Parameter estimates averaged across all best performing models (ΔAICc < 2) in Table 5-­‐2A for factors influencing the survival of Karoo prinia nests during the laying and incubation period. Treatment estimates are given for nests receiving augmented Eriocephalus compounds (exp) relative to nests receiving no additional compounds (con). Sample size: control, n = 45; experimental, n = 52. Parameters Estimate SE z p Predator type 23.052 6.490 3.55 0.0004 92 First egg date Temperature Nest height Predator type:first egg date Predator type:temperature Nest density Eriocephalus mass Treatment 0.092 -­‐0.081 0.010 -­‐0.098 0.251 -­‐0.015 -­‐0.014 -­‐0.014 0.024 0.097 0.008 0.025 0.111 0.032 0.074 0.083 3.77 0.83 1.15 3.85 2.25 0.48 0.19 0.18 0.0002 0.41 0.25 0.0001 0.024 0.63 0.84 0.86 93 Table 5-­‐3A. Best performing models (ΔAICc < 2) testing the influence of treatment (Eriocephalus compounds augmented (exp) vs. not augmented (con)) on the survival of Karoo prinia nests that were suspected to have been depredated by rhombic egg-­‐
eating snakes during the laying and incubation period. Models also included the following predictor variables: maximum temperature at night, nest height, first egg date, residual Eriocephalus mass (controlling for the number of days the nest was active), and nest bush species. Sample sizes: control, n = 43; experimental, n = 44. Models df Log likelihood AICc ΔAICc Weight Temperature + nest height 2 -­‐281.47 566.97 0.00 0.31 Temperature 1 -­‐283.04 568.09 1.12 0.18 Temperature + nest height + treatment 3 -­‐281.21 568.48 1.51 0.14 Temperature + nest height + Eriocephalus mass 3 -­‐281.27 568.61 1.64 0.13 Temperature + nest height + nest density 3 -­‐281.35 568.77 1.80 0.12 Temperature + nest height + first egg date 3 -­‐281.43 568.93 1.96 0.11 Table 5-­‐3B. Parameter estimates averaged across all best performing models (ΔAICc < 2) in Table 5-­‐3A for factors influencing the survival of Karoo prinia nests that were thought to have been depredated by rhombic egg-­‐eating snakes during the laying and incubation periods. Treatment estimates are given for nests receiving augmented Eriocephalus compounds (exp) relative to nests receiving no additional compounds (con). Sample sizes: control, n = 43; experimental, n = 44. Parameters Estimate SE z p Temperature 0.208 0.042 4.96 <0.0001 Nest height 0.011 0.008 1.31 0.19 Treatment -­‐0.024 0.103 0.23 0.82 Eriocephalus mass -­‐0.014 0.073 0.19 0.84 Nest density -­‐0.003 0.016 0.15 0.88 First egg date -­‐0.0002 0.003 0.08 0.93 94 Table 5-­‐4A. Best performing models (ΔAICc < 2) testing the influence of treatment (Eriocephalus compounds augmented (exp) vs. not augmented (con)) on the survival of Karoo prinia nests that were thought to have been depredated by predators other than rhombic egg-­‐eating snakes during the laying and incubation periods. Model also included the following predictor variables: maximum temperature at night, first egg date, and nest bush species. We dropped Eriocephalus mass and nest height from our model because these variables were not retained in any top model (ΔAICc < 2) during initial model selection, and because dropping them allowed us to include nests with data missing for these variables. Sample sizes: control, n = 19; experimental, n = 26. Models df Log likelihood AICc ΔAICc Weight First egg date 1 -­‐113.44 228.90 0.00 0.30 First egg date + temperature 2 -­‐112.72 229.50 0.60 0.23 First egg date + treatment 2 -­‐112.86 229.76 0.86 0.20 First egg date + temperature + treatment 3 -­‐112.04 230.17 1.27 0.16 First egg date + nest density 2 -­‐113.41 230.87 1.97 0.11 Table 5-­‐4B. Parameter estimates averaged across all best performing models (ΔAICc < 2) in Table 5-­‐4A for factors influencing survival of Karoo prinia nests that were thought to have been depredated by non egg-­‐eating snake predators during the laying and incubation periods. Treatment estimates are given for nests receiving augmented Eriocephalus compounds (exp) relative to nests receiving no additional compounds (con). Sample sizes: control, n = 19; experimental, n = 26. Parameters Estimate SE z p First egg date 0.034 0.014 2.46 0.014 Temperature 0.030 0.053 0.56 0.57 Treatment 0.152 0.307 0.50 0.62 Nest density 0.002 0.021 0.08 0.93 95 Chapter 6: Discussion and future research. 6.1 Trait co-­‐option and an evolutionary perspective of community ecology Recent commentaries on community ecology have addressed a pointed question: What is the value of community ecology (Lawton 1999, Vellend 2010, Martin 2014)? On one hand, it provides exceptional detail for understanding local, species-­‐specific interactions and informing management and conservation strategies (Simberloff 2004). On the other hand, community ecology struggles to provide general rules or frameworks that create a broader understanding of species interactions and community composition (Lawton 1999, Martin 2014). Integrating an evolutionary framework into community ecology has the potential to strengthen its predictive power and improve our understanding of species interactions and community composition. For example, the processes that determine community composition parallel the evolutionary processes that determine genetic diversity—selection (whether a species persists in a community), drift (stochastic changes in a species' abundance), mutation (the generation of new species), and migration (the movement of species across a landscape) (Vellend 2010). Adopting and applying this framework to the study of community ecology holds promise for identifying repeatable patterns and rules for community assembly and composition (Vellend 2010). Similarly, understanding the abundance and distribution of species—key goals of community ecology—may be most productively approached from the perspective of evolutionary trade-­‐offs (Martin 2014). Trade-­‐offs create ecological opportunities that prevent a single species from dominating resources and occupying all habitats, allowing diverse species to coexist 96 (Martin 2014). Such trade-­‐offs can be understood from their mechanistic basis (e.g., genetic, developmental, physiological trade-­‐offs) to their consequences for broad patterns of abundance, distribution, and diversity, and highlight repeatable patterns that extend across taxonomic groups and environments (Martin 2014). Both of these perspectives (Vellend 2010, Martin 2014) argue that the generalities of community assembly, distribution, and abundance—the patterns that community ecologists study—will be best understood by integrating an evolutionary perspective into community ecology. The perspective of trait co-­‐option that I present in my dissertation echoes some of the central ideas of Vellend (2010) and Martin (2014): incorporating an evolutionary perspective into studies of species interactions will increase the generality and predictive power of community ecology. Central to the perspective of trait co-­‐option is viewing species interactions as an alternative to typical trait evolution. Many interactions provide species with adaptive traits, yet rarely have species interactions been viewed as an alternative to trait evolution (Mittelbach 2012). 6.2 The predictive power of a trait co-­‐option framework The perspective of trait co-­‐option leads to at least ten hypotheses for when and where selection may favor the co-­‐option of traits as an alternative to trait evolution. For example, when constraints limit the possibility of evolving novel or adaptive traits (e.g., evolutionary history restricts developmental pathways, long generation times limit rates of mutation and recombination, strong selection reduces genetic diversity), or when organisms are faced with strong selective pressures (e.g., rapid 97 environmental change, antagonistic co-­‐evolutionary dynamics), we should expect many species to co-­‐opt traits from other species rather than evolve these traits themselves (Rohwer dissertation, Chapter 2). Thus, viewing species interactions from the perspective of trait co-­‐option leads to broad and testable hypotheses, with the potential to reveal generalizable patterns across a diversity of species and environments. Testing these hypotheses, however, will require creativity. For example, primates use medicinal plants by rubbing them on their fur and by eating them. Plants used for fur-­‐rubbing represent clear cases of co-­‐option (e.g., primates gain access to chemical compounds that reduce ectoparasites; Baker 1996, Huffman 1997), but plants that are eaten may be food or medicine, or both (Huffman 1997). Diets of wild gorillas (Gorilla gorilla) include over 118 plant species, most of which have medicinal properties (Cousins and Huffman 2002), making the distinction between food and adaptive traits difficult (Huffman 1997). For widely distributed species of primates, geographic variation in both the traits available for co-­‐option (e.g., diversity of plants), and the strength of selection favoring co-­‐option (e.g., diversity of parasites, intensity of infection), may confound tests of hypotheses outlined in Chapter 2. Additionally, the number of studies addressing medicinal plant use in primates is biased towards the great apes. Limitations in the availability of trait co-­‐option data will make testing these predictions challenging. Primates represent an exceptionally well-­‐studied group. Rarely do we have a thorough understanding of the number of organisms with which a species interacts, the evolutionary history of the interactions, the selective advantages of the 98 interactions, and the traits important to the interactions. This lack of knowledge may, in part, result from species interactions being studied within a traditional framework, rather than viewing interactions as a mechanism for the co-­‐option of traits. Regardless, the general lack of data highlights the need for detailed studies of species interactions in nature, which was the second part of my dissertation. 6.3 Describing and understanding bird-­‐Eriocephalus interactions The second part of my dissertation (Chapters 3-­‐5) provided a detailed study of a poorly known, but intriguing interaction between birds and Eriocephalus plants in southern Africa. The results of this work illustrate that the interactions between these species are complex. Several species of birds incorporate the seed material of Eriocephalus into their nests, suggesting that both birds and plants may benefit from the interaction; plants via seed dispersal, and birds via increased reproductive success. However, different bird species interact with Eriocephalus in different ways. Karoo prinias appear to be fluff thieves rather than seed dispersers. Prinias actively avoided gathering Eriocephalus capitula when collecting Eriocephalus fluff by holding capitula in place with their toes and by dropping capitula that they inadvertently removed from plants. By contrast, other species of birds (e.g., Cape bulbul, Cape white-­‐eye, see Rohwer dissertation, Chapter 3) collected and incorporated many Eriocephalus capitula into their nests. Thus, the interaction between Karoo prinias and Eriocephalus appears costly to plants, as prinias reduce their seed dispersal potential, but interactions between other birds and Eriocephalus may benefit plants via directed seed dispersal. 