Social and Ecological Correlates of Avian Infection by Haemosporidian Blood Parasites

University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

Dissertations & Theses in Natural Resources Natural Resources, School of

8-2021

Ian R. Hoppe University of Nebraska - Lincoln, [email protected]

Follow this and additional works at: https://digitalcommons.unl.edu/natresdiss

Part of the Behavior and Ethology Commons, Natural Resources and Conservation Commons, and the Natural Resources Management and Policy Commons

Hoppe, Ian R., "Social and Ecological Correlates of Avian Infection by Haemosporidian Blood Parasites" (2021). Dissertations & Theses in Natural Resources. 334. https://digitalcommons.unl.edu/natresdiss/334

This Article is brought to you for free and open access by the Natural Resources, School of at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Dissertations & Theses in Natural Resources by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln.

SOCIAL AND ECOLOGICAL CORRELATES OF AVIAN INFECTION

BY HAEMOSPORIDIAN BLOOD PARASITES

Ian R. Hoppe

A THESIS

Presented to the Faculty of

The Graduate College at the University of Nebraska

In Partial Fulfillment of Requirements

For the Degree of Master of Science

Major: Natural Resource Sciences

Under the Supervision of Professor Elizabeth VanWormer

Lincoln, Nebraska

August, 2021 SOCIAL AND ECOLOGICAL CORRELATES OF AVIAN INFECTION

BY HAEMOSPORIDIAN BLOOD PARASITES

Ian R. Hoppe, M.S.

University of Nebraska, 2021

Advisor: Elizabeth VanWormer

Haemosporidian parasites are a significant source of morbidity and mortality for birds. There is growing recognition of the negative consequences of haemosporidian infections for wild birds at individual and population levels. Avian haemosporidians are geographically widespread, have been detected from a phylogenetically diverse array of hosts, and have been the focus of extensive research due to their impacts on birds and their similarity to vector-borne diseases of humans. However, factors influencing haemosporidian transmission, especially transmission between species, are poorly understood. To better understand these influences, we compared prevalence and diversity of haemosporidian blood parasite infections among species in a behaviorally and ecologically diverse host assemblage. We studied whether interspecific associations could explain community-wide trends in infection by pairing molecular diagnostics with direct observations of species interactions. Haemosporidian prevalence in the community was low (8.6%), but varied substantially with host phylogeny. Most (94.8% of all infections) infections were identified as Haemoproteus spp. Few Plasmodium spp. infections were detected, and no Leucocytozoon spp. infections were found. We found no evidence for an effect of interspecific sociality on Haemoproteus infections, but we did find evidence for an effect of intraspecific sociality. Individuals of species that had smaller average conspecific group sizes were more likely to be infected than those of species with larger groups. We also found that species with relatively long lifespans (as an index of immune investment) had higher prevalences than species with shorter lifespans. No other individual- or species-level traits were associated with Haemoproteus infection. We identified 7 Haemoproteus mitochondrial cytochrome b lineages, which clustered at the host family level. Two Plasmodium lineages were also identified, each of which had been previously detected in different host species in the region. The apparent host-family specificity of the parasite lineages may partly explain the lack of effect of interspecific sociality on Haemoproteus infection probability, and implies the presence of barriers to transmission that are associated with host phylogeny.

Keywords: birds; Haemosporida; social ecology; South Australia; species interactions

DEDICATION

For Mary

ACKNOWLEDGMENTS

Australia is a land like no other. It is populated with fantastic creatures that supply endless fascination to anyone who takes the time watch them. I am incredibly privileged to have had the opportunity to learn something about the natural world by observing the inhabitants of that strange and beautiful country. I also recognize that the Australia of today is but a shadow of the wild and untamed place of former years. That wild places and things still exist is a testament both to recent efforts by government agencies, conservation organizations, and intrepid individuals, and to the caretakers of ages past. I acknowledge the Ngaiawang people, Traditional Custodians of the land on which this study was undertaken, and pay my respects to their Elders. It is my sincere hope that the values they placed on stewardship of the land, and on appreciating connections to land, life, and community, will be maintained, both in Australia and abroad, so that no future generation will know a world without wild places. This body of work could not have been completed without the labors of many. The core ideas behind the thesis took root where all good ideas begin—in the field, during conversations about birds and blood parasites with Allison Johnson. They began to adopt a more definite shape with the help of my advisor, Liz VanWormer. My deepest thanks to both of you for that early guidance—to Liz for kindling my interest in disease ecology; and to Allison for introducing me to the birds at Brookfield, then welcoming me back time and time again. Throughout the growth of this project, the members of my committee—Liz and Allison, Clay Cressler, and Dai Shizuka—have supplied invaluable intellectual and technical support and guidance. I extend my gratitude to each of you for placing your confidence in me when I could not, for offering from the wealth of your expertise when I needed it, and for contributing your thoughts and ideas to the improvement of this manuscript. To Laura Vander Meiden, who dedicated untold hours to the pursuit of small and swiftly-moving targets through shimmering canopies—I offer my thanks and humble respect. Aside from her critical support in the field, Laura showed me the ropes about networks and helped me to think through and develop an approach to a particular network analysis problem. Most of all, thank you for being a friend. Malcolm Conner, Britt Dryer, and Erik Johansson performed countless hours of untiring work in the field setting, moving, and mending nets; banding birds; and collecting samples and behavioral observations. Without their talent and hard work, this thesis could not stand. Without their good humor and excellent company, the field season would have passed in tedium. Many thanks are also owed to past field crews at Brookfield, whose years of work supply valuable context and background to this study. Special thanks to Dylan Meyer, for helping to make my own first visit to Brookfield a marvelous adventure. Chad Brassil, Larkin Powell, and Drew Tyre provided many helpful comments on the statistical analyses presented in this thesis. Their guidance helped to clarify my thinking and steer me out of trouble. I’m also grateful to Brian Dillard for sharing his sequence analysis expertise when I was unsure of myself, to Justin Eastwood for providing advice for troubleshooting the PCR diagnostic, and to Holly Lutz for supplying the initial positive controls when I had none.

Throughout my work on this project, I’ve had the good fortune of having not one but two homes on campus. Members of both the One Health research group and the “Dezukabets” lab cluster have provided encouragement, ideas, and friendship. Special thanks to Kleidy Camela, Dominic Cristiano, Kyle Dougherty, Maria Goller, Faiza Hafeez, Louise Lynch-O’Brien, Annie Madsen, Ben Ndayambaje, David Nguyen, Anisha Pokharel, Marnee Roundtree, Miranda Salsbery, Abbey Snyder, Stella Uiterwaal, and Brianne Wolf for their friendship, ideas, and encouragement. Coordinating the logistics of field and lab work proved a much more demanding task than I anticipated. The University’s administrative teams were a tremendous asset in helping to ensure that everything was accomplished smoothly and in a timely manner. I’m especially grateful to Terri Eastin, Karen Gilbert, Lisa Greif, Patty Swanson, and Pat Wemhoff for their patience with my many enquiries and requests. Thanks also to Ellen Marsh and Anthony Mazza for helping to set up and facilitate sequencing activities. Much of the field and lab equipment needed for the study was purchased using funds from a crowd-sourced grant competition coordinated by the Wildlife Disease Association in cooperation with Experiment. I extend my humble gratitude to the many friends, family members, associates, and strangers who were willing to support this study financially. I also appreciate the work of Nicole Sharpe and Cindy Wu at Experiment, who organized and facilitated the competition. Nicole also offered excellent insight and advice for improving my original proposal. Most spring afternoons in Brookfield are incredibly bright, but visits to the field site from beloved friends never failed to make the day brighter. Over the years, while I was developing a love for the sun-scoured mallee and scrub, I became aware of a community of others who share this love for Brookfield and the wild things that live there. In my mind, as in my heart, these people are inextricably linked with the place. With great fondness and the sting of separation, I thank Tricia Curtis, Claire Gaughwin, Matt and Kristina Gaughwin, Terry Peacock and Karen Davies, and Joe Stelmann and Stephanie Williams for their friendship, and for continuing to champion Brookfield and all of South Australia’s wild places. This thesis answers some questions, but it also raises new ones. It is not a laying to rest of the issue of social behavior and disease, but merely a different take on an old question with many answers. It is neither an end nor a beginning; but rather, it represents one step towards a greater understanding. So too is this project for me neither the end of the path nor its start. In the years ahead I will continue my journey of discovery and the pursuit of understanding. But the journey I’m on is one I began long before I set to work on the study that has culminated in the preparation of the document now before you. Along the way, I benefitted tremendously from many wise and caring mentors and friends who helped me to find my own way. I am especially indebted to Mary Bomberger Brown, Larkin Powell, and Stephen Pruett-Jones. Each played a pivotal role in the development of my sense of self as a scientist and scholar. Each has had a lasting impact on the trajectory my life has taken since our meeting. And each will continue to be an important part of my life and journey. Thank you. My love for and curiosity about the natural world has taken me to exotic and faraway lands, but they began in the commonest of places: a backyard abutting extensive fields and fencerows. They began, moreover, with the freedom to explore. I am tremendously grateful for the great privilege of growing up where I did, and to my family

vii for inspiring and encouraging my curiosity and wanderlust. My parents, Weldon and Paula Hoppe, have not ceased supporting and believing in me from the beginning. Thanks to each of you for helping to make this project possible, and for your work in improving it. Any errors within are purely my own. Finally, to the birds of Brookfield: watching your lives unfold has given me immeasurable joy. When I stop to think that even now you are there, busy about your daily dramas—that even in the midst of my absence Brookfield exists, and you in it, forming flocks and building nests, experiencing freedoms that I can never know; and when I think that you have been thus engaged since time immemorial, and will be so (hopefully) long after I am gone—I’m reminded of the enormity of the world, and of my smallness within it. Thank you for tolerating our many intrusions into your lives, and for being irresistibly intriguing.