99 The bird-­‐Eriocephalus interaction adds to a growing literature recognizing the dynamic nature of species interactions (Palmer et al. 2010, Heath and Stinchcombe 2014). Interactions among species often fluctuate from costly to beneficial (Bronstein 1994). When interactions involve multiple species, such as in the case of Eriocephalus plants interacting with a diversity of birds, the costs and benefits of the interactions often vary across species, creating a dynamic selective background for traits involved in the interaction (Palmer et al. 2008, Galetti et al. 2013, Heath and Stinchcombe 2014). Our experiments testing the selective benefits of Eriocephalus compounds in prinia nests supported the hypothesis that compounds reduced nest ectoparasites (Rohwer dissertation, Chapter 4). This finding is consistent with broader predictions of trait co-­‐option—chemical compounds in plants often function as a defense against insect herbivores, microbes, and fungi (Berenbaum 1995), and thus chemical compounds should be most effective against similar selective pressures, such as insect parasites in bird nests (Rohwer dissertation, Chapter 2, hypothesis 10). Indeed, this explanation could help to explain why our field experiments revealed no evidence that Eriocephalus compounds reduced predation from rhombic egg-­‐eating snakes; Eriocephalus plants have no evolutionary history of interacting with snakes, and thus these compounds should be less likely to impact snake predation on bird nests. Although experiments suggest a role for Eriocephalus secondary compounds in reducing nest parasitism but not predation, the results of the experiments are not entirely conclusive. During field experiments in South Africa, too few nests survived 100 until the nestling stage when they could support ectoparasites, preventing us from directly testing the ectoparasite hypothesis in South Africa. In the few nests that did survive long enough to support ectoparasites (i.e., until the nestling stage), we discovered no ectoparasites (Rohwer unpublished data). This result suggests that nest ectoparasitism may be relatively uncommon and, at most, a weak selective pressure at this site (Earlé 1981), at least during the 2013 breeding season. Similarly, our experiments found no evidence that Eriocephalus material reduced nest predation at our sites. However, we cannot yet rule out a role for Eriocephalus in reducing nest predation rates, particularly if these chemicals function as a locally specific chemical camouflage. Our findings that Eriocephalus compounds increased fledging success of tree swallows by reducing nest predation rates (Chapter 3) further suggests that a connection between Eriocephalus use in bird nests and nest predation (either direct or indirect) cannot yet be ruled out. 6.4 Difficulties and opportunities of the Eriocephalus-­‐bird system Not only do a diversity of birds interact with Eriocephalus plants, but the possible benefits of constructing nests with Eriocephalus material are many (Rohwer dissertation, Chapter 3). Despite the complexities of the bird-­‐Eriocephalus interaction, understanding the selective mechanisms favoring these types of interactions is critical for our broader understanding of diversity. This research project, like many others, generated more questions than it answered. All of these questions present intriguing avenues for further study. The relative ease by which Eriocephalus seed mortality can be assessed using bird nests compared to other agents of seed dispersal (e.g., frugivory), offers a 101 unique potential for studying the costs and benefits of interacting with diverse avian seed dispersers. Four questions arise from my work on the prinia-­‐Eriocephalus interactions with respect to seed dispersal in bird nests: (i) Why do Karoo prinias avoid gathering Eriocephalus capitula but other species do not? (ii) Do nests of some bird species provide better opportunities for seed germination compared to nests of other species?, (iii) Do seed germination rates vary between seeds incorporated into bird nests versus seeds that were not incorporated into nests?, and (iv) Do seed germination rates vary between nests that successfully fledge young versus nests that failed? Addressing these questions would improve our understanding of how interactions between Eriocephalus and a diverse avian community influences selection on Eriocephalus. My experimental addition of Eriocephalus compounds to tree swallow nests reduced ectoparasite loads and increased the number of young fledged compared to nests that received only mineral oil (Chapter 4). Many species of birds use chemical compounds (primarily from plants) to construct their nests (Hansell 2000), yet we know surprisingly little about these bird-­‐plant interactions. For example: How widespread is the use of aromatic materials in nest construction? How do aromatic materials increase the reproductive success of birds (e.g., do they reduce infestation of ectoparasites, interact with nest predation risk, increase adult female condition, reduce bacterial or fungal growth on eggs)? How many different types of aromatic materials (e.g., species of plants) do birds use in nest construction? Are some compounds more effective at increasing reproductive success than others? Do birds that use aromatic material in nest construction choose materials that share similar 102 chemical compounds? How does the use of plant secondary chemicals by birds alter the selective pressures acting on the plant chemicals themselves? Addressing these questions would provide important details about how selection acts on birds and plants, and help us to understand when such interactions should occur. In addition to these questions, we know very little about the natural history of rhombic egg-­‐eating snakes (and other species of egg-­‐eating snakes) and how they interact with birds. For example: How do egg-­‐eating snakes find nests? How far do foraging snakes travel in a single night? Do snakes remember the location of nests? Are egg-­‐eating snakes territorial? While natural history observations suggest egg-­‐
eating snakes rely primarily on olfactory cues, no studies have tested among alternative cues (e.g., visual, thermal, olfactory). Additionally, on multiple occasions, we observed snakes returning to the same nest two nights in a row, suggesting that egg-­‐eating snakes have a developed spatial memory. Tracking rhombic egg-­‐eating snakes would provide helpful new data on snake movement patterns and home range size. Overall, a better understanding of the natural history of egg-­‐eating snakes would help us to understand how they interact with the bird community, and to test if and how Eriocephalus could influence nest predation. 6.5 Extending back to trait-­‐co-­‐option Numerous species of birds interact with Eriocephalus plants; although these interactions are probably beneficial for birds, they likely range from costly to beneficial for plants. Approaching these interactions from the perspective of trait co-­‐option shifts our focus away from the fitness outcomes of the interaction and allows us to better understand how interactions with diverse birds affect 103 Eriocephalus traits, and how Eriocephalus traits can change how selection acts on birds. From the perspective of trait evolution, the co-­‐option of traits broadens the possibilities for accessing adaptive traits to the diversity of species with which organisms co-­‐occur. These interactions not only provide creative solutions to complex problems facing organisms but also tie the evolutionary trajectories of interacting species with one another. Overall, viewing species interactions from the perspective of trait co-­‐option—as an alternative to typical trait evolution—has the potential to broaden and enrich our understanding of nature and biodiversity. 104 References Agrawal AF, Stinchcombe JR. 2009. How much do genetic covariances alter the rate of adaptation? Proc R Soc B 276: 1183–1191. Amer A, Mehlhorn H. 2006. Larvicidal effects of various essential oils against Aedes, Anopheles, and Culex larvae (Diptera, Culicidae). Parasitol Res 99: 466–472. Arnold SJ. 1992. Constraints on phenotypic evolution. Am Nat 140: S85–S107. Augsperger CK. 1986. Morphology and dispersal potential of wind-­‐dispersed diaspores of Neotropical trees. Am J Bot 73: 353–363. Bailey IE, Muth F, Morgan K, Meddle SL, Healy SD. 2015. Birds build camouflaged nests. Auk 132: 11–15. Baker M. 1996. Fur rubbing: use of medicinal plants by capuchin monkeys (Cebus capucinus). Am J Primatol 38: 263–270. Bartón K. 2015. MuMIn: Multi-­‐model inference (R package version 1.13.4) http://CRAN.R-­‐project.org/package=MuMIn Bates MF, Little IT. 2013. Predation on the eggs of ground-­‐nesting birds by Dasypeltis scabra (Linnaeus 1758) in the moist highland grasslands of South Africa. Afr J Herpetol 62: 125–134. Benjamini Y, Krieger A, Yekutieli D. 2006. Adaptive linear step-­‐up procedures that control the false discovery rate. Biometrika 93: 491–507. Bennett GF, Whitworth TL. 1991. Studies on the life history of some species of Protocalliphora (Diptera: Calliphoridae). Can J Zool 69: 2048–2058. Berenbaum MR. 1995. The chemistry of defense. Proc Natl Acad Sci USA 92: 2–8. Best AR, Lewis Z, Hurst GDD, Lizé A. 2012. Thermal environment during and outside courtship jointly determine female remating rate in Drosophila melanogaster. Anim Behav 83: 1483–1490. Birkhead TR, Briskie JV, Møller AP. 1993. Male sperm reserves and copulation frequency in birds. Behav Ecol Sociobiol 32: 85–93. Bisseleua HBD, Gbewonyo SWK, Obeng-­‐Ofori D. 2008. Toxicity, growth regulatory and repellent activities of medicinal plant extracts on Musca domestica L. (Diptera: Muscidea). Afr J Biotechnol 7: 4635–4642. Bradshaw HD Jr, Schemske DW. 2003. Allele substitution at a flower colour locus produces a pollinator shift in monkeyflowers. Nature 426: 176–178. Bravo C, Bautista LM, García-­‐París M, Blanco G, Alonso JC. 2014. Males of a strongly polygynous species consume more poisonous food than females. PLoS ONE 9(10): e111057. Breidenbach CH. 1990. Thermal cues influence strikes in pitless vipers. J Herpetol 24: 448–450. 105 Brockhurst MA, Chapman T, King KC, Mank JE, Paterson S, Hurst GDD. 2014. Running with the Red Queen: the role of biotic conflicts in evolution. Proc R Soc B 281: http://dx.doi.org/10.1098/rspb.2014.1382 Brodie ED Jr. 1977. Hedgehogs use toad venom in their own defence. Nature. 268: 627–628. Brodie ED III, Brodie ED Jr. 1999. Predator-­‐prey arms races. Bioscience 49: 557–
568. Bronstein JL. 1994. Conditional outcomes in mutualistic interactions. Trends Ecol Evol 9: 214–217. Brooks WR, Mariscal RN. 1986. Population variation and behavioral changes in two pagurids in association with the sea anemone Calliactis tricolor (Lesueur). J Exp Mar Biol Ecol 103: 275–289. Bruno JF, Stachowicz JJ, Bertness MD. 2003. Inclusion of facilitation into ecological theory. Trends Ecol Evol 18: 119–125. Chalfoun AD, Martin TE. 2007. Latitudinal variation in avian incubation attentiveness and a test of the food limitation hypothesis. Anim Behav 73: 579–
585. Chapuisat M, Oppliger A, Magliano P, Christe P. 2007. Wood ants use resin to protect themselves against pathogens. Proc R Soc B 274: 2013–2017. Clark L, Mason R. 1985. Use of nest material as insecticidal and anti-­‐pathogenic agents by the European starling. Oecologia 67: 169–176. Clucas B, Owings DH, Rowe MP. 2008. Donning your enemy’s cloak: ground squirrels expoit rattlesnake scent to reduce predation risk. Proc R Soc B 275: 847–852. Colwell RK, Dunn RR, Harris NC. 2012. Coextinction and persistence of dependent species in a changing world. Annu Rev Ecol Evol Syst 43: 183–203. Conner JK. 2012. Quantitative genetic approaches to evolutionary constraint: how useful? Evolution 66: 3313–3320. Cousins D, Huffman MA. 2002. Medicinal properties in the diet of gorillas: an ethno-­‐