viii

CONTENTS

Abstract ...... ii Dedication ...... iv Acknowledgments ...... v Table of contents ...... viii List of tables...... ix List of figures ...... x

Introduction ...... 1 Haemosporidian parasites and their avian hosts ...... 3 Connecting avian ecology and blood parasite transmission ...... 8 History and geographic context of haemosporidian infection ecology research ...... 18 Research motivation and objectives ...... 20

Chapter 1: Relating social ecology to blood parasite infections in a diverse avifauna Introduction ...... 22 Materials and methods ...... 27 Results ...... 36 Discussion ...... 44

Chapter 2: Blood parasite diversity in an under-surveyed avifauna Introduction ...... 49 Materials and methods ...... 52 Results ...... 54 Discussion ...... 59

Conclusion ...... 63

References ...... 67

Appendix A: List of sampled host species...... 86 Appendix B: Banding data sheet ...... 88 Appendix C: Banding reference guide ...... 89 Appendix D: Molecular assays ...... 91 Appendix E: Priors, convergence, and posterior predictive checks ...... 96 Appendix F: Lineage-specific prevalence analyses...... 108 Appendix G: Biology of haemosporidian parasites ...... 115 Appendix H: Glossary ...... 127

TABLES

Table 2.1 Haemosporidian lineages identified in birds sampled at Brookfield Conservation Park, Australia, and their host associations...... 55

Table A.1 Distribution of samples and haemosporidian infections among species sampled at Brookfield Conservation Park, as well as species-level social and ecological attributes ...... 86

Table D.1 Primers used for molecular sexing of host species ...... 91

Table D.2 Polymerase chain reaction cycling conditions for molecular sexing of host species ...... 92

Table D.3 Primers used for molecular diagnosis of haemosporidian infections ...... 92

Table D.4 Polymerase chain reaction cycling conditions for molecular diagnosis of haemosporidian infections ...... 93

Table F.1 Summary of posterior distributions for population effect parameters of CLIPIC01 infection in honeyeaters ...... 111

Table F.2 Summary of posterior distributions for population effect parameters of CLIPIC02 infection in fairywrens ...... 113

FIGURES

Figure 0.1 Stages of haemosporidian infection in avian hosts, as reflected by changes in parasitemia in the peripheral circulation (after Valkiūnas 2005) ...... 5

Figure 1.1 Map of Brookfield Conservation Park, Australia, shaded to indicate dominant habitat types ...... 28

Figure 1.2 Locations of mixed-species flock observations within Brookfield Conservation Park, Australia ...... 30

Figure 1.3 Prevalences of Haemoproteus infections among 889 birds of 23 species in Brookfield Conservation Park, Australia ...... 38

Figure 1.4 Mixed-species flock network and distributions of flock size, flock richness, and node centrality of sampled host species ...... 40

Figure 1.5 Posterior densities for population effects of species’ traits on Haemoproteus infection probability ...... 42

Figure 1.6 Model-predicted shifts in Haemoproteus prevalence with changes in habitat and host species’ flock centrality, group size, and longevity ...... 43

Figure 2.1 Haemosporidian lineage phylogenies ...... 56

Figure 2.2 Time series of weekly sampling effort and Haemoproteus infections for each lineage and host family ...... 58

Figure E.1 Comparison of changes in probability space mapped from unit changes in real space by the inverse logit transformation, when unit changes occur in different locations in the domain...... 97

Figure E.2 Implied probability prior densities for parameter estimates mapped from real space by the inverse logit transformation. A comparison of flat (uniform) priors and weakly informative Gaussian priors ...... 98

Figure E.3 Markov chain Monte Carlo trace plots for population effect estimates and group scale parameter estimates ...... 102

Figure E.4 Posterior densities for population effect estimates and group scale parameter estimates ...... 103

Figure E.5 Graphical posterior predictive check for Haemoproteus prevalence ...... 104

Figure E.6 Graphical posterior predictive check for groupwise (species) Haemoproteus prevalence ...... 106

Figure E.7 Graphical prior predictive check for Haemoproteus prevalence ...... 107

Figure F.1 Posterior densities for population effects of species’ traits on CLIPIC01 infection probability among honeyeaters ...... 110

Figure F.2 Model-predicted change in CLIPIC01 prevalence with changes in average group size among honeyeaters ...... 112

Figure F.3 Posterior densities for population effects of species' traits on CLIPIC02 infection probability among fairywrens ...... 113

INTRODUCTION

The emergence of novel zoonoses has become a subject of increasing global interest over the past several decades. In particular, the recent COVID-19 pandemic has highlighted the importance of understanding connections linking the health of different species. Spillover events between species are expected to become more frequent if current trends towards increased global mobility, urbanization, agricultural expansion, and encroachment into natural areas continue (Epstein and Price 2009). Understandably, much popular and scientific attention has focused on the proximate and ultimate causes of disease spillover into human populations. However, spillover events are not limited to occurring in humans. Several viral and bacterial pathogens of human origin have been transmitted to wildlife and domestic animals, leading to outbreaks of varying scale and severity (reviewed in Epstein and Price 2009). Pathogen transmission between non-

human animal species is thought to play a major role in maintaining infections in certain

systems (e.g., canine distemper virus and canine parvovirus in Serengeti carnivores [Craft

et al. 2008, Behdenna et al. 2019]), and multi-host pathogens have been implicated as an important factor in the decline and extinction of wildlife populations (e.g., avian malaria in endemic Hawai’ian avifauna [Warner 1968, van Riper et al. 1986], chytrid fungus in amphibians [Berger et al. 1998]). Despite the important impacts of disease on wildlife and the potential for spillover into human and domestic animal populations, we lack a rigorous understanding of the environmental and ecological factors influencing pathogen transmission in many multi-host wildlife systems.

At a fundamental level, transmission events rely on direct or indirect contacts

between infected and susceptible individuals. Understanding patterns of transmission,

then, becomes partly a matter of understanding the patterns of contacts between

individuals. The nature of contacts necessary for pathogen transmission is determined by

the mode of transmission, and whether or not contact actually results in the spread of a pathogen depends on the individuals involved and pathogen’s host specificity. For trophically-transmitted parasites (e.g., cestodes), such contacts involve predation events.

For sexually-transmitted pathogens (e.g., Trichomonas, Chlamydia, herpesviruses), effective contacts involve mating. When there is an important environmental component to pathogen transmission (e.g., foot and mouth disease, canine parvovirus), or for vector-

borne pathogens, effective contacts need not involve direct interaction. Instead,

transmission may occur over relatively long timespans or broad spatial scales. Recently,

network analyses examining animal contact patterns have shed light on how these

contacts relate to pathogen and parasite transmission in wildlife populations (e.g., devil

facial tumor disease [Hamede et al. 2009], canine distemper virus [Craft et al. 2009,

2011], and ticks [Leu et al. 2010]). However, collecting both infection data and the

behavioral data needed to construct adequate contact networks can be challenging,

particularly in wild populations.

In this thesis, I explore the relationship between infection and host behavior

(specifically, social behavior) in the context of haemosporidian blood parasites of birds.

Haemosporidians can infect an ecologically and behaviorally diverse complement of

avian hosts (Bensch et al. 2009, Clark et al. 2014). This diversity, along with the widespread geographic distribution of both the hosts and their parasites, makes this multi-

3 host, multi-parasite system ideal for studying how host behavior and ecology relates to infection. In chapters 1 and 2, I present the results of a field study of an Australian avifauna that investigates whether variation in infection among host species can be explained by host social behavior and ecology. Before getting into the details of that study, I introduce the host–parasite system itself, beginning with a general description of haemosporidian infections of birds, their mode of transmission, and their impacts on avian hosts. For a more detailed overview of the biology of avian blood parasites, refer to

Appendix G. Next, I briefly summarize previous research relating blood parasite infections to host ecology. To close this general introduction, I offer some historical and geographic context to the current study. After describing the study and its results in greater detail, I conclude the thesis by providing a general discussion of those results and their implications for parasite transmission in communities with multiple competent host species. I also identify key questions in the system that remain unanswered and make recommendations for future research.

Haemosporidian parasites and their avian hosts

General characteristics of infections. The life cycle, range of competent hosts and vectors, and the physiological processes by which haemosporidians cause disease in their hosts are known in detail for only a handful of parasite species. However, most follow the same basic pattern. All haemosporidians are obligate intracellular parasites, with complex life cycles involving both a definitive host (in which the parasite reproduces sexually) and an intermediate host (where asexual reproduction occurs).