pharmacological evaluation. Afr Study Monogr 23:65–89. Cushman JH, Whitham TG. 1989. Conditional mutualism in a membracid-­‐ant association: temporal, age-­‐specific, and density dependent effects. Ecology 70: 1040–1047. Dawkins R, Krebs JR. 1979. Arms races between and within species. Proc R Soc B 205: 489–511. Davies J, Davies D. 2010. Origins and evolution of antibiotic resistance. Microbiol Mol Biol R 74: 417–433. Dean WRJ, Milton SJ, Siegfried WR. 1990. Dispersal of seeds as nest material by birds in semiarid Karoo shrubland. Ecology 71: 1299–1306. 106 de Cock Buning T. 1983. Thermal sensitivity as a specialization for prey capture and feeding in snakes. Am Zool 23: 363–375. de Roode JC, Lefèvre T, Hunter MD. 2013. Self medication in animals. Science 340: 150–151. Dobzhansky T. 1964. Biology, molecular and organismic. Am Zool 4: 443–452. Dubiec A, Góźdź I, Mazgajski TD. 2013. Green plant material in avian nests. Avian Biol Res 6: 133–146. Duckworth RA. 2009. The role of behavior in evolution: a search for mechanism. Evol Ecol 23: 513–531. Dumbacher JP, Wako A, Derrickson SR, Samuelson A, Spande TF, Daly JW. 2004. Melyrid beetles (Choresine): a putative source for the batrachotoxin alkaloids found in poison-­‐dart frogs and toxic passerine birds. Proc Nat Acad Sci USA 101: 15857–15860. Earlé RA. 1981. Factors governing avian breeding in Acacia savanna, Pietermaritzburg, part 3: breeding success, recruitment and clutch size. Ostrich 52: 235–243. Eisner T, Johnessee JS, Carrel J. 1974. Defensive use by an insect of a plant resin. Science 184: 996–999. English PA, Montgomerie R. 2011. Robin’s egg blue: does egg color influence male parental care? Behav Ecol Sociobiol 65: 1029–1036. Eyre-­‐Walker A, Keightley PD. 2007. The distribution of fitness effects of new mutations. Nat Rev Genet 8: 610–618 Ferko G. 2006. Composition and method for repelling squamate reptiles. US Patent No.: US 7,033,601 B1. Fox J. 2002. Cox proportional-­‐hazards regression for survival data. Appendix to An R and S-­‐PLUS companion to applied regression. http://cran.r-­‐
project.org/doc/contrib/Fox-­‐Companion/appendix-­‐cox-­‐regression.pdf. Frankino WA, Zwaan BJ, Stern DL, Brakefield PM. 2005. Natural selection and developmental constraints in the evolution of allometries. Science 307: 718–720. Fredrickson ME. 2013. Rethinking mutualism stability: cheaters and the evolution of sanctions. Q Rev Biol 88: 269–295. Futuyma DJ. 2010. Evolutionary constraint and ecological consequences. Evolution 64: 1865–1884. Futuyma DJ, Keese MC, Funk DJ. 1995. Genetic constraints on macroevolution: the evolution of host affiliation in the leaf beetle genus Ophraella. Evolution 49: 797–
809. Galetti M, Guevara R, Cortes MC, Fadini R, Von Matter S, Leite AB, Labecca F, Ribeiro T, Carvalho CS, Collevatti RG, Pires MM, Guimarães PR Jr, Brancalion PH, Ribeiro 107 MC, Jordano P. 2013. Functional extinction of birds drives rapid evolutionary changes in seed size. Science 340: 1086–1090. Gartner GEA, Greene HW. 2008. Adaptation in the African egg-­‐eating snake: a comparative approach to a classic study in evolutionary functional morphology. J Zool 275: 368–374. Ghalambor CK, Martin TE. 2001. Fecundity-­‐survival trade-­‐offs and parental risk-­‐
taking in birds. Science 292: 494–497. Goldblatt P, Manning JC. 2002. Plant diversity of the Cape region of Southern Africa. Ann Mo Bot Gard 89: 281–302. Greene S, McConnachie S, Secor S, Perrin M. 2013. The effects of body temperature and mass on the postprandial metabolic responses of the African egg-­‐eating snakes Dasypeltis scabra and Dasypeltis inornata. Comp Biochem Phys A 165: 97–105. Hall AR, Scanlan PD, Morgan AD, Buckling A. 2011. Host-­‐parasite coevolutionary arms races give way to fluctuating selection. Ecol Lett 14: 635–642. Hansell M. 2000. Bird nests and construction behaviour. Cambridge, UK: Cambridge University Press. Harper and Brothers Publishers. 1874. Miss Beecher’s Housekeeper and Healthkeeper. New York, NY: Harper and Brothers. Heath KD, Stinchcombe JR. 2014. Explaining mutualism variation: a new evolutionary paradox? Evolution 68: 309–317. Hemmes RB, Alvarado A, Hart, BL. 2002. Use of California bay foliage by wood rats for possible fumigation of nest-­‐borne ectoparasites. Behav Ecol 13: 381–385. Hilton GM, Hansell MH, Ruxton GD, Reid JM, Monaghan P. 2004. Using artificial nests to test importance of nesting material and nest shelter for incubation energetics. Auk 121: 777–787. Hockey PAR, Dean WRJ, Ryan PG. 2005. Roberts birds of southern Africa, 7th edition. Cape Town, South Africa: The Trustees of the John Voelcker Bird Book Fund. Hong TD, Ellis RH. 1990. A comparison of maturation drying, germination, and desiccation tolerance between developming seeds of Acer pseudoplatanus L. and Acer platanoides L. New Phytol 116: 589–596. Howe HF, Miriti MN. 2004. When seed dispersal matters. Bioscience 54: 651–660. Huffman MA. 1997. Current evidence for self-­‐medication in primates: a multidisciplinary perspective. Yearb Phys Anthropol 40: 171–200. Jackman S. 2014. pscl: Classes and Methods for R Developed in the Political Science (R package version 1.4.6). http://pscl.stanford.edu/ Jalink HR, van der Schoor R, Frandas A, van Pijlen JG, Bino RJ. 1999. Seed chlorophyll content as an indicator for seed maturity and seed quality. Acta Hortic 504: 219–
227. 108 Janzen DH. 1966. Coevolution of mutualism between ants and acacias in Central America. Evolution 20: 249–275. Janzen DH. 1977. Why fruits rot, seeds mold, and meat spoils. Am Nat 111: 691–713. Johnson LJ, Tricker PJ. 2010. Epigenomic plasticity within populations: its evolutionary significance and potential. Heredity 105: 113–121. Jones ZS. 1996. Captive maintenance and breeding of rhombic egg-­‐eating snakes (Dasypeltis scabra). Adv Herpetoculture 1: 119–126. Kerley GIH. 1991. Seed removal by rodents, birds and ants in the semi-­‐arid Karoo, South Africa. J Arid Environ 20: 63–69. Kellermann V, van Heerwaarden B, Sgrò CM, Hoffmann AA. 2009. Fundamental evolutionary limits in ecological traits drive Drosophila species distributions. Science 325: 1244–1246. Kerr B. 2007. Niche construction and cognitive evolution. Biological Theory 2: 250–
262. Kirkpatrick M. 2010. Rates of Adaptation: Why is Darwin’s machine so slow? In Evolution since Darwin. The first 150 years (eds Bell MA, Futuyma DJ, Eanes WF, Levinton JS), pp. 177–195. Sunderland, MA: Sinauer Associates. Krebs JR. Davies NB. 1997. Behavioural ecology: an evolutionary approach, 4th ed. Oxford, UK: Blackwell Scientific Publications. Kumar P, Mishra S, Malik A, Satya S. 2014 Biocontrol potential of essential oil monoterpenes against housefly, Musca domestica (Diptera: Muscidae). Ecotox Environ Safe 100: 1–6. Lawton JH. 1999. Are there general laws in ecology? Oikos 84: 177–192. Leimar O, Tullberg BS, Mallet J. 2012. Mimicry, saltational evolution, and the crossing of fitness valleys. In The Adaptive Landscape in Evolutionary Biology. (eds Svensson E, Calsbeek R.), pp. 259–270. Oxford, UK: Oxford University Press. Lloyd P, Abadi F, Altwegg R, Martin TE. 2014. South temperate birds have higher apparent adult survival than tropical birds in Africa. J Avian Biol 45: 493–500. Lloyd P, Little RM, Crowe TM. 2001. The breeding biology of the Namaqua sandgrouse, Pterocles namaqua. J Afr Ornithol 72: 169–178. Losos JB. 2008. Phylogenetic niche conservatism, phylogentic signal and the relationship between phylogenetic relatedness and ecological similarity among species. Ecol Lett 11: 995–1007. Manning J. 2007. Field guide to Fynbos. Cape Town, South Africa: Struik Nature, Random House Struik. Martin PR. 2014. Trade-­‐offs and biological diversity: integrative answers to ecological questions. In Integrative Organismal Biology (eds LB Martin, CK Ghalambor, HA Woods), pp. 291–308. New York, NY: Wiley. 109 Martin TE. 1987. Food as a limit on breeding birds: a life-­‐history perspective. Ann Rev Ecol Syst 18: 453–487. Martin TE. 1995. Avian life history evolution in relation to nest sites, nest predation, and food. Ecol Monogr 65: 101–127. Martin TE, Bassar RD, Bassar SK, Fontaine JJ, Lloyd P, Mathewson HA, Niklison AM, Chalfoun A. 2006. Life-­‐history and ecological correlates of geographic variation in egg and clutch mass among passerine species. Evolution 60: 390–398. Martin TE, Geupel GR. 1993. Nest-­‐monitoring plots: methods for locating nests and monitoring success. J Field Ornithol 64: 507–519. Martin TE, Lloyd P, Bosque C, Barton DC, Biancucci AL, Cheng Y-­‐R, Ton R. 2011. Growth rate variation among passerine species in tropical and temperate sites: an antagonistic interaction between parental food provisioning and nest predation risk. Evolution 65: 1607–1622. Martin TE, Martin PR, Olson CR, Heidinger BJ, Fontaine JJ. 2000. Parental care and clutch sizes in North and South American birds. Science 287: 1482–1485. Martin TE, Roper JJ. 1988. Nest predation and nest-­‐site selection of a western population of the hermit thrush. Condor 90: 51–57. Mathew S, Perreault C. 2015. Behavioural variation in 172 small-­‐scale societies indicates that social learning is the main mode of human adaptation. Proc R Soc B 20150061. http://dx.doi.org/10.1098/rspb.2015.0061 Mayer VE, Frederickson ME, McKey D, Blatrix R. 2014. Current issues in the evolutionary ecology of ant-­‐plant symbioses. New Phytologist 202: 749–764. McClintock JB, Janssen J. 1990. Pteropod abduction as a chemical defence in a pelagic Antarctic amphipod. Nature 346: 462–464. McGill BJ, Enquist BJ, Weiher E, Westoby M. 2006. Rebuilding community ecology from functional traits. Trends Ecol Evol 21: 178–185. McKeller, AE. 2010. Female American redstart, Setophaga ruticilla, reuses red-­‐eyed vireo, Vireo olivaceus, nest. Can Field Nat 124: 272–273. McKinnon L, Smith PA, Nol E, Martin JL, Doyle FI, Abraham KF, Gilchrist HG, Morrison RIG, Bêty J. 2010. Lower predation risk for migratory birds at high latitudes. Science 327: 326–327. Mittelbach GG. 2012. Community ecology. Sunderland, MA: Sinauer Associates. Moczek A. 2008. On the origins of novelty in development and evolution. BioEssays 30: 432–447. Morris JJ, Lenski RE, Zinser ER. 2012. The black queen hypothesis: evolution of dependencies through adaptive gene loss. mBio 3: e00036-­‐12, doi:10.1128/mBio.00036-­‐12. Mousseau TA, Fox CW 1998. The adaptive significance of maternal effects. Trends Ecol Evol 13: 403–407. 110 Müller MAN, Herman PPJ, Kolberg HH. 2001. Fascicle 1: Eriocephalus and Lasiospermum. Flora of South Africa. 33: 1–63. Murphy ME, King JR. 1986. Composition and quality of feather sheaths produced by white-­‐crowned sparrows during the postnuptial molt. Auk 103: 822–825. Nalwanga D, Lloyd P, du Plessis MA, Martin TE. 2004. The influence of nest-­‐site characteristics on the nesting success of the Karoo prinia (Prinia maculosa). Ostrich 75: 269–274. New Directions Aromatics Inc. Material safety data sheet: Eriocephalus africanus essential oils. http://www.newdirectionsaromatics.com/msds/MSDS_CapeSnowbushEssential
Oil.pdf Njenga EW. 2005. The chemotaxonomy, phylogeny and biological activity of the genus Eriocephalus L. (Asteraceae). Ph.D. dissertation, Johannesburg, South Africa: University of the Witwatersrand. Njenga EW, van Vuuren SF, Viljoen AM. 2005. Antimicrobial activity of Eriocephalus L. species. S Afr J Bot 71: 81–87. Nur N, Holmes AL, Geupel GR. 2004. Use of survival time analysis to analyze nesting success in birds: an example using loggerhead shrikes. Condor 106: 457–471. Orr HA. 2005. The genetic theory of adaptation: a brief history. Nat Rev Genet 6: 119–127. Orr HA, Coyne JA. 1992. The genetics of adaptation: a reassessment. Am Nat 140: 725–742. Paine RT. 1966. Food web complexity and species diversity. Am Nat 100: 65–75. Palmer TM, Doak FD, Stanton ML, Bronstein JL, Kiers ET, Young TP, Goheen JR, Pringle RM. 2010. Synergy of multiple partners, including freeloaders, increases host fitness in a multispecies mutualism. Proc Nat Acad Sci USA 107: 17234–