Definitive hosts for haemosporidians are insects of the Order Diptera. The insect hosts are referred to as “vectors” on account of their important role in transmitting blood

4 parasites between birds. Haemosporidians are transmitted to susceptible birds when an infected vector takes a blood meal. In general, each group of haemosporidian parasites is transmitted by a different vector family. Avian malaria parasites (Plasmodium spp.) are transmitted by mosquitoes (Culicidae), Haemoproteus spp. by biting midges

(Ceratopogonidae) and louse flies (Hippoboscidae), and Leucocytozoon spp. parasites by black flies (Simuliidae) (Valkiūnas 2005, Atkinson et al. 2008). In addition to host (bird) ecology, differences in the ecology and behavior of vector species may shape patterns of infection.

The stereotypical haemosporidian infection progresses through four general stages in the avian host, each characterized by changes in the number of parasites circulating in the bloodstream (Fig. 0.1; Atkinson and van Riper 1991, Valkiūnas 2005).

During the incubation stage, parasites invade the liver, kidneys, muscle, and other fixed tissues, where they multiply rapidly. Infections are not detectable in the blood at this stage. During the acute stage, parasitemia (the number of parasites circulating in the bird’s bloodstream) rises sharply as parasites emerge from the fixed tissues in large numbers, enter the circulation, and penetrate the host’s red blood cells. This phase of infection culminates in the crisis stage when parasitemia reaches a maximum before dropping off to lower levels. Birds may then maintain these low levels of parasitemia

(chronic stage) for months or years as parasites that remain in the fixed tissues continue to release cells into the bloodstream. Occasionally (often coincident with the onset of breeding or migration), hosts experience periodic relapses to more acute levels of infection. Some individuals may succeed in clearing infection from the bloodstream, although infections in the latent stage typically remain in the fixed tissues of the liver and

5 other organs, and may give rise to subsequent episodes of chronic or acute infection. In general, once infected, most hosts are thought to remain infected for the remainder of their life (Valkiūnas 2005). Blood-based diagnostics (including most conventional microscopy and PCR-based tests for haemosporidian infection) are unable to detect incubation- or latent-stage infections as these are characterized by an absence of parasites in the bloodstream, but these assays are generally considered sensitive to even the low levels of parasitemia characterizing chronic infections (Richard et al. 2002, Krams et al.

2012).

Figure 0.1 Schematic showing stereotypical progression of haemosporidian infection in birds, as measured by the number of parasites circulating in the host bloodstream (i.e. parasitemia). Shaded regions indicate periods when diagnosis from peripheral blood samples is theoretically possible. Actual detectability of infection is also influenced by the intensity of infection and the sensitivity of the assay. Birds are infected when bitten by an infectious vector (mosquito, midge, or other fly) taking a blood meal. Thereafter, the infection progresses through four general stages: (1) During the incubation stage, parasites invade the liver and other fixed tissues and multiply rapidly. (2) Acute infection is characterized by a sharp increase in parasitemia when parasites flood the bloodstream and penetrate host red blood cells. After reaching a crisis point, the infection typically subsides to a lower intensity. (3) Low-intensity chronic infections may persist for months or years, and may or may not give rise to periodic relapses to acute levels. (4) If the host manages to clear all parasites from the circulation, but additional parasites remain in the fixed tissues of the liver and other organs, the infection is considered to be in a latent stage. The remaining cells may give rise to new chronic and acute infections. Note that the durations of stages are not necessarily to scale. After Valkiūnas (2005).

Pathogenicity and impacts on hosts. Haemosporidian infections in the brain, lungs, heart, and other organs can lead to inflammation, internal bleeding, and tissue

6 death as circulating parasites block capillaries and promote inflammatory immune responses. Capillary occlusions associated with heavy infections can lead to cerebral paralysis, pneumonia, and death. Parasites active in the bloodstream cause anemia by effecting hemolysis and promoting changes to blood chemistry that reduce the oxygen- binding capacity of hemoglobin (Valkiūnas 2005).

Despite these severe physiological impacts of haemosporidian infection, the vast majority of infected birds encountered in the wild exhibit little (if any) overt clinical signs of disease (Bennett et al. 1993, Valkiūnas 2005). This observation can be partly accounted for by the apparent benignancy of low-intensity infections, and the relative infrequency with which high-intensity cases are encountered. Yet neither infection status nor the intensity of infection by blood parasites appears to have a strong relationship with body mass or fat deposition (Ashford 1971, Smith and Cox 1972, Bennett et al. 1988), two conventional metrics of condition in birds. Based on a review of literature reports of mortality and gross pathology resulting from avian blood parasite infections, Bennett et al. (1993) note that most cases of severe disease and death are reported in domestic fowl, represent novel host–parasite associations, or else cannot unequivocally be attributed to blood parasite infection. They then interpret these observations as suggesting that haemosporidioses are unlikely to be a major source of mortality among their natural hosts. At the same time, the authors acknowledge that reports of virulence in wild host populations are usually based on opportunistic encounters or circumstantial evidence, and that most studies of blood parasite infections in the wild are ill-designed for detecting pathogenic effects. Thus, it would seem that we cannot draw strong conclusions about the pathogenicity of avian blood parasites from the largely observational and survey-based

7 field studies that dominated the avian blood parasite literature during much of the twentieth century.

Contrasting with the interpretation of Bennett et al. (1993) that avian blood parasites are unlikely to be an important source of mortality for birds, Valkiūnas (2005) highlights a number of reports of severe pathology, often leading to death, among wild birds infected with parasites that are apparently benign in conspecific hosts. Many of these case studies involve opportunistically-encountered individuals experiencing high- intensity parasitemia and/or an unusually large number of parasites in various fixed tissues (e.g., Markus and Oosthuizen 1971, Peirce 1984, Atkinson and Forrester 1987).

From this information, Valkiūnas (2005) concludes that blood parasite infections can and do cause significant morbidity and mortality in their natural avian hosts, but that standard methods of encounter (especially passive mist-netting) probably undersample diseased individuals. Additionally, predation events may selectively remove diseased birds from the population, further reducing the apparent prevalence of infection (Valkiūnas 2005).

Recent experimental studies have more clearly illustrated the extent and subtlety of impacts of blood parasite infection on host fitness. By experimentally infecting three wild-caught passerines with Plasmodium (Giovannolaia) homocircumflexum, Ilgūnas et al. (2016) demonstrated the ability of that parasite to cause mortality in hosts.

Intriguingly, death in each case occurred after the crisis stage of infection, when parasitemia in the peripheral circulation was in decline. Upon examination of tissues,

Ilgūnas et al. (2016) attributed mortality to cerebral paralysis due to the blockage of brain capillaries by parasites.

Even when infection by blood parasites does not lead to death, a growing body of

evidence indicates that there are significant lifetime impacts of chronic infections on host fitness. Controlled studies of free-living blue tits (Cyanistes caeruleus) have revealed that chronically-infected females with experimentally-reduced parasite loads experience higher fecundity (greater hatching and fledging success) than do sham-treated individuals

(Merino et al. 2000, Knowles et al. 2010). Malarial infections among juvenile songbirds can influence the structure of learned songs and the development of neural circuitry associated with song-learning (Spencer et al. 2005), with potential effects on lifetime reproductive success mediated by mate choice. Taken together, these results suggest that blood parasite infections can impose major costs on lifetime fitness potential of affected individuals, even if they do not directly cause mortality.

Connecting avian ecology and blood parasite transmission

The relationships between blood parasite infection and different facets of avian ecology have been explored in several host communities, yet few generalizable patterns have emerged. Instead, trends in blood parasite infection with respect to host ecology that are clearly demonstrated in one time and place are often reversed or nonexistent in others.

The aspects of host ecology highlighted below provide a foundation for understanding the importance of host behavior and environment in driving patterns of infection. The apparent inconsistencies in the results of otherwise similar studies underscore the need for additional research to clarify these relationships.

Microhabitat associated with foraging stratum and nest structure. Although individual vectors are capable of making long-distance flights, movements between

individual feeding bouts tend to be relatively circumscribed (Mullen and Durden 2019).

Furthermore, the dipteran vectors of avian blood parasites are often distributed heterogeneously across landscapes. For instance, individual species of mosquitoes and

midges frequently occupy comparatively narrow vertical zones, with distinct patterns of

stratification that tend to covary with host preference. Typically, ornithophilic species are

found in higher strata, and mammalophilic species in lower strata (Swanson and Adler

2010, Johnston et al. 2014). Bird species also demonstrate distinct patterns of space use

while foraging, roosting, and nesting, such that we might expect hosts to encounter

vectors at different rates. Together, these traits should yield an uneven spatial distribution

of exposure and infection risk for hosts.