17239. Palmer TM, Stanton ML, Young TP, Goheen JR, Pringle RM, Karban R. 2008. Breakdown of an ant-­‐plant mutualism follows the loss of large herbivores from an African savanna. Science 319: 192–195. Petit C, Hossaert-­‐McKey M, Perret P, Blondel J, Lambrechts MM. 2002. Blue tits use selected plants and olfaction to maintain an aromatic environment for nestlings. Ecol Lett 5: 585–589. Pigliucci M, Murren CJ, Schlichting CD. 2006. Phenotypic plasticity and evolution by genetic assimilation. J Exp Biol 209: 2362–2367. Pike N. 2011. Using false discovery rates of multiple comparisons in ecology and evolution. Method Ecol Evol 2: 278–282. 111 Pinheiro J, Bates D, DebRoy S, Sarkar D. 2014. nlme: linear and nonlinear mixed effects models. R package version 3.1–117. http://CRAN.Rproject.org/package=nlme Pravosudov VV, Roth TC II. 2013. Cognitive ecology of food hoarding: the evolution of spatial memory and the hippocampus. Ann Rev Ecol Evol Syst 44: 173–193. Price TD, Qvarnstrom A, Irwin DE. 2003. The role of phenotypic plasticity in driving genetic evolution. Proc R Soc B 270: 1433–1440. Queller DC, Strassmann JE. 2009. Beyond society: the evolution of organismality. Phil Trans R Soc B 364: 3143–3155. R Core Team. 2014. R: A language and environment for statistical computing (R Foundation for Statistical Computing, Vienna, Austria). http://www.R-­‐
project.org/ Rasband WS. ImageJ. U. S. National Institutes of Health, Bethesda, Maryland, USA, 1997–2014. http://imagej.nih.gov/ij/ Regnault-­‐Roger C. 1997 The potential of botanical essential oils for insect pest control. Integrated Pest Manag Rev 2: 25–34. Reinking L. 2007. Examples of image analysis using ImageJ. http://rsb.info.nih.gov/ij/docs/pdfs/examples.pdf; (accessed September 2014). Rice PJ, Coats JR. 1994. Insecticidal properties of several monoterpenoids to house fly (Diptera: Muscidae), red flour beetle (Coleoptera: Tenebrionidae), and southern corn rootworm (Coleoptera: Chrysomelidae). J Econ Entomol 87: 1172–1179. Rivers JW, Liebl AL, Owen JC, Martin LB, Betts MG. 2012. Baseline corticosterone is positively related to juvenile survival in a migrant passerine bird. Funct Ecol 26: 1127–1134. Roff DA. 2002. Life-­‐history evolution. Sunderland, MA: Sinauer Associates. Rogers CA, Robertson RJ, Stutchbury BJ. 1991. Patterns and effects of parasitism by Protocalliphora sialia on tree swallow (Tachycineta bicolor) nestlings. In Bird-­‐
parasite interactions: ecology, evolution, and behavior (eds Loye J, Zuk M), pp. 123–139. New York, NY: Oxford University Press. Romero LM, Reed JM. 2005. Collecting baseline corticosterone samples in the field: is under 3 min good enough? Comp Biochem Physiol A 140: 73–79. Ross DM. 1971. Protection of hermit crabs (Dardanus spp.) from octopus by commensal sea anemones (Calliactis spp.). Nature 230: 401–402. Rowan MK, Broekhuysen GJ. 1962. A study of the Karoo prinia. Ostrich 33: 6–30. Salazar-­‐Ciudad I, Marín-­‐Riera M. 2013. Adaptive dynamics under development based genotype-­‐phenotype maps. Nature 497: 361–365. SAS 2007. JMP Version 11.0, SAS Institute Inc., Cary, NC. 112 Schluter D. 1996. Adaptive radiation along genetic lines of least resistance. Evolution 50: 1766–1774. Schönknecht G, Weber APM, Lercher MJ. 2015. Horizontal gene acquisitions by eukaryotes as drivers of adaptive evolution. Bioessays 36: 9–20. Sibly RM, Calow P. 1986. Physiological ecology of animals: an evolutionary approach. Oxford, UK: Blackwell Scientific Publications. Simberloff D. 2004. Community ecology: is it time to move on? Am Nat 163: 787–
799. Soler JJ, Neve L, Martínez JG, Soler M. 2001. Nest size affects clutch size and the start of incubation in magpies: an experimental study. Behav Ecol 12: 301–307. Sonerud GA, Fjeld PE. 1987. Long-­‐term memory in egg predators: an experiment with a hooded crow. Ornis Scandinavica 18: 323–325. Stanton ML, Palmer TM. 2011. The high cost of mutualism: effects of four species of east African ant symbionts on their myrmecophyte host tree. Ecology 92: 1073–
1082. Stachowicz JJ. 2001. Mutualism, facilitation, and the structure of ecological communities. Bioscience 51: 235–246. Stachowicz JJ, Hay ME. 1999. Reducing predation through chemically mediated camouflage: indirect effects of plant defenses on herbivores. Ecology 80: 495–
509. Stearns SC. 1992. The evolution of life histories. Oxford, UK: Oxford University Press. Stevens M, Merilaita S. 2009. Animal camouflage: current issues and new perspectives. Phil Trans Roy Soc B 364: 423–427. Suárez-­‐Rodríguez M, López-­‐Rull I, Garcia CM. 2013. Incorporation of cigarette butts into nests reduces nest ectoparasite load in urban birds: new ingredients for an old recipe? Biol Lett 9: doi: 10.1098/rsbl.2012.0931 Swennen C. 1968. Nest protection of eiderducks and shovelers by means of feaces. Ardea 56: 248–258. Systema Dipterorum, Version 1.5 (ed Pape T, Evenhuis NL). http://www.diptera.org/ accessed 12 November 2014. Tarboton W. 2011. Roberts nests and eggs of southern African birds. Cape Town, South Africa: John Voelcker Bird Book Fund. Tewksbury JJ, Reagan KM, Machnicki NJ, Carlo TA, Haak DC, Peñaloza ALC, Levey DJ. 2008. Evolutionary ecology of pungency in wild chilies. Proc Nat Acad Sci USA 105: 11808–11811. Thompson JN. 1994. The coevolutionary process. Chicago, IL: University of Chicago Press. 113 Tomás G, Merino S, Martínez-­‐de la Puente J, Moreno J, Morales J, Rivero-­‐de Aguilar J. 2013. Nest size and aromatic plants in the nest as sexually selected female traits in blue tits. Behav Ecol 24: 926–934. Vander Wall SB, Longland WS. 2004. Diplochory: are two seed dispersers better than one? Trends Ecol Evol 19: 155–161. Van Valen L. 1973. A new evolutionary law. Evol Theory 1: 1–30. Vellend M. 2010. Conceptual synthesis in community ecology. Q Rev Biol 85: 183–
206. Webb DR. 1987. Thermal tolerance of avian embryos: a review. Condor 89: 874–
898. Weldon PJ, Aldrich JR, Klum JA, Oliver JE, Debboun M. 2003. Benzoquinones from millipedes deter mosquitoes and elicit self-­‐anointing in capuchin monkeys (Cebus spp.). Naturwissenschaften 90: 301–304. West-­‐Eberhart MJ. 2003. Developmental plasticity and evolution. New York, NY: Oxford University Press. Wimberger PH. 1984. The use of green plant material in bird nests to avoid ectoparasites. Auk 101: 615–618. Wingfield JC, Vleck CM, and Moore MC. 1992. Season changes of the adrenocortical response to stress in birds of the Sonoran Desert. J Exp Zool 264: 419–428. Winkler, DW, Hallinger KK, Ardia DR, Robertson RJ, Stutchbury BJ, Cohen RR. 2011. Tree Swallow (Tachycineta bicolor). In The Birds of North America Online (ed Poole A). Ithaca, NY: Cornell Lab of Ornithology Whitworth TL, Bennett GF. 1992. Pathogenicity of larval Protocalliphora (Diptera: Calliphoridae) parasitizing nestling birds. Can J Zool 70: 2184–2191. Wright S. 1932. The Roles of mutation, inbreeding, crossbreeding and selection in evolution. In Proceedings of the 6th Annual Congress of Genetics pp. 356–366. Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smith GM. 2009. Mixed effects models and extensions in ecology with R. New York, NY: Springer. 114 Appendix A: Supplementary material for Chapter 3 Detailed methods for quantifying Eriocephalus fluff on the interior and exterior of nests Nest interiors. We estimated the proportion of the nest interior that was covered in Eriocephalus fluff using ImageJ software (Rasband 2014). We calculated proportions by dividing the total area of pale colored material (primarily Eriocephalus fluff) on the nest interior by the total area of the nest interior. We estimated the area of Eriocephalus fluff and nest interior for each nest by photographing cut-­‐in-­‐half nests against a dark background; we took all photos using a Panasonic Lumix FZ150 camera (Panasonic Corporation, Japan) and included a 10cm ruler beside all nests for scale. We opened photos in ImageJ and: (i) converted color images to 8-­‐bit grey scale (ii) set the scale for each image, (iii) adjusted the pixel threshold to capture the total nest area, and then the area of pale colored nest lining, (iv) converted the adjusted image to binary so that the nest or the pale lining became dark colored, (v) summed the area of dark color representing the nest or pale lining material; this procedure follows Reinking (2007). While we took all nest pictures against a dark background with similar ambient light levels, the pixel threshold used to quantify Eriocephalus material inside each nest was slightly different because of small differences in exposure levels, ambient light levels, and materials used in nest construction. To account for this, VGR did all measures in ImageJ to control for inter-­‐
observer variation, and a subset of 15 of the 117 nests were scored five times, each on different days, to assess the repeatability of our measures; repeatability was high 115 (Intraclass correlation coefficient: 0.93, 95% CI range 0.861–0.972, F14,60 = 67.1, p < 0.00001). Nest exteriors. We estimated the proportion of the nest exterior that was covered by Eriocephalus fluff using nests that were cut in half. We placed nest halves with the nest exterior facing up, then sectioned each nest into thirds where the bottom third represented the base of the nest, the middle third the nest-­‐walls, and the top third the nest-­‐roof. We visually estimated the proportion of Eriocephalus fluff in each nest section, covering the other sections with gray-­‐colored cardboard. All estimates were made by VGR. We then averaged the estimated proportions of Eriocephalus material visible on the exterior base, walls and roof of each nest to generate our measure of the proportion of Eriocephalus material visible on the nest exterior. Statistical analyses of Eriocephalus capitula and fluff in Karoo prinia nests. For the analysis of Eriocephalus capitula in prinia nests, we excluded nests that contained no Eriocephalus material (and thus no capitula). We examined the factors that influenced the number of Eriocephalus capitula in prinia nests using hurdle models that are suitable for zero-­‐inflated count data, following Zuur et al. (2009), chapter 11. Hurdle models separately test the effects of predictor variables on the zero versus non-­‐zero (bivariate) component of the response variable, and on variation in the non-­‐zero component of the response variable (Zuur et al. 2009). In our analysis, the response variable was the number of Eriocephalus capitula in nests and our predictor variables were: proximity of the nest to the nearest Eriocephalus bush, first egg date, minimum ambient temperature, nest bush species, nest height, 116 and the number of days the nest remained active. We transformed the following variables prior to analysis: proximity to closest Eriocephalus bush [natural logarithm (closest Eriocephalus bush + 3], first egg date (first egg date^0.2), nest height (nest height^0.1), number of days a nest remained active (-­‐1/(days active ^0.3)), and minimum ambient temperature (temperature^0.7). For nest bush species, we grouped all bush species with fewer than 4 nests into a single category “other”, leaving a total of eight categories for bush species. For analyses of Eriocephalus fluff, we ran four models each using a different measure of Eriocephalus fluff [i.e., total mass of Eriocephalus fluff, average depth of fluff in the base of nests, the proportion of pale colored material (primarily Eriocephalus fluff) on the interior of the nest, and the proportion of fluff on the exterior of nests]. We examined these four different measures of Eriocephalus fluff separately because Eriocephalus material in different locations of the nest may have different functions, and because birds may prioritize the placement of Eriocephalus material in nests depending on the availability of material. For analyses of depth and mass of Eriocephalus fluff, we used hurdle models, and for analyses of the amount of Eriocephalus material on the interior and exterior of nests, we used Generalized Linear Models (GLMs). Both Eriocephalus depth and mass were bi-­‐modal, with peaks at zero and near the mid-­‐point of each variable. We converted these continuous variables with decimal measures into whole-­‐number integers (similar to count data), so that we could run hurdle models. For depth of Eriocephalus fluff, we binned measures into 2mm intervals as follows: nests with < 2mm of Eriocephalus fluff were scored as 0, 117 2-­‐4mm as 1, 4-­‐6mm as 2, etc.; this binning scale resulted in bins from 0–13, spanning Eriocephalus depth measures 0–27mm. For total mass of Eriocephalus fluff, we binned measures into 0.25g intervals with nests having <0.25g scored as 0, 0.26-­‐