Researchers examining patterns of haemosporidian prevalence among avian

species on the basis of microhabitat routinely focus on the height at which birds nest and

forage as indices of microhabitat use. Following the observation that ornithophilic vectors

tend to occur in higher vegetative strata, Garvin and Remsen (1997) predicted that the

prevalence of haemosporidian parasitism should be highest among species that forage or

nest in the canopy (Bennett and Fallis 1960, Garvin and Remsen 1997). Although this

pattern has been observed in a number of systems, it is far from universal. For instance,

González et al. (2014) reported a higher prevalence of Haemoproteus infection among

species nesting in the understory and midstory than among either ground- or canopy- nesting species in Colombia. In contrast, Fecchio et al. (2011) found that Haemoproteus prevalence increased with nest height among savannah-dwelling species in the same biogeographic region. Yet in a subsequent analysis of the same system using different diagnostic methods, no relationship between Haemoproteus infection and nest height was detected (Fecchio et al. 2013), nor was nest height associated with either Haemoproteus

or Plasmodium prevalence in a large-scale retrospective survey of Andean bird specimens (Barrow et al. 2019). The same retrospective analysis did reveal a higher rate

of Leucocytozoon parasitism among species nesting in low strata than among those

nesting on the ground (Barrow et al. 2019). In sub-Saharan Africa, both nest height (Lutz

et al. 2015) and roost height (Hellard et al. 2016) were associated with Haemoproteus

infection, with canopy-nesting and canopy-/midstory-roosting species showing higher prevalences than those nesting and roosting on the ground. Hellard et al. (2016) also reported the highest prevalence of Plasmodium infection among species roosting in the midstory compared to those roosting in either the canopy or on the ground. In contrast, no association between nest height and Plasmodium infection was found by Lutz et al.

(2015). Working in the tropics of Australia, Laurance et al. (2013) found both

Plasmodium and Haemoproteus infections to be associated with ground-foraging behavior. However, neither González et al. (2014) nor Barrow et al. (2019) detected an effect of host foraging height on the prevalence of infection for any of the three blood- parasite genera in the Neotropics.

The structure of nests might also influence the rate of encounter between hosts and vectors. Both nestlings and incubating adults are at risk of infection in the nest, and its structure could mediate that risk by modulating the chemical and visual cues used by host-seeking vectors to find blood meals. In both Neotropical and Afrotropical systems, closed-cup-nesting species tended to have higher prevalences of Plasmodium infection, but lower prevalences of Haemoproteus infection, than did species that nest in open cups

(Fecchio et al. 2011, Lutz et al. 2015, Barrow et al. 2019). To explain this distinction,

Lutz et al. (2015) point out that culicoid midges, which transmit many Haemoproteus

spp., tend to rely more heavily on visual cues than do the vectors of Plasmodium spp., which rely predominantly on olfactory cues. They suggest that enclosed structures might concentrate these olfactory cues, facilitating host-finding by mosquitoes, while obstructing the visual cues relied upon by midges. Nonetheless, as with nest height and foraging stratum, there is variation among studies with respect to the relationship between nest structure and blood parasite infection. González et al. (2014) reported no significant differences in either Haemoproteus or Plasmodium prevalences between open- and closed-cup-nesting species in Colombia (though species with these nest types had higher Plasmodium prevalences than did cavity-nesting species). Likewise, Fecchio et al. (2013) found no difference in infection rates by either parasite genus among bird species using different nest types.

Social behavior. Interactions between individual hosts form the basis for social behavior and influence parasite transmission (reviewed in Kappeler et al. 2015). Some social interactions might augment transmission by providing opportunities for pathogens to spread between individuals. This could lead to selection for avoidance (behavioral immunity, wherein individuals recognize and avoid diseased associates as a preemptive means of preventing exposure) or resistance (physiological immunity, wherein the host immune system actively fights infections in the body) traits among social individuals to offset the costs of enhanced transmission potential. However, the social environment itself may also provide means of reducing the risk of parasitism in social animals.

Antiparasite social behaviors such as allopreening could reduce disease transmission if they limit encounters between susceptible hosts and infected vectors. Socially-derived benefits from improved foraging efficiency and predator avoidance may also lessen

nutritional and physiological stresses, freeing up resources for investment in

immunological resistance. The consequences of sociality for host infection patterns, then,

are not necessarily straightforward. Instead, they reflect the outcomes of complex

evolutionary interactions between host social organization, behavior, and immune

function on one hand and parasite virulence and transmission on the other. In a review of behavioral mechanisms mediating transmission in the context of animal social systems,

Loehle (1995) posits that many of the features of host societies that are commonly regarded as having effects on parasite transmission may themselves be shaped by disease-

induced selection.

It is useful to conceptualize variation in animal social behavior as occurring along

two related but independent axes describing differences in sociality within a species

(intraspecific social behavior) and between species (interspecific social behavior).

Among the most relevant and generalizable social variables in the context of avian blood

parasite transmission are conspecific group size (a measure of intraspecific sociality),

breeding system (an intraspecific measure closely tied to group size), and flocking

behavior (a measure of interspecific sociality). Contrasting with the classification

schemes for microhabitat discussed above, the distinctions between these variables are

more difficult to define. Part of this difficulty is due to seasonal changes in host

behavior—for example, some species may only participate in mixed-species flocks

during the non-breeding season, or conspecific group size may vary between breeding

and non-breeding contexts. Another challenge is that there is considerable conceptual

overlap between these metrics, and interpretations and classification schemes vary

considerably in the literature. For instance, Barrow et al. (2019) separately classify

species on the basis of foraging group size (solitary, family groups, conspecific flocks, or

mixed-species flocks) and breeding system (uniparental, cooperative, or colonial). Lutz et

al. (2015) classify flocking behavior (solitary/pairs, single-species groups/flocks, or mixed-species flocks), but do not address breeding sociality. Each of these classification schemes includes aspects of both intra- and interspecific social behavior, making it difficult to tease apart the unique effects of either. In other cases, species are simply

identified according to their broadly-defined social system (pairs or groups/flocks), and it is not always clear whether these classifications refer to breeding or foraging systems, or

the extent to which intra- and interspecific sociality is considered (Fecchio et al. 2011,

2013; González et al. 2014). Considering that individual and group behavior may differ

between foraging and breeding contexts, and that transmission dynamics within a host

species are likely different from those between species, these distinctions may prove

important in terms of their impacts on blood parasite infection risk.

The difficulty in understanding the impact of social behavior on avian blood

parasite infections is compounded by the lack of consistent patterns across systems. With

respect to flocking behavior, species participating in mixed-species flocks had higher

prevalences of Plasmodium, but lower prevalences of Haemoproteus, than did solitary

and pair-living species in the tropics of sub-Saharan Africa (Lutz et al. 2015). Patterns of

haemosporidian infection with respect to flocking behavior appear different in the

Neotropics. González et al. (2014) reported no effect of group size (solitary vs. group-

living) on either Plasmodium, Haemoproteus, or Leucocytoozoon prevalence, but did

observe higher rates of Haemoproteus and Leucocytozoon infection among species

participating in mixed-species flocks than among non-flocking species. Fecchio et al.

(2011, 2013) found that group-living species had higher Haemoproteus prevalences than

did solitary species. In contrast, Barrow et al. (2019) found no effect of foraging group

size on blood parasite infection, but did observe lower prevalence of Leucocytozoon

infection among species that nested or roosted colonially than among non-colonial

species. Additional research is needed to reconcile these disparate findings and to identify

the aspects of intra- and interspecific social behavior that are most relevant for

haemosporidian transmission in birds.

Movement ecology. The relationship between infection patterns and host movement is complex, and a variety of outcomes of large-scale movements are possible

(reviewed by Altizer et al. 2011). Long-distance movements could introduce hosts to

novel parasites (migratory exposure). Animal migration might then facilitate the spread of parasites and pathogens between widely-separated geographic areas and host communities. Alternatively, such movements might act as a means of parasite avoidance, particularly if host absence coincides with periods of high transmission potential

(migratory escape). The physiological demands of migration could also selectively remove infected individuals from the population, thereby reducing the overall prevalence of infection and risk of transmission (migratory culling). On the other hand, these demands could impair immune function, reducing the host’s ability to resist infection

(migratory susceptibility). Thus, infection may represent a cost, or infection avoidance a benefit, of migration. The extent to which each of these processes occurs largely depends on ecological and epidemiological context. Local transmission ecology on the breeding and non-breeding grounds, parasite virulence, host specificity of the parasite, and host

tolerance each likely play a role in determining whether host movements result in an

increase or decrease in parasite transmission (Altizer et al. 2011, Becker et al. 2020).

Given the potentially opposing mechanisms by which migration might influence

the prevalence of infection in a population, it seems unsurprising that researchers have

reported contrasting patterns of haemosporidian infections in birds with respect to their

migratory status. Several large-scale, longitudinal studies in the mid- to late twentieth century found evidence suggesting migration was an important risk factor for blood parasite infections. A combination of high prevalence and high intensity of

Haemoproteus and Leucocytozoon infections among birds arriving on the breeding grounds in Ontario led Bennett and Fallis (1960) to suggest that these birds had acquired new infections during spring migration. High prevalences of blood parasites (principally

Haemoproteus, Leucocytozoon, and haemogregarine parasites) were also observed in trans-Saharan spring migrants arriving in Britain (Bennett et al. 1974, Peirce and Mead

1978). Contrasting with the situation observed by Bennett and Fallis (1960) in North

America, the intensity of infections among migrants in Britain was low (Peirce and Mead

1978). Since the intensity of secondary parasitemia is generally lower than that of initial infections, Peirce and Mead (1978) posited migratory relapse (rather than migratory exposure or susceptibility) as a contributing factor to the high prevalence among migratory species in Britain. More recently, however, studies of the impact of migration on haemosporidian infections have challenged the view that blood parasites constitute a major cost of migration. In a comparison of haemosporidian prevalence (combined

Plasmodium and Haemoproteus infections) between populations of migratory and sedentary dark-eyed juncos (Junco hyemalis) overwintering together in Virginia,

Slowinski et al. (2018) found a significantly lower prevalence among migratory than among sedentary birds. This observation led them to conclude that migration afforded juncos a means of escape from blood parasite infection, or that migratory culling reduced the apparent prevalence of infection in the migratory population. In song sparrows

(Melospiza melodia), migration distance (inferred by stable hydrogen isotope ratios in claw samples) was related to blood parasite infection risk (pooled Plasmodium,

Haemoproteus, and Leucocytozoon infections; Kelly et al. 2016). Individuals that travelled farther between wintering and breeding grounds had a higher probability of infection than did those migrating shorter distances. In contrast, migration status was not a good predictor of either Plasmodium or Haemoproteus infections in resident and

Neotropical migrant species (Fecchio et al. 2013, González et al. 2014), but Neotropical residents did have higher Leucocytozoon prevalences than did migrant species (González et al. 2014).

Reconciling the results of these studies poses an intriguing challenge, and there are many potentially confounding influences between migration and infection. The timing of sampling in relation to host breeding and migration phenology is a critical but often overlooked component of studies of migration and its effects on infection.

Sedentary and migratory populations may experience very distinct annual cycles, and these differences should be taken into consideration when comparing infection outcomes between the two groups. In some areas of the tropics, sedentary populations may breed at any time of year as conditions permit. Thus, some tropical communities could include breeding residents, nonbreeding residents, and nonbreeding migrants. Both reproduction and migration are energetically expensive, but they may have different impacts on body

condition, physiology, and immune function. These differences in host physiological

state may also play a role in driving observed infection patterns independent of migration,

such that direct comparisons of prevalence between migrant and sedentary birds could be

confounded (e.g., by effects of breeding state). The location of sampling is also an important consideration. Populations overwintering in Virginia (e.g., Slowinski et al.

2018) probably experience very different conditions than do those in tropical and subtropical regions of South America (e.g., Fecchio et al. 2013, González et al. 2014) and

Africa (e.g., Waldenström et al. 2002). These differences, in turn, likely have important

implications for breeding patterns in the sedentary populations and for vector activity

patterns.

Despite their limitations, cross-sectional surveys of sedentary and migrating birds

can be informative, especially for identifying when and where transmission occurs. When

parasite lineages are identified that are not known from a species’ breeding range but

which are common among sedentary species sharing the wintering range, it can be inferred that transmission occurs on the wintering grounds. This was the case in a survey of migrant and resident birds in northern Nigeria, where lineages from African residents

were found infecting Palearctic migrants (Waldenström et al. 2002).

The lack of clear consensus between studies of blood parasite infections and

microhabitat, social behavior, and movement ecology could have a number of

nonexclusive explanations. First, it may indicate that exposure is not strongly influenced

by these aspects of host ecology, and that any observed relationships are merely spurious.

Second, the disparities between studies could reflect different mechanisms driving

exposure in different systems. This may explain variation in the direction and magnitude

of effects observed in infection by different parasite taxa, for instance, as transmission in

each of these systems is likely subject to different ecological and epidemiological

constraints. Third, it may be that sampling bias due to parasite-induced mortality

obscures the relationships between ecology, exposure, and apparent infection. Finally, the

classification schemes (and/or analytical methods) used by different studies to categorize

species’ ecological traits may be incompatible with one another, thus invalidating direct comparison. Regardless of the generative processes responsible, the result of these inconsistencies is a lingering uncertainty about the impact of host ecology on avian blood parasite infections. Closer examination of individual ecological variables and experimental approaches are needed to fully resolve these questions.

Historical and geographic context of haemosporidian infection ecology research

Although sparse, the history of haemosporidian studies in Australia is long. Only a quarter of a century after Russian physiologist Vasily Danilewsky became the first to describe parasitic protists in the blood of vertebrates (Valkiūnas 2005), researchers working in Australia started a concerted effort to identify parasites present in the fauna of that continent. In the first few decades of the twentieth century, J. Burton Cleland and T.

Harvey Johnston pioneered this work in Australia (Cleland and Johnston 1909, 1910,

1911; Johnston and Cleland 1909; Johnston 1910, 1912; Cleland 1915, 1922). Early efforts to catalog the diversity of blood parasites of birds were largely systematic in nature, including comprehensive descriptions of the parasites involved. However, most offer little detail about the individual afflicted hosts or their environments apart from species and the general location of encounter. Such reports are invaluable from a taxonomic perspective, and can also provide important insights into the host specificity,

pathology, and geographic distributions of avian haemosporidians. However, making

inferences about the ecology of host–parasite interactions from these early reports presents serious challenges given the paucity of available data about sampling protocols, encounter rates, and habitat.

Beginning in 1910, Cleland and Johnston made a key advance in the field from an ecological perspective by including a listing of individual birds examined and found to be free of parasites (Cleland and Johnston 1910, 1911; Cleland 1915, 1922). This was apparently done in an attempt to establish with greater certainty the limits on host ranges for parasite species, but the inclusion of this list also helps to shed light on regional patterns in the prevalence of blood parasite infections. Nonetheless, these early reports

invariably suffer from small sample sizes (almost always <10, and frequently 1–2, per

species), inconsistent protocols, and incomplete reporting of habitat and/or location.

Without such information, any inferences from comparative analyses of the ecological

factors influencing infection rates are severely weakened.

Unfortunately, and in addition to the shortcomings described above, the accuracy

of parasite identifications (especially separation of Plasmodium spp. from Haemoproteus

spp.) reported in many of the early records is dubious. In their taxonomic review of the

blood parasites of Australian birds, Mackerras and Mackerras (1960:239) note that “the

discovery of species of Plasmodium which produce elongate, halter-shaped gametocytes

throws considerable doubt upon identifications of Haemoproteus based on single dried

blood films.” The presence of such gametocytes was formerly considered a diagnostic

morphological trait of Haemoproteus (syn. Halteridium) parasites.

In recent decades, molecular diagnostic methods that enable rapid detection and

identification of haemosporidian parasites have fueled a global resurgence in studies of the ecology of avian blood parasite infections (Bensch et al. 2009). This has been especially true in North and South America, Europe, and Africa. Comparatively few surveys of haemosporidian infection in Australian birds have been conducted in recent years, despite its endemic bird diversity (Clark et al. 2014). Moreover, many of the studies that have been performed sampled a relatively small number of host species (e.g.,

Kleindorfer et al. 2006, Colombelli-Négrel and Kleindorfer 2008, Balasubramaniam et al.

2013, Clark and Clegg 2015, Clark et al. 2015a, Eastwood et al. 2019; but see Beadell et al. 2004, Zamora-Vilchis et al. 2012, Laurance et al. 2013, Clark et al. 2015b). Although targeted sampling is an efficient way to answer specific questions about the dynamics of particular host–parasite relationships, the development of these questions is often contingent upon having at least modest foreknowledge of the focal host(s) and parasite(s). Community-wide surveys offer a means of obtaining such information while controlling for variation in space and time that could confound comparisons made between study systems. In addition, community-wide surveys can be designed to answer questions that are difficult to address with a narrower focus, thereby contributing important information to the broader understanding of haemosporidian infection ecology.

Thus, there is a need for additional broad-scope haemosporidian surveys in Australian avifauna, similar to those conducted in other regions (e.g., Sehgal et al. 2005; Ishtiaq et al. 2007; Fecchio et al. 2011, 2013; Loiseau et al. 2012; González et al. 2014; Lutz et al.

2015; Hellard et al. 2016; Barrow et al. 2019).

Research motivation and objectives

We aim to provide fine-scale insight into the potential risks of blood-parasite infection associated with sociality. Building on past work that examined the influence of broad patterns of social behavior on haemosporidian infections, we pair infection data

with direct observations of social behavior in the avian focal community at Brookfield

Conservation Park, South Australia. Further, we seek to supply needed baseline infection

data for a poorly-surveyed region. In this study we will address two principal lines of questioning:

(1) How do species’ social (flocking behavior and conspecific group size) and

ecological (foraging stratum, nest structure, and regional movement patterns) traits

relate to blood parasite infection status in a diverse avian community? We predict

that more social species (i.e., those with larger conspecific group sizes and that are

more central within the mixed-species flocking community) will have a higher

prevalence of blood parasite infections. We further expect that species foraging in

higher strata will exhibit greater prevalence of infection.