0.5g as 1, 0.51-­‐0.75g as 2, 0.76-­‐1.0g as 3, etc.; this binning scales resulted in bins ranging from 0–13, spanning Eriocephalus mass measures 0–3.4g. Binning continuous variables risks creating artificial categories, thus we ran both of these models with different binning categories (i.e., for Eriocephalus depth, we used 1mm and 3mm bins; and for Eriocephalus mass we used 0.1g and 0.5g bins); bin size did not change the best performing models or which predictor variables were important. We excluded one nest from analyses of the depth of Eriocephalus fluff because the depth of fluff was over 5 standard deviations larger than the mean of all other nests (mean ± 1sd of nests with outlier excluded 12.9 ± 7.7mm, depth of fluff of outlier 52.5mm); no other prinia nests approached this value (max depth of Eriocephalus fluff in all other nests 27mm). For GLMs, we transformed the amount of Eriocephalus material on the exterior and interior of nests prior to analyses using the following formulas: exterior (exterior^0.6), and for interior, we used a Johnson Su transformation in JMP (SAS 2007) with the following formula: 𝐴𝑟𝑐𝑆𝑖𝑛𝐻
(𝑖𝑛𝑡𝑒𝑟𝑖𝑜𝑟 − 0.967)
∗ 1.666 + 5.190 0.0232
118 For all four analyses of Eriocephalus fluff, we used a single measure of fluff as the dependent variable (i.e., depth of fluff in nest base, total mass, proportion pale material in the interior, proportion fluff on the exterior) and, in all analyses, we used the same predictor variables: proximity to the nearest Eriocephalus bush, first egg date, minimum ambient temperature, nest bush species, nest height, and the number of days the nest remained active. We transformed predictor variables prior to analyses to better fit the assumptions of our models using the following transformations: first egg date [natural logarithm (first egg date)], distance to closest Eriocephalus bush (closest bush^0.3), minimum ambient temperature (temperature^0.7), number of days a nest remained active [-­‐1/(days active^0.3)], nest height (height^0.2). For nest bush species, we again grouped all bush species with fewer than 4 nests into a single category “other”. We identified the best performing models using AIC, corrected for small sample sizes (AICc), and the dredge command in package MuMIn (Bartón 2015) in R (R Core Team 2014). Prior to model selection, we checked the assumptions of hurdle models and GLMs following Zuur et al. (2009). For hurdle models, we plotted Pearson’s residuals against predictor variables to ensure that model residuals showed no patterns with predictor variables. For GLMs, we plotted model residuals against predictor variables and checked that the distribution of model residuals did not deviate from normality using Shapiro-­‐Wilk tests. For both hurdle models and GLMs, we check for similar residual variance among the categorical predictor variable “nest bush species” using Bartlett’s test. We present best performing models (those with ΔAICc<2) and averaged parameter estimates, which take into 119 account the frequency that predictor variables were included in best performing models and the fit of those models. Finally, because our different measures of Eriocephalus fluff were not independent, we corrected for false discovery rates following Benjamini et al. (2006) and Pike (2011). 120 Table S1. Descriptions of the phenological categories of Eriocephalus. Stage Description 1 Bud. <50% of the capitula are blooming. 2 Early flowering. >50% of the capitula are blooming, flowers just beginning to open giving bush a slightly rosy appearance. 3 Peak flowering. Nearly all capitula are blooming. Overall bush has rosy and yellowish tints from blossoms. Fluff just beginning to develop but its length does not exceed the length of its associated capitulum. 4 Late flowering. Flowers still visible, but the fluff is beginning to develop. Fluff appears as sparse plumes, which are generally shorter than capitula. 5 Sparse fluff. Florets still visible but surrounded by sparse plumes of fluff. Length of fluff generally equals length of capitula. 6 Early fluff. Individual capitula are recognizable but nearly invisible because of thick, short fluff surrounding them. Bush begins to appear white. 7 Medium fluff. Center of capitula has brown dimple and length of fluff exceeds length of capitula. Capitula generally appear white, but fluff is often projected toward the tip of the capitula. Bush has overall white appearance. 8 Thick fluff. Bush has blazing white appearance. Fluff is so thick that it obscures leaves and often makes recognition of individual capitula difficult. Capitula continue to have brown dimple in their center. Fluff appears fully developed and often extends away from the capitula in a globe like fashion. 121 Table S2A. Best performing models (ΔAICc < 2) for the number of Eriocephalus capitula in Karoo prinia nests as a function of: first egg date, proximity of nests to nearest Eriocephalus bush, ambient temperature, number of days nests remained active, nest height and bush species in which nests were placed. We used hurdle models that examine the variation between nests with and without Eriocephalus capitula (ne), and the variation among nests with Eriocephalus capitula (we), separately. Sample sizes: nests without capitula n = 19; nest with capitula: n = 80; total nests: n = 99. Model df AICc ΔAICc Weight Log likelihood First egg date (we) + proximity (ne) 5 573.09 0 0.17 -­‐281.22 First egg date (we) + proximity (ne) + proximity (we) 6 573.16 0.07 0.16 -­‐280.12 First egg date (we) + proximity (ne) + ambient temperature (ne) 6 573.97 0.88 0.11 -­‐280.53 First egg date (we) + proximity (ne) + proximity (we) + ambient temperature (ne) 7 574.08 0.99 0.10 -­‐279.43 First egg date (we) + proximity (ne) + first egg date (ne) 6 574.30 1.21 0.09 -­‐280.69 First egg date (we) + proximity (ne) + proximity (we) + first egg date (ne) 7 574.41 1.32 0.09 -­‐279.59 First egg date (we) + proximity (ne) + ambient temperature (we) 6 574.47 1.38 0.08 -­‐280.78 First egg date (we) + proximity (ne) + days active (we) 6 574.76 1.67 0.07 -­‐280.92 First egg date (we) + proximity (ne) + proximity (we) + days active (we) 7 574.81 1.72 0.07 -­‐279.79 First egg date (we) + proximity (ne) + proximity (we) + ambient temperature (we) 7 574.96 1.87 0.06 -­‐279.87 Table S2B. Parameter estimates averaged from all best performing models (ΔAICc < 2) in Table S2A, for factors influencing the number of Eriocephalus capitula in Karoo prinia nests. Parameter First egg date (we) Proximity (ne) z p 9.55 2.91 3.29 0.0010 -­‐0.10 0.30 3.30 0.0009 Proximity (we) 0.08 0.11 0.70 0.48 Ambient temperature (ne) 0.07 0.19 0.37 0.71 First egg date (ne) 1.32 4.22 0.31 0.75 0.02 0.08 0.26 0.80 -­‐0.07 0.29 0.24 0.81 Ambient temperature (we) Days active (we) Estimate SE 122 Table S3A. Best performing models (ΔAICc < 2) for the average depth of Eriocephalus fluff in the base of Karoo prinia nests as a function of: first egg date, proximity of nests to nearest Eriocephalus bush, ambient temperature, number of days nests remained active, nest height and bush species in which nests were placed. We used hurdle models that examine the variation between nests with and without Eriocephalus fluff (ne), and the variation among nests with Eriocephalus fluff (we), separately. Sample sizes: nests without fluff n = 16; nest with fluff n = 88; total number of nests: n = 104. Model df AICc ΔAICc Weight Log likelihood Proximity (we) + first egg date (we) + proximity (ne) 5 445.23 0 0.52 -­‐217.31 Proximity (we) + first egg date (we) + proximity (ne) + days active (we) 6 446.62 1.39 0.26 -­‐216.88 Proximity (we) + first egg date (we) + proximity (ne) + nest height (ne) 6 446.96 1.73 0.22 -­‐217.04 Table S3B. Parameter estimates averaged from all best performing models (ΔAICc < 2) in Table S3A, for factors influencing the average depth of Eriocephalus fluff in the base of Karoo prinia nests. P-­‐adj corrects for false discovery rates following Benjamini et al. 2006, because we measured four different, but related, components of Eriocephalus fluff in prinia nests. Parameter Estimate SE z p p-­‐adj Proximity (we) -­‐0.18 0.04 4.33 <0.0001 <0.0001 First egg date (we) -­‐1.90 0.70 2.71 0.0068 0.017 Proximity (ne) -­‐4.41 1.27 3.46 0.0005 0.002 Days active (we) -­‐0.41 1.10 0.37 0.71 0.62 Nest height (ne) 0.80 2.86 0.28 0.78 0.62 123 Table S4A. Best performing models (ΔAICc < 2) for the total mass of Eriocephalus fluff in Karoo prinia nests as a function of: first egg date, proximity of nests to nearest Eriocephalus bush, ambient temperature, number of days nests remained active, nest height and bush species in which nests were placed. We used hurdle models that examine the variation between nests with and without Eriocephalus fluff (ne), and the variation among nests with Eriocephalus fluff (we), separately. Sample sizes: nests without fluff n = 14, nest with fluff n = 91; Total number of nests: n = 105. Model df AICc ΔAICc Weight Log likelihood First egg date (we) + proximity (we) + proximity (ne) + days active (we) 6 395.32 0 0.13 -­‐191.23 First egg date (we) + proximity (we) + proximity (ne) + days active (we) + nest height (we) 7 395.56 0.23 0.12 -­‐190.20 First egg date (we) + proximity (we) + proximity (ne) 5 395.87 0.55 0.10 -­‐192.63 First egg date (we) + proximity (we) + proximity (ne) + days active (we) + days active (ne) 7 396.66 1.33 0.07 -­‐190.75 First egg date (we) + proximity (we) + proximity (ne) + days active (we) + nest height (ne) 7 396.73 1.40 0.07 -­‐190.79 First egg date (we) + proximity (we) + proximity (ne) + nest height (we) First egg date (we) + proximity (we) + proximity (ne) + days active (we) + days active (ne) + nest height (we) 6 396.88 1.56 0.06 -­‐192.01 8 396.94 1.62 0.06 -­‐189.72 First egg date (we) + proximity (we) + proximity (ne) + days active (we) + first egg date (ne) First egg date (we) + proximity (we) + proximity (ne) + days active (we) + nest height (we) + nest height (ne) 7 396.99 1.67 0.06 -­‐190.92 8 397.01 1.69 0.06 -­‐189.75 First egg date (we) + proximity (we) + proximity (ne) + ambient temperature (we) 6 397.08 1.76 0.06 -­‐192.11 First egg date (we) + proximity (we) + proximity (ne) + days active (ne) First egg date (we) + proximity (we) + proximity (ne) + days active (we) + ambient temperature (we) 