(2) Which lineages of blood parasites are present in the avian community at Brookfield

Conservation Park? How does the diversity of blood parasite lineages in the focal

community relate to regional or global diversity patterns? How does lineage

diversity relate to host species’ social behavior, ecology, and movement patterns

(i.e., whether a species is resident or nomadic/migratory)? We predict that species

that have a relatively central position within the mixed-species flocking community

will harbor a greater diversity of blood parasite lineages.

CHAPTER 1

RELATING SOCIAL ECOLOGY TO BLOOD PARASITE INFECTIONS IN A

DIVERSE AVIFAUNA

Infectious diseases of wildlife are ecologically important, and have major implications for conservation. Though the effects of parasites and pathogens are often considered at the level of the individual, infections can have devastating impacts on

wildlife populations. For example, recurrent epizootics of Phocine morbillivirus

beginning in the late 1980s contributed to several mass mortalities among marine seals in

the north Atlantic (Härkönen et al. 2006). Similarly, bacterial and fungal pathogens have

been implicated in regional and global declines of wildlife populations (e.g., sylvatic

plague in prairie dogs [Pauli et al. 2006], and chytrid fungus in amphibians [Vredenburg

et al. 2010]).

Pathogens can pose serious threats to naïve populations, particularly in an era marked by global climate change and accelerated movements of people, goods, and animals. Climate change is expected to cause large-scale shifts in patterns of infection,

and may expose vulnerable populations to novel threats or make it more difficult to cope

with existing pathogens (Garamszegi 2011, Raffel et al. 2013, Clark et al. 2018, reviewed

in Gallana et al. 2013). Environmental change, including climate change and land-use

change, can lead to complex, interacting effects on hosts, pathogens, and vectors

(Murdock et al. 2012). The outcome of these interactions is not always straightforward.

For instance, rising temperatures increase the rate of development for malarial parasites

within mosquito vectors, but also stimulate the vectors’ immune function, with a net

reduction in vectorial capacity at higher temperatures (Paaijmans et al. 2012, Mordecai et al. 2013, reviewed in Gallana et al. 2013). To effectively manage these growing and evolving challenges, we need a clearer understanding of ecological effects on transmission and disease risk in diverse host–parasite systems.

A critical ecological consideration for many host–parasite systems is the social

framework of the hosts. Social structure comprises a series of interactions between

organisms, whether direct (e.g., parents teaching offspring, communal roosting) or

indirect (e.g., scent-marking), by which individuals accrue benefits. These connections

act as channels through which information—about foraging resources, territory

boundaries, predators, and more—spreads. The same connections can serve as pathways

for the spread of parasites and pathogens, as social interactions may enable the

transmission of an infection by the direct exchange of pathogens or by facilitating the

spread of vectors such as lice, ticks, or flies. Across host species, greater sociality is often

associated with higher prevalence of infection. This trend is evident on evolutionary (i.e.,

from solitary to colonial living; Tella 2002) and ecological (i.e., social group size; Brown

and Bomberger Brown 1986, Brown et al. 2001) scales.

Studies of how variation in social behavior influences infection risk are often

limited to examining interactions between conspecifics. However, many pathogens and

parasites readily pass between host species. Such spillover events may be important

drivers of pathogen and parasite persistence in some instances (Craft et al. 2009). This

raises questions about how interspecific associations affect the risk of infection in

communities with multiple competent host species.

When species differ in exposure rates, susceptibility, or competence, interspecific

associations may amplify or dilute transmission in multi-host systems relative to single-

host systems (Keesing et al. 2006). The potential for amplification and dilution is

typically investigated from a local host diversity perspective, with infection outcomes

being compared across environments with varying levels of host diversity. Yet even

within an environment, hosts may experience local diversity differently by engaging in

more or fewer interspecific interactions with other members of the community. The

influence of such interspecific associations on the risk of infection remains poorly characterized. Insight into the epidemiological consequences of social interactions, especially between members of different species, could shed light on the mechanisms by which pathogens, including zoonotic pathogens, spill over from one host species to another.

The avian host–haemosporidian blood parasite (Apicomplexa: Haemosporidia) system is well suited for studies relating infection outcomes to host ecology.

Haemosporidians are among the most widespread groups of blood parasites, infecting birds from 30 taxonomic orders representing a vast spectrum of habitat associations, life history strategies, and ecologies (Bensch et al. 2009). This parasite group, which includes the causative agents of avian malaria (Plasmodium spp.), has received attention from researchers due to its importance in driving declines among island-endemic bird

populations (e.g., LaPointe et al. 2012), and as a model for predicting how vector-borne diseases may respond to anthropogenic change (e.g., Garamszegi 2011, Clark et al.

2018). The wide geographic distribution of avian blood parasites and the taxonomic diversity of their hosts make avian haemosporidians a valuable focus of study for

building a broad understanding of infection dynamics in multi-host systems.

Reflecting their taxonomic richness, birds demonstrate a remarkable diversity of social ecologies. Species show variation in the number and nature of interactions with both conspecific (intraspecific social behavior) and heterospecific individuals

(interspecific social behavior). Previous field studies of haemosporidian infections have looked for covariation of infection risk with host social behavior, but with mixed results.

Among avian communities in the central Brazilian Cerrado, Fecchio et al. (2011, 2013) found that species living in larger conspecific family groups had higher prevalences of

Haemoproteus infection than did those living in pairs. This finding is broadly consistent with an increase in prevalence reflecting evolutionary transitions to greater sociality, as suggested by Tella (2002). However, no association between sociality and Plasmodium infection was observed in Brazil (Fecchio et al. 2011, 2013), nor among nearly 250 species inhabiting a variety of habitats in Colombia (González et al. 2014). With respect to flocking behavior, González et al. (2014) found that species that participate in mixed- species flocks had higher prevalences of Haemoproteus infection than did solitary species and those forming single-species flocks. In contrast, Lutz et al. (2015) reported a weak trend in the opposite direction (i.e., Haemoproteus prevalence among species forming single-species flocks was marginally higher than that among solitary species and those engaging in mixed-species flocks). Barrow et al. (2019) found no differences in

Haemoproteus or Plasmodium prevalences between solitary and pair-forming species, those living in groups, those forming single-species flocks, and those participating in large mixed-species flocks. Taken together, these disparate findings make it difficult to infer general patterns about the effects of social behaviors on infection risk.

A possible explanation for the apparent lack of consensus between previous

studies is that the coarse perspective by which social behavior is frequently considered is

inefficient at capturing the complexity of the behavior–infection relationship. By this

traditional framework, species are classified into broad behavioral categories based on

general knowledge of the species’ life history, and mean prevalence is then compared

between categories. Although this approach may be sufficient to detect strong ecological

patterns, it reduces the rich spectrum of sociality to a small and rigid set of arbitrary

categories. To address this issue, we paired haemosporidian infection data with high-

resolution observations of social behavior specific to the focal host community. These

data were then combined in a social network analysis framework to provide a nuanced

view of host social behavior.

To understand the impact of host ecology and social behavior on risk of infection, we investigated relationships between haemosporidian blood parasite infections and host flocking behavior, group size, foraging and movement ecology, habitat, and nest structure in a diverse Australian avifauna. We used direct observations of flocking behavior from

the focal community to infer each host species’ position in the social structure, allowing

us to test for effects of fine-scale differences in social behavior between host species.

Specifically, we aimed to address the following questions:

(1) How does the prevalence of haemosporidian blood parasite infections differ among

members of the avian community at Brookfield Conservation Park, South

Australia?

(2) Can species-level variation in prevalence be explained by life history traits (nest

structure, foraging stratum, longevity), social behavior (conspecific group size,

flock network centrality), and/or ecology (microhabitat, migratory status)? We

predicted that species with larger conspecific group sizes and those that are more

central in the mixed-species flock network would exhibit comparatively high

prevalence of avian blood parasite infections.

Materials and methods

Study area. We sampled birds during the 2019 breeding season (late August– mid-December) at Brookfield Conservation Park near Blanchetown, South Australia, in the lower Murray–Darling River Basin (Fig. 1.1; −34.35°, 139.50°). The landscape consists of rolling plains with a persistently dry grassland climate (modified Köppen classification scheme; BOM 2005). The park is dominated by second-growth mallee eucalypt woodland and chenopod scrub.

Figure 1.1 Brookfield Conservation Park, South Australia. Green-shaded areas are predominantly open mallee eucalypt woodland. Areas shaded brown are dominated by chenopod scrub. Circles mark sampling locations buffered by 150 m, colored according to the percent cover of mallee habitat near the capture site. Field methods. During the 2019 breeding season, we captured adult and juvenile birds in mist nets using a combination of flushing and audio playback. We marked the location of each capture site using handheld GPS units (Garmin International Inc.,

Olathe, KS, USA). Captured birds were fitted with uniquely numbered, permanent aluminum leg bands issued by the Australian Bird and Bat Banding Scheme (ABBBS).

After banding, we processed and sampled each bird before releasing it at the capture site.

We collected a small (5–25 µL) quantity of blood from each bird by brachial venipuncture into heparinized micro-capillary tubes. Samples were immediately

29 transferred to Whatman® FTA® classic cards (GE Healthcare, Chicago, IL, USA) for storage and transport.