6 397.15 1.83 0.05 -­‐192.15 7 397.16 1.84 0.05 -­‐191.00 First egg date (we) + proximity (we) + proximity (ne) + nest height (ne) First egg date (we) + proximity (we) + proximity (ne) + days active (we) + nest height (we) + first egg date (ne) 6 397.22 1.90 0.05 -­‐192.18 8 397.27 1.95 0.05 -­‐189.89 124 Table S4B. Parameter estimates averaged from all best performing models (ΔAICc < 2) in Table S4A, for factors influencing the total mass of Eriocephalus fluff in Karoo prinia nests. P-­‐adj corrects for false discovery rates following Benjamini et al. 2006, because we measured four different, but related, components of Eriocephalus fluff in prinia nests. Sample sizes: n = 105 nests. Parameter Estimate SE p-­‐adj -­‐4.26 0.92 4.62 <0.0001 <0.0001 Proximity (we) -­‐0.16 0.05 3.38 0.0007 0.0023 Proximity (ne) -­‐5.39 1.90 2.84 0.0046 0.016 Days active (we) 2.23 2.20 1.01 0.31 0.57 Nest height (we) 0.17 0.32 0.55 0.59 0.62 Days active (ne) 9.37 31.30 0.30 0.76 0.62 Nest height (ne) 0.97 3.31 0.29 0.77 0.62 1.00 4.87 0.21 0.84 0.62 -­‐0.01 0.03 0.22 0.83 0.62 Ambient temperature (we) p First egg date (we) First egg date (ne) z 125 Table S5A. Best performing models (ΔAICc < 2) for the proportion of Eriocephalus fluff on the nest exterior and pale colored material (primarily Eriocephalus fluff) on the nest interior of Karoo prinia nests as a function of: first egg date, proximity of nests to the nearest Eriocephalus bush, ambient temperature, number of days a nest remained active, nest height, and bush species in which nests were placed. Both models used GLMs and both have samples sizes: n = 105. Model Exterior df AICc ΔAICc Weight Log likelihood Proximity + days active 4 451.92 0 0.38 -­‐221.76 Proximity + days active + first egg date 5 452.51 0.59 0.28 -­‐220.95 Proximity + first egg date 4 453.52 1.61 0.17 -­‐222.56 Proximity + days active + nest height 5 453.64 1.73 0.16 -­‐221.52 Proximity + days active + first egg date 5 248.25 0 0.55 -­‐118.82 Proximity + days active + ambient temperature 5 249.99 1.73 0.23 -­‐119.69 Proximity + days active + first egg date + nest height 6 250.04 1.79 0.22 -­‐118.59 Model Interior Table S5B. Parameter estimates averaged from all best performing models (ΔAICc < 2) in Table S5A, for factors influencing the proportion of Eriocephalus fluff on the exterior and interior of Karoo prinia nests. P-­‐adj corrects for false discovery rates following Benjamini et al. 2006, because we measured four different, but related, components of Eriocephalus fluff in prinia nests. Sample sizes: n = 105 nests. Parameters Exterior z p p-­‐adj Proximity -­‐1.30 0.16 7.92 <0.0001 <0.0001 Days active 13.91 9.90 1.40 0.16 0.36 First egg date -­‐2.16 3.29 0.65 0.51 0.62 0.17 0.74 0.23 0.82 0.62 Nest height Parameters Interior Proximity -­‐0.40 0.06 6.46 <0.0001 <0.0001 Days active 14.40 3.20 4.44 <0.0001 <0.0001 First egg date Estimate SE 2.07 1.59 1.30 0.19 0.39 126 Ambient temperature 0.03 0.07 0.45 0.65 0.62 Nest height 0.09 0.33 0.27 0.79 0.62 127 Appendix B: Supplementary material for Chapter 4 Measuring the amount and dissipation rates of volatile compounds in Eriocephalus essential oils New Directions Aromatics Inc.’s Eriocephalus africanus essential oils are produced by steam distillation of floral heads (New Directions Aromatics Inc.). Eriocephalus plants flower en mass covering plants in clusters of white blossoms. After flowering, floral heads produce a pale-­‐colored, cotton-­‐like material, which many species of birds use in nest construction, often with the associated floral-­‐head (Dean et al. 1990). While the part of the plant that birds use to construct their nests (seed material) is not the same part of the plant used by New Directions Aromatics Inc. to produce essential oils (floral heads), both plant tissues are highly aromatic (VGR pers. obs.). It is also possible that the seed material used by birds is more chemically defended than the floral heads, as plant fitness may be most affected by the loss of viable seeds. New Direction Aromatics Inc.’s E. africanus essential oil contains only compounds extracted from E. africanus plants and is composed of four dominant compounds: camphor, sabinene, p-­‐cymene, and eucalyptol (New Directions Aromatics Inc.). While these compounds represent a small number of the total compounds found in Eriocephalus plants (Njenga 2005), they have ovicidal, larvicidal, and repellent properties to house flies (Musca domestica) and other insects (Rice and Coats 1994, Regnault-­‐Rogers 1997, Bisseleua et al. 2008, Kumar et 128 al. 2014) when applied directly to the surface of parasites or indirectly as a fumigant in enclosed flasks. We estimated amounts of the four dominant compounds (camphor, sabinene, eucalyptol, p-­‐cymene) in E. africanus essential oil by creating separate solutions of known concentration for each compound. We generated calibration curves by injecting different quantities of each compound into a gas chromatograph (GC), which created a linear relationship between the volume of solution injected and the amount of each compound in the solution; all four compound calibration curves had r2 > 0.97. Using these calibration curves, we estimated quantities of the four dominant compounds in E. africanus essential oil by injecting known amounts of essential oil and comparing the quantity of compounds measured in the oil to our calibration curves using a linear regression equation: x = (y-­‐b)/m where y is the amount of a compound measured in the essential oil, m is the slope of the compound’s calibration curve, and b is the intercept from the compound’s calibration curve. Specifically, we mixed 4.3mg of camphor into 20mL of isopropyl alcohol, and 2.8mg of sabinene hydrate, and 0.0020mL of eucalyptol and p-­‐cymene into 10mL of isopropyl alcohol to achieve concentrations (g/mL) of 1.84e-­‐4 (eucalyptol), 2.15e-­‐4 (camphor), 2.80e-­‐4 (sabinene hydrate), and 1.72e-­‐4 (p-­‐cymene). We injected sequentially larger amounts of these solutions (starting with 0.2μL, ending with 1.2μL, and increasing amounts by 0.2μL) into the GC to generate our calibration 129 curves. Sabinene was not available from Sigma-­‐Aldrich Co. (St. Louis, MO, USA) when we ordered compounds so we used sabinene-­‐hydrate. For E. africanus essential oils, we added 10μL of essential oil into 1000μL of isopropyl alcohol and injected 1μL of diluted essential oil through the GC. We then used our calibration curves to estimate the amount of each compound in E. africanus essential oil. We used a Varian CP-­‐3900 gas chromatograph with autosampler, flame ionization detector (FID), and a 15m (0.25mm internal diameter) capillary column. We used helium as the carrier gas at a flow rate of 4mL/min. We optimized chromatographic separation of compounds using a ramped temperature program that started at 40°C and increased temperature at a rate of 3°C/min to 75°C, then increased at a rate of 80°C/min to 280°C, and finally increased at a rate of 100°C/min to 325°C, for a total run time of 17 minutes. Injector and FID were held at 250°C and 300°C, respectively. We estimated dissipation rates of the four dominant compounds in E. africanus essential oil by adding 0.023mL of diluted essential oil (1:1, E. africanus essential oil:mineral oil) to five cotton balls placed in a windless room at 23°C. Cotton balls sat for time intervals of 0, 6, 12, 24, and 48 hours, before they were placed into 10mL of isopropyl alcohol to measure amounts of chemical compounds remaining at different time intervals. All samples were soaked in isopropyl for 5 days prior to analysis. We injected 0.5μL of extract into the GC and ran each sample with the same temperature and time profiles as described above. We generated new calibration 130 curves for each compound to control for possible variation caused by GC settings, and all calibration curves had r2 > 0.97. Measuring compounds in Eriocephalus plant material We assessed the chemical profiles of seven Eriocephalus plants using 2 extraction methods. For our first extraction method, we added 0.8g of seed material from a single E. africanus plant into 100mL isopropyl alcohol, covered with a paraffin seal, and placed into a warm, ultra-­‐sonic water bath for 30 minutes. For our second extraction method, we added 1.0g of E. racemosus seed material into 20mL of isopropyl alcohol for 10 days, after which, 10mL were transferred into a glass vial until GC analysis of chemical composition. We assessed chemical profiles of each plant sample by injecting 1μL of extract into the GC. GC settings of temperature and time for each run were the same as those used for essential oil analysis above. We used the same standard solutions as described above but generated new calibration curves for each compound to control for possible variation in GC temperatures and amounts of each compound detected during each analysis, all calibration curves had r2 > 0.98. Parasite analysis For transformations of first egg date, we used a Johnson Su transformation in JMP (SAS 2007) using the following formula: 𝐴𝑟𝑐𝑆𝑖𝑛𝐻
first egg date − 131.84
∗ 0.84 − 0.82 2.87
131 Measuring corticosterone (cort) in adults and nestlings We measured adult female cort levels twice during the breeding period: first on day 3±1 of incubation, second on day 13±1 of the nestling period. We caught all females using door traps that closed when females entered the box, or by covering the box-­‐
entrance with our hand while females were inside; capture method did not influence cort measures (Wilcoxon rank-­‐sum test, w = 101, p > 0.1; hand n = 8, door trap n = 18). We measured nestling cort when nestlings were 13 days old. We collected all blood samples in under three minutes from the time trap doors closed or our hands covered the nest entrance for adult females, or from the time we opened the box to extract nestlings. Care was also taken to be quiet when approaching nest boxes to further ensure that hormone measures from these samples provided estimates of baseline circulating hormone levels, rather than a response to the stress of capture (Romero and Reed 2005). For all cort measures, we punctured the brachial vein and collected small blood samples into micro-­‐capillary tubes then immediately placed samples on ice. We centrifuged blood samples to separate plasma from red blood cells, and kept plasma and red blood cells frozen at -­‐20°C until hormone assays. We used direct radio-­‐immunoassay (RIA) to measures cort in blood plasma (See: Wingfield et al. (1992) for details), and ran all samples in duplicate in a single assay. We estimated within assay variation to be 12.1 per cent using five standards of known concentration. We tested for differences in adult female and nestling cort using linear mixed effect models (Pinheiro et al. 2014) in R (R Core Team 2014), with nesting location 132 (nest grid) as a grouping variable to control for variation caused by nesting site. For adult females, we used two models with slightly different variables. Our first model used adult female cort taken on day 13 of the nestling period as our response variable and treatment, number of nestlings, parasite mass, and female mass (measured on day 13, the same day the cort sample was taken) as our predictor variables. Our second model used change in adult female cort measured as [(ad female cort from day 13 of the nestling period)-­‐(ad female cort from day 3 of incubation)] as our response variable, and treatment, number of nestlings, parasite mass, and change in adult female mass [(adult female mass measured on day 13 of the nestling period)-­‐(adult female mass measured on day 3 of incubation)], as