We determined the age of birds based on the amount of skull ossification and characteristics of the plumage and soft parts as described by Hardy (2019). Each individual was classified as either an adult (after hatch-year) or juvenile (hatch-year).

When possible, we determined the sex of birds based on plumage. For individuals that could not be reliably sexed in the field, we later determined the genetic sex using PCR- based methods (see Appendix D for details about molecular sexing protocols).

Each day during the field season, we conducted flock and foraging observations opportunistically. The spatial distribution of sampling effort for flock observations reflected that for captures, although the open scrub habitat was not extensively used for flock observations (Fig. 1.2). We identified flocks as mixed-species groups of birds either aggregating at a stationary resource (e.g., flowering eucalypt trees) or foraging and moving together in a coordinated manner. Whenever we encountered a flock, we recorded its composition, size (number of individuals of each species), duration, and location. Foraging behavior was observed both in flocking and non-flocking contexts. For each foraging attempt we observed, we noted the stratum in which the focal individual was foraging (on the ground, in scrub <0.5 m in height, in a bush, in the low canopy/midstory, or in the upper canopy), the number of any conspecifics present, and whether or not the individual was foraging as part of a mixed-species flock.

Figure 1.2 Locations of mixed-species flock observations recorded during Aug–Dec, 2019, in Brookfield Conservation Park, South Australia. Green-shaded areas indicate mallee woodland habitat, and brown areas are dominated by chenopod scrub. Seventeen flock observations are not represented, as they lacked spatial data.

Research ethics. All field activities received prior approval from the University

of Nebraska Institutional Animal Care and Use Committee (Project ID: 1822), the South

Australian Wildlife Ethics Committee (approval numbers 9/2019, 13/2019), and the

Government of South Australia Department for Environment and Water (DEW; permit number M26867-2). Banding and research activity were conducted under licenses to A.

E. Johnson (DEW research license number 365, ABBBS banding authority number

3352).

Ecological data. For each species, we calculated the average conspecific foraging

group size and modal foraging stratum (i.e., the stratum in which members of a species

were most frequently observed foraging) across flocking and non-flocking contexts from

the behavioral observations. To confirm foraging height data for species with few direct

observations, we compared our observations to data available in the Australian Bird

Database (Garnett et al. 2015) and the EltonTraits foraging database (Wilman et al.

2014). We compared group sizes and foraging strata between flock contexts separately for each species using Wilcoxon Mann-Whitney tests (group size) and Fisher’s exact tests

(foraging strata). In addition to direct observations of social and foraging behavior, we gathered data concerning other ecological and life-history traits from external databases.

We classified each species as nesting in an open cup, closed cup, or hollow (cavity or burrow) based on information available in the Birds of the World species accounts

(Billerman et al. 2020). We identified whether species engage in migratory or nomadic movements using data from the Australian Bird Database (Garnett et al. 2015). As an index for longevity, we used the maximum time elapsed between captures for each species, which is supplied by the ABBBS database of banding records (ABBBS 2020).

We characterized the habitat at each sampling location using a land cover model

based on Landsat imagery captured between 2010 and 2015 (DEWNR 2017). The model

has a resolution of 25 m, and classifies most land cover in the study area as either woody

or non-woody native vegetation (Fig. 1.1). At Brookfield, the former category

corresponds to mallee woodland, and the latter to chenopod scrub. To describe habitat

variability in the vicinity of sampling locations, we defined buffer zones 150 m in radius

around each capture site and calculated the proportion of mallee grid cells within the

resulting circles (Fig. 1.1). We selected this radius to approximate the home ranges of

captured individuals. Home range size data are not available for most species in the

sample, but prior data for splendid fairywrens (Malurus splendens) at the field site

indicate a home range size of 5–6 ha (equivalent in size to a buffer zone of radius 125–

140 m; Van Bael and Pruett-Jones 2000). Similar territory sizes have also been reported for breeding spiny-cheeked honeyeaters (Acanthagenys rufogularis; Reid 1984,

Rawsthorne et al. 2011). In a few instances (n = 19 locations, 35 birds), GPS data were not available for capture locations. For individuals captured at those sites, we used one of

two alternative methods for geocoding samples. For fairywrens, thornbills, and robins, territory maps compiled during focal follows and behavioral observations allowed us to locate the centroid of each social group’s breeding territory, and we used this location to calculate the habitat metric when available. For other species, we selected the approximate capture locations post hoc based on brief descriptions of the sites made at the time of sampling. We quantified land cover in R using the raster package (Hijmans

2020, R Core Team 2020).

Lab analysis. We extracted DNA from dried blood spots using DNeasy Blood &

Tissue Kits (Cat. No. 69506; QIAGEN Inc., Germantown, MD, USA). We screened extractions for apicomplexan DNA using the nonspecific primers MalMitoF1 and

MalMitoR1 (Eastwood et al. 2019; see Appendix D for detailed PCR methods). Positive samples were then barcoded for lineage identification using a nested PCR targeting a 580 bp region of the mitochondrial cytochrome b gene (Waldenström et al. 2004, Eastwood et al. 2019). This barcoding protocol amplifies Plasmodium, Haemoproteus, and

Leucocytozoon DNA. After each round of PCR, we ran products on agarose gels to

confirm amplification.

For each sample that tested positive on the nested PCR, we diluted the internal reaction product and performed an enzyme-based cleanup (70995.1.KT; Applied

Biosystems™ PCR Product Pre-Sequencing Kit; Thermo Fisher Scientific, Waltham,

MA, USA). Purified samples were then submitted to the University of Minnesota

Genomics Center (Minneapolis, MN, USA) for Sanger sequencing (Applied

Biosystems™ 3730xl DNA Analyzer; Thermo Fisher Scientific, Waltham, MA, USA).

Sequencing was carried out bidirectionally using the primers HaemNF and HaemNR2.

We identified each infection to the genus level by submitting the trimmed consensus sequences as BLASTn queries to the GenBank database.

Flock network. We constructed the mixed-species flock network with a gambit- of-the-group approach (Franks et al. 2010), using count data from the flock-affiliation observations to construct an association matrix. As an index of association between pairs of species, we calculated the scalar product of the two species’ counts in flocks, then performed a square-root transformation to scale the result (Equation 1).

= 𝑀𝑀 (1)

𝑥𝑥𝑖𝑖𝑖𝑖 � � 𝑛𝑛𝑖𝑖𝑖𝑖 𝑛𝑛𝑗𝑗𝑗𝑗 𝑚𝑚=1 In Equation 1, xij is the association between species i and species j, nim and njm are the

counts for species i and j, respectively, in flock m. M is the total number of flocks. This is

equivalent to multiplying the species-by-flock affiliation matrix with its transpose, then taking the elementwise square root of the product. This index extends the joint occurrence index (Whitehead 2008) to incorporate information about species counts. In

doing so, it captures aspects of both the frequency and the intensity (as measured by the

number of individuals of each species associating) of associations between species. We

did not perform any filtering of the association scores when constructing the network, as

our primary interest was in quantifying the frequency and intensity of species’ in-flock

encounters rather than in identifying preferential associations (Franks et al. 2010).

We summarized the frequency- and intensity-based dyadic association scores for each species by computing the scaled eigenvector centrality, using the association scores from the flock network as edge weights. The centrality of a node indicates its relative position within the network, with more central nodes tending broadly to have more and/or stronger connections to other nodes in the network. Eigenvector centrality in particular

measures both the first- and second-order connections of a node. Given our association index for defining the strength of connections, species with comparatively high eigenvector centrality in our flock network are expected to have more frequent or more intense direct or indirect contacts with comparatively more species. Thus, our metric of interspecific sociality (centrality) is based upon the hypothesis that the frequency and intensity of species interactions are important in determining how infection risk covaries with associations in flocks. All network analyses were performed in R using functions from the igraph and tidygraph packages (Csardi and Nepusz 2006, Pedersen 2020, R

Core Team 2020).

Statistical analysis. We calculated prevalence for each species and constructed

95% highest density credible intervals (95% HD CrI) using Jeffreys priors (Beta(0.5,

0.5)). We evaluated demographic, ecological, and environmental risk factors for infection using Bayesian phylogenetic multilevel regression. Initially, we planned to model

35 infection separately for each genus of parasite. However, we found no evidence of

Leucocytozoon infection among the birds in our sample, and a very low prevalence of

Plasmodium infections (0.4%; 4 positive samples out of 889 individuals tested). Thus, we only considered Haemoproteus infections for the modelling portion of this analysis. We excluded hatch-year birds from the model, as we expected that the social exposure risks comprising the primary focus of our study would be less relevant for nestlings and fledglings than for adults.

We considered the following variables as population (“fixed”) effects in the global model of infection status: sex, habitat (proportion of land cover by mallee near capture site), average conspecific group size, species longevity, modal foraging stratum, nest structure, eigenvector centrality, and migratory status. We included species as a group (“random”) intercept term, using phylogenetic correlation matrices to define relationships between species. Briefly, we sampled 100 trees from a published avian phylogeny (Jetz et al. 2012), pruned the trees to the set of focal species, and calculated the phylogenetic correlation matrix for each tree using the R package ape (Paradis and

Schliep 2019). We then fit a multilevel Bayesian regression model to each matrix before combining the posteriors. We performed prior checking, convergence testing, and posterior predictive checks on the global model as described in Appendix E. Models were specified using the brms package, with Markov-chain Monte Carlo sampling performed using rstan (Bürkner 2018, R Core Team 2020, Stan Development Team 2020).

For ease of interpretation, we describe the posterior densities of parameter estimates in terms of their implications for effect existence and significance (Makowski et al. 2019b). As an index of existence, the probability of direction (pd) quantifies the

certainty with which a predictor has a positive or negative (depending on the sign of the median posterior estimate) effect on the response. Values of pd approaching 1 indicate high certainty about the direction (i.e., the existence) of an effect, whereas pd near 0.5

suggests high uncertainty (indicating a posterior centered at 0). We further identify the

significance of an effect with the degree to which the full marginal distribution of the

parameter falls within a region of practical equivalence to zero, or ROPE (Makowski et

al. 2019b). We define the ROPE as the interval 0 ± 0.18, which is the default ROPE for logistic regression coefficients in the bayestestR package (Makowski et al. 2019a), and corresponds to a ~4% change in the predicted probability of infection for unit changes in the predictors. Overlap with the ROPE is affected by both the location and the scale of the posterior, such that a low overlap could result from either strong separation of the posterior from zero (i.e., high certainty of existence of an effect) or diffusion in the

posterior.

Results

We sampled 889 birds of 23 species from 9 families (see Appendix A for species

details). The sample included 298 adult females, 465 adult males, 66 hatch-year (nestling

or fledgling) females, and 60 hatch-year males. Six adult fairywrens (Malurus spp.) were sampled twice each (once early and once late in the field season), for a total of 895 blood samples. Both samples were tested in those cases, but only one conversion of infection status was detected (a five-year old male purple-backed fairywren (M. assimilis) was negative in late August, but a sample collected from the same bird in early November tested positive for Haemoproteus infection). This bird was considered positive for all subsequent analyses.

Of the 889 sampled individuals, we identified 73 birds (8.2%, 95% HD CrI [6.5%,

10.1%]) with Haemoproteus infections and 4 (0.4%, 95% HD CrI [0.1%, 1.0%]) with

Plasmodium infections. No Leucocytozoon infections were detected. Haemoproteus

prevalence varied substantially between species (Fig. 1.3, Appendix A). We found no

haemosporidian infections in 15 of the species sampled (404 individuals). Among

infected species, Haemoproteus prevalence ranged from 1.0% (95% HD CrI [0.002%,

3.8%]) in weebills (Smicrornis brevirostris) to 92.9% (95% HD CrI [75.2%, 100%]) in

spiny-cheeked honeyeaters (Acanthagenys rufogularis). All 77 infected birds were adults, with 28 infections (26 Haemoproteus, 2 Plasmodium) occurring in females and 47 (45

Haemoproteus, 2 Plasmodium) in males. We detected Haemoproteus infections in 8

species (4 families) and Plasmodium in 3 species (2 families). Although no

Haemoproteus/Plasmodium co-infections were found in individual birds, all Plasmodium

infections occurred in species in which other individuals were infected with

Haemoproteus. For further description and discussion of the diversity of blood parasite

lineages identified, refer to Chapter 2.

Figure 1.3 Prevalence of Haemoproteus spp. infections in 23 avian species at Brookfield Conservation Park, Australia, with posterior distributions calculated using Jeffreys priors (Beta(0.5, 0.5)) and shaded to indicate limits of the 95% highest density credible intervals. Overall prevalence was 8.2% (73 positive/889 birds tested), with considerable variation between species. The highest prevalences were found among honeyeaters (Meliphagidae spp.). No infections were detected in 15 species, although small sample sizes for several host species made precise estimation of prevalence difficult. We observed 481 flocks involving 39 species (Fig. 1.4). The mean flock size

(total number of participating individuals) was 7.75 (SD = 4.57, range = 2–36), and mean

richness (number of participating species) was 3.46 (SD = 1.74, range = 2–10).

Conspecific group size for each species was not different between flocking and non-

flocking contexts (all p > 0.05). However, foraging stratum was associated with flock

context for three species (weebill, Fisher’s exact p = 0.00108; purple-backed fairywren

[Malurus assimilis], p = 0.00164; chestnut-rumped thornbill [Acanthiza uropygialis], p =

0.0118). However, in the latter two cases, the overall modal foraging stratum was the

same as the modal foraging stratum in both flocking and non-flocking contexts. In the

case of the weebill, the overall mode (low canopy) was the same as that in flocking

contexts, but was tied with high canopy in non-flocking contexts. For the prevalence model, we used low canopy (overall and flocking mode) as the modal foraging stratum

for this species.

Figure 1.4 (a) Species connections in a foraging flock network. Edges between species were defined on the basis of co- occurrence in one or more mixed species flocks (n = 481), with edge weights determined by both the frequency of co- occurrence and the number of individuals of each species present when the two species co-occurred. For visual clarity, only the strongest 30% of connections are shown. (b) Joint and marginal distributions of size (total number of individuals) and species richness of mixed-species foraging flocks (points and tassels are jittered to avoid overplotting). (c) Node centrality of sampled avian hosts were used as an index of species’ degree of interspecific sociality. Species

with higher centrality scores have more numerous and/or stronger first- and second-order connections in the flock network (a). Species varied in the extent to which they participated in mixed-species foraging

flocks (Fig. 1.4). Among flocking species with blood samples collected, mean

unweighted degree (number of connections in the flock network) was 23.77 (SD = 7.03).

Chestnut-crowned babblers (Pomatostomus ruficeps) co-occurred with the fewest species

(8), and chestnut-rumped thornbills with the most (34). Chestnut-rumped thornbills also

had the greatest weighted eigenvector centrality (1.00), and singing honeyeaters

(Gavicalis virescens) had the lowest (0.09). Mean eigenvector centrality among sampled

species was 0.45 (SD = 0.28; Fig. 1.4c).

There was considerable uncertainty about the existence of most of the marginal

effects, with pd < 0.975 for all but two parameters (Fig. 1.5). The existence of effects of

group size and longevity was more certain. Haemoproteus infection probability decreased

with group size (pd = 1.000; median = –2.078, 95% HD CrI [–3.127, –1.047]) and

increased slightly with species longevity (pd = 0.994; median = 0.267, 95% HD CrI

[0.060, 0.486]; Fig. 1.6). The effect of group size can be considered significant (0% in

ROPE), but the significance of the longevity effect is less certain (20.5% in ROPE). In

addition, there were trends towards greater Haemoproteus prevalence in species with

higher centrality in the mixed-species flock network, migratory species, and birds caught

in habitats with more mallee coverage (Fig. 1.6). Prevalence was also highest on average

among species with open nests and those foraging in the high canopy stratum, but these

differences were slight.

Figure 1.5 Posterior densities for population effects on Haemoproteus infection probability in birds captured August– December, 2019, at Brookfield Conservation Park, South Australia. The tails of the distribution indicate the ranges of parameter estimates across 400,000 MCMC iterations. Points indicate the posterior median for each distribution, with error bars showing the 95% highest posterior density intervals. The region of practical equivalence to zero (ROPE) for logistic regression coefficients is delimited by dashed lines (0 ± 0.18).

Figure 1.6 Haemoproteus infection probability is slightly higher (a) among individuals occupying mallee-dominated habitat than those in scrub habitat, and (b) among more central species in the mixed-species flocking network. However, the 95% credible intervals for both effects broadly overlap zero. Haemoproteus prevalence (c) decreases with average conspecific group size and (d) increases with average lifespan. Model predictions for each covariate are based on the median expected response for females of open-nesting, non-migratory species at the median observed values of all other continuous covariates. Shaded regions indicate the 95% highest posterior density intervals about the median

predicted values. Points (b–d) and marginal tassels (a) indicate observed species’ prevalences and individual infection statuses, respectively.

Discussion

We found a low prevalence of Haemoproteus infection, and an extremely low

prevalence of Plasmodium infection, in an avian community inhabiting a semi-arid woodland in temperate South Australia. We also found substantial variation in blood parasite prevalence between host species, consistent with other studies in the region (e.g.,

Beadell et al. 2004, Zamora-Vilchis et al. 2012, Laurance et al. 2013). These results

highlight the importance of phylogenetically-correlated effects in shaping patterns of infection by haemosporidian parasites. However, most of the species-level behavioral and ecological variables we investigated were not able to efficiently describe this variation.

Sex, foraging height, nest structure, centrality in mixed-species flocks, and migratory status were not strongly related to the probability of Haemoproteus infection in the birds we sampled. We did detect effects of mean conspecific group size and species’ longevity

(as an index of immune function; Tella et al. 2002) on infection prevalence. These effects appear to be driven in part by high prevalence in three species of honeyeaters (Fig. 1.6; see also Appendix F for a Haemoproteus lineage-specific analysis in honeyeaters).

Previous studies have identified influences