predictor variables. Including measures of female mass could be problematic if increased parasite loads lead to greater food demand on females, causing both a reduction in female mass and a change in female cort. We thus ran all cort analyses with and without measures of female mass. We did not include interaction terms because of small sample sizes. We transformed parasite mass [1/(parasite mass+2)] and change in female cort [((late sample-­‐early sample)+10)1.5] to fit normal distributions. For analyses of nestlings, nestling cort was our response variable and treatment, average nestling mass within broods (measured as: brood mass/number of nestlings at time of cort sample), number of nestlings, and parasite mass were our predictor variables. We excluded 1 nestling because cort levels were more than four standard deviations higher than the mean when the outlier was included (outlier 133 cort measure = 159.7 ng/mL; mean nestling cort with outlier = 19.8 ± 39.0 sd (n = 15); mean nestling cort without outlier = 9.8 ± 5.2 sd (n = 14)). For adult female and nestling cort analyses, we checked the assumptions of our full models prior to model selection by plotting model residuals on predictor variables, testing for differences in residual variance between treatments using Bartlett’s tests, and testing if model residuals deviated from normality using Shapiro-­‐Wilk tests. We used AICc values to select the best performing models (ΔAICc < 2) using the dredge command in the R package MuMIn (Bartoń 2015), and checked the assumptions of our best performing models using the same procedure outlined for our full models. Measuring nestling growth rates We measured nestling growth rates by weighing each nestling on days 4 and 12; we cut one toenail on each nestling on day 4 to identify the same individuals on day 12 allowing us to calculate growth rates for individual nestlings. We used total change in mass for each nestling measured as [(day 12 mass)-­‐(day 4 mass)] for our measure of nestling growth rate and grouped nestlings by nest box. We tested for differences in nestling growth rate between nests with and without Eriocephalus compounds using linear mixed-­‐effects models in the nlme package (Pinheiro et al. 2014) in R (R Core Team 2014) with nest box as a grouping variable to account for the non-­‐independence of multiple nestlings originating from the same box. Nestling growth rate was our response variable and a treatment*parasite mass interaction term, and average maximum temperature (average maximum temperature between nestling days 4 and 12) were our 134 predictor variables; average maximum temperature was transformed using a Johnson SU transformation in JMP (SAS 2007) using the following formula: 𝐴𝑟𝑐𝑆𝑖𝑛𝐻
average maximum temperature − 23.05
∗ 0.54 − 0.15 0.28
We tested the assumptions of our full model prior to model selection by plotting standardized residuals on fitted values, testing for differences in residual variance between treatments using Bartlett’s tests, and testing if model residuals deviated from normality using Shapiro-­‐Wilk tests. We used AICc values to select the best performing models (ΔAICc < 2) using the dredge command in the R package MuMIn (Bartoń 2015). We checked assumptions of the best performing models using the same procedure outlined for our full model. Results and Discussion Dissipation rates of compounds in Eriocephalus essential oil Compounds in E. africanus essential oils are highly volatile and dissipated rapidly (Appendix B; Figure S1). After 12 hours from the initial application of 0.023mL of essential oil mixture (1:1, E. africanus essential oil:mineral oil), compounds were undetectable using the GC (Appendix B; Figure S1). Compounds in Eriocephalus oils and plants The amount of Eriocephalus compounds in bird nests depends on both the amount of Eriocephalus material used in nest construction and the amount of secondary 135 compounds in the specific plants from which the birds gathered material. Eriocephalus plants show individual-­‐, population-­‐, and species-­‐level variation in the amount of compounds in plants (Njenga 2005). For example, some compounds will dominate (compose >40 per cent) the chemical composition of plants in some populations but be entirely absent from plants of the same species in other populations (Njenga 2005). Our application of 0.031mL of E. africanus essential oil resulted in the addition of ~21.5mg ± 1.56(SE) (n = 3 measures of essential oil) of the four dominant compounds per application (camphor, eucalyptol, sabinene, p-­‐
cymene). Levels of these same four compounds ranged from 1.42–5.61mg/2g of seed material in the 7 plants we examined (Appendix B; Figure S2); two grams of Eriocephalus seed material is an amount common in nests of some small passerine birds (e.g., Karoo prinia Prinia maculosa, grey-­‐backed cisticola Cisticola subruficapilla, VGR unpublished data). The amounts of these compounds we measured in plants are at least 3.8 times lower than those we added to tree swallow nests. However, with high dissipation rates and a three-­‐day time interval between applications, adding slightly more compounds likely helped slow their complete dissipation from nests. Additionally, other species of birds (e.g., fiscal shrike, Lanius collaris, Cape sparrow, Passer melanurus) and species of nest-­‐building rodents (four striped grass mouse Rhabdomys pumilio, Southern African vlei rat Otomys irroratus) that use Eriocephalus in nest construction likely use more than two grams of Eriocephalus seed material (VGR pers. obs.). Thus our oil application to tree swallow nests likely resulted in chemical profiles that peaked immediately after application 136 to levels above those found in nature, then rapidly dissipated to zero or near zero levels. Corticosterone (cort) levels in adult females and nestlings No best performing models identified treatment as an important predictor of adult female cort when using female cort measured on day 13 of nestling period or when using within individual change in cort (Appendix B; Table S1) (for adult female cort measured during day 13 of the nestling stage, the null model was the only best performing model so we do not present summary tables for this measure of cort). Parasite mass was a significant predictor of change in female cort, with increasing parasite mass associated with more negative changes in adult female cort (i.e., change in female cort was calculated as [(ad female cort from day 13 of the nestling period)-­‐(ad female cort from day 3 of incubation)], thus a negative change in cort results from individuals having lower cort levels later in the breeding cycle relative to their cort levels earlier in the breeding cycle) (Appendix B; Table S1). While the mechanism responsible for this association remains unknown, one possibility is that parasites may change how nestlings interact with adults (e.g., begging behavior, attentiveness), which may influence changes in adult female cort. No best performing models included measures of female mass, and analyses of female cort with and without mass returned the same best performing models. For nestlings, no best performing models identified treatment as an important predictor of nestling cort (Appendix B; Table S2). While we detected no differences in adult female and nestling cort levels, our samples do not include 137 females that lost their broods or nestlings that did not survive until our sampling date. Nestling growth rates AICc comparison identified a single best performing model for nestling growth rate, so we do not present summary tables of best performing models (all other models had ΔAICc > 2 relative to the best performing model). Average maximum temperature was the only predictor of nestling growth rate (average maximum temperature estimate: 0.68, z = 2.31, p = 0.021) with nestlings growing more quickly with warmer temperatures (sample sizes are: control n = 27 nestlings from n = 7 boxes; experimental n = 39 nestlings from n = 11 boxes). Exclusion of first larval instars Two lines of evidence suggest that first larval instars in nests surviving to nestling day 15 originated from eggs laid once treatments finished, just prior to fledging. First, Eriocephalus compounds dissipated rapidly (Appendix B; Figure S1); thus, once oil applications ended, chemicals in the nests likely dropped to zero or near zero levels, eliminating the effect of our treatment. Nestlings remained in the nest on average 20.8 ± 1.8 days (sd), nearly six additional days after oil applications finished; these six additional days are double the three-­‐day interval between treatment applications. Second, comparing numbers of first larval instars between nests that survived to nestling day 15 and those that did not (i.e., nests where first larval instars did not overlap with our treatment versus nests where first larval instars overlapped completely with our treatment) suggests that the effect of 138 Eriocephalus compounds disappeared shortly after oil applications finished (Appendix B; Figure S3). Median numbers of first larval instars increased in nests that previously received Eriocephalus compounds suspended in mineral oil (exp), but not in nests that previously received mineral oil only (con) (Appendix B; Figure S3). These two lines of evidence suggest that upon stopping applications of Eriocephalus compounds, adult P. sialia found nests, laid eggs, and numbers of first larval instars from newly laid eggs increased across treatments. Additionally, two of five nests that received Eriocephalus chemicals and survived until nestling day 15 had only first larval instars of P. sialia larvae (no advanced larval stages), suggesting that they were parasitized once treatments ended; no control nests that survived to nestling day 15 had only first larval instar P. sialia. Parasite analyses with and without 1st larval instars Re-­‐running our parasite analyses with first larval instars from nests that survived to nestling day 15 shows that nests receiving Eriocephalus compounds suspended in mineral oil had reduced parasite mass, but no statistically significant differences in the number of ectoparasites (Appendix B; Tables S3, S4, Figure S4) compared to nests receiving only mineral oil. The overall increase in parasite numbers in both control and experimental nests, when first larval instars of nests surviving to nestling day 15 are included (Appendix B; Figure S4), may be caused by increased blowfly numbers as the season progresses because: (i) flies are more active as ambient temperatures warm, (ii) multiple generations of blowflies are searching for active bird nests, and (iii) numbers of active nests may be lower during the end of 139 the tree swallow breeding season (Bennett and Whitworth 1991, Rogers et al. 1991). Rapid dissipation rates of Eriocephalus compounds and the increased ectoparasite infestation upon stopping our treatments in nests that survived to nestling day 15, provide evidence that the loss of chemical compounds increases the risk of nest ectoparasite infestation. These observations suggest that selection should favor bringing fresh aromatic materials to the nests in order to maintain levels of secondary compounds throughout the breeding cycle (Petit et al. 2002). However, replenishing aromatic material during the nestling stage may be challenging as time spent searching for, gathering, and replenishing aromatic material may be traded-­‐off against provisioning hungry nestlings. This trade-­‐off should be especially pronounced with large nestlings because their demands on attending parents should be highest. 140 amo
in 0.
0
euc
cam
sab
cym
0.0
(mg)
ial
2
amount of compounds (mg)
0.8
eucalyptol
p-cymene
sabinene
camphor
0.4
0.0
0
6
12
24
time (hours)
48
amount of p-cymene (mg) in
0.031 mL of essential oil
1
0.0
oil
fluff
0.0
amount of p-cymene (mg)
per (g) plant material
1.5
1.5
Figure S1. Dissipation rates of the four dominant compounds in E. africanus essential oils. Amount of compounds represent milligrams of each compound measured in a 0.023 mL mixture (1:1 essential oil:mineral oil) added to cotton balls then placed at time intervals 0, 6, 12, 24, and 48 hours after 1.0
1.0 in isopropyl alcohol 7
12 dissipated initial oil application. C
ompounds r
apidly s
uch t
hat compounds fell to 3
levels below detection for our gas chromatograph 12 hours after initial application. 0.5
0.5
seed Africanus
141 compounds (mg/2g plant material)
6
4
2
0
E. africanus
E. racemosus
Figure S2. The summed amount of four chemical compounds (eucalyptol, camphor, p-­‐cymene, and sabinene) measured in one E. africanus (left) and six E. racemosus plants. 142 all nests
complete overlap treatment stopped
of treatment and
nestlings
parasites
>15 days old
parasite numbers
first larval instars only
400
n=4
300
n=8
200
n=5
n=9
100
n=5
n=3
0
con
100
con
exp
con
exp
nd
3rdinstar instar P. sialia between 1st instar
instar of first larval puparianests that did Figure S3. Contrasting n2umbers not survive to nestling day 15 and parasites overlapped completely with oil treatments (middle panel) and nests that survived to nestling day 15 and treatments ended (right panel). Experimental nests (exp) received Eriocephalus chemicals suspended in mineral oil and control nests (con) received mineral oil without compounds. Boxplots represent medians (thick black line), 25th and 75th percentiles n = (6points that exceed (box), 1.5 times the interquartile range (whiskers), and outliers 1.5 time the interquartile range). n = 5
50
n=3
n=8
n=7
n=8
n=5
n=1
0
con
exp
con
exp
con
exp
con
exp
1
proportion of nests
with ectoparasites
parasite numbers
150
exp
143 1st instars
all
parasites excluded
all
1st instars
parasites excluded
parasite mass (g)
2.5
1.5
0.5
parasite numbers
120
80
40
0
Figure S4. Average (±1SE) mass and number of P. sialia larvae in nests receiving mineral oil with Eriocephalus compounds (experimental, gray, n = 15) and nests receiving mineral oil without compounds (control, black, n = 16), plotted with and without first larval instars in nests that survived to nestling day 15 and oil applications ended. 144 n=3
0
0
Con
Exp
with without
with without
3rd instar
puparia
1st instar
2nd instar
parasite numbers
150
100
n=6
n=5
50
n=3
n=8
n=7
n=8
n=5
n=1
0
con
exp
con
exp
con
exp
con
exp
proportion of nests
with ectoparasites
Figure S5. Numbers of blowfly larvae for different developmental stages in nests 1
receiving Eriocephalus chemicals suspended in mineral oil (exp) and nests receiving mineral oil without compounds (con); 1st instar panel includes only nests where first larval instars overlapped completely with our treatment (i.e., nests that did not survive until nestling day 15). Boxplots represent medians (thick black line), 25th and 75th percentiles (box), 1.5 times the interquartile range (whiskers), and outliers (points that exceed 1.5 time the interquartile range). 0
n = 16
n = 15
con
exp
145 exp
con
exp
con
exp
1
proportion of nests
with ectoparasites
n
n=1
0
n = 16
n = 15
con
exp
Figure S6. The proportion of nests that contained blowfly larvae for nests receiving Eriocephalus chemicals suspended in mineral oil (exp) and nests receiving mineral oil without compounds (con). Two nests receiving Eriocephalus compounds (exp) were infested with only first larval instars after nestling day 15 when our oil additions ended but were categorized here as infested. 146 number of young
number of nests
5
a
10
5
b
n=4
n = 10
0
0
unhatched
eggs
depredated
nestlings dead
in nest
fledged
young
con
exp
Figure S7. (a) Fate of nests receiving Eriocephalus compounds suspended in mineral oil (experimental, gray, n = 19), and nests receiving mineral oil without compounds (control, white, n = 17). Unhatched eggs were entire clutches that failed to hatch; depredated nests were entire clutches or nestlings that went missing prior to fledgling; nestlings dead in nests were entire broods found dead in the nest — in these cases, the precise cause of failure (e.g., disease, starvation, abandonment by adults) was unknown; fledged nests were nests that fledged at least one young. (b) Number of young by which brood size was reduced by the time of fledging [calculated as: (clutch size)-­‐(number of young fledged)] for control and experimental nests; numbers in box plots are sample sizes of nests that fledged at least one young and from which partial brood loss was calculated. Box plots represent medians (thick black line), 25th and 75th percentiles (box), and 1.5 times the interquartile range (whiskers). 147 Table S1A. Best performing models (ΔAICc < 2) testing for within individual change in adult female cort [measured as: (ad female cort from day 13 of the nestling period)-­‐(ad female cort from day 3 of incubation)] as a function of treatment (mineral oil with Eriocephalus compounds (experimental) versus mineral oil without compounds (control)), number of nestlings, parasite mass, and change in adult female mass [measured as: (ad female mass from day 13 of the nestling period)-­‐(ad female mass from day 3 of incubation)]. Treatment was not an important predictor in any best performing model, and analyses with and without female mass returned the same best performing models. Sample sizes are: control n = 5; experimental n = 8. model df log likelihood AICc ΔAICc weight parasite mass 4 -­‐52.06 117.13 0.00 0.68 null 3 -­‐55 118.68 1.55 0.32 Table S1B. Parameter estimates averaged from best performing models in Table S1A for predictor variables influencing within individual change in adult female cort. Sample sizes are: control n = 5, experimental n = 8. parameter estimate SE z p parasite mass -­‐77.27 30.8 2.17 0.030 148 Table S2A. Best performing models (ΔAICc < 2) testing for differences in nestling cort as a function of treatment (mineral oil with Eriocephalus compounds (experimental) versus mineral oil without compounds (control)), nestling mass, number of nestlings in the nest box, and parasite mass. Treatment was not included in any best performing models. Sample sizes are: control nests n = 5; experimental nests n = 9. model df log likelihood AICc ΔAICc weight average nestling mass 4 -­‐39.91 92.25 0.00 0.46 null 3 -­‐42.3 92.99 0.74 0.32 parasite mass 4 -­‐40.62 93.69 1.44 0.22 Table S2B. Parameter estimates averaged from best performing models in Table S2A for predictor variables influencing levels of nestling cort. Sample sizes are: control nests n = 5; experimental nests n = 9. parameter estimate SE z p average nestling mass -­‐0.76 0.34 1.95 0.052 parasite mass 3.59 1.99 1.58 0.113 149 Table S3A. Best performing models (ΔAICc < 2) testing for differences in total parasite mass as a function of treatment (mineral oil with Eriocephalus compounds (experimental) versus mineral oil without compounds (control)), first egg date, and exposure days (number of days nest received treatment). Table shows best performing models for analyses that included and excluded first instar P. sialia larvae from nests that survived to nestling day 15 when treatments ended. Sample sizes are: control nests n = 16; experimental nests n = 15. model df log likelihood AICc ΔAICc weight included treatment + exposure 5 -­‐31.73 75.85 0.00 0.65 treatment + exposure + first egg 6 -­‐30.8 77.1 1.25 0.35 date excluded treatment + exposure 5 -­‐32.41 77.23 0.00 0.53 treatment + exposure + first egg 6 -­‐31.62 78.74 1.52 0.25 date treatment 4 -­‐34.73 78.99 1.77 0.22 Table S3B. Parameter estimates averaged from best performing models in Table S3A for predictor variables influencing total parasite mass in analyses that included and excluded first instar P. sialia larvae from nest surviving to nestling day 15 when treatments ended. Treatment estimates are given for experimental relative to control nests. Sample sizes are: control nests n = 16; experimental nests n = 15. parameter estimate SE z p included exposure 0.06 0.03 2.15 0.032 treatment -­‐0.59 0.26 2.19 0.029 first egg date -­‐0.16 0.12 1.23 0.220 excluded exposure 0.06 0.03 2.06 0.040 treatment -­‐0.58 0.27 2.04 0.041 first egg date -­‐0.15 0.13 1.13 0.257 150 Table S4A. Best performing models (ΔAICc < 2) testing for differences in parasite numbers as a function of treatment (mineral oil with Eriocephalus compounds (experimental) versus mineral oil without compounds (control)), first egg date, and exposure days (number of days nest received treatment). Table shows best performing models for analyses that included and excluded first instar P. sialia larvae from nests that survived to nestling day 15 when treatments ended. Sample sizes are: control nests n = 16; experimental nests n = 15. model df log likelihood AICc ΔAICc weight included null 3 -­‐92.56 192.01 0.00 0.66 treatment 4 -­‐91.88 193.31 1.3 0.34 excluded treatment 4 -­‐79.63 168.8 0.00 0.71 treatment + first egg date 5 -­‐79.11 170.63 1.82 0.29 Table S4B. Parameter estimates averaged from best performing models in Table S4A for predictor variables influencing numbers of parasites in analyses that included and excluded first instar P. sialia larvae from nest surviving to nestling day 15, when treatments ended. Treatment estimates are given for experimental relative to control nests. Sample sizes are: control nests n = 16; experimental nests n = 15. model estimate SE z p included treatment -­‐1.98 1.74 1.09 0.277 excluded treatment -­‐2.98 1.13 2.52 0.012 first egg date -­‐0.55 0.56 0.93 0.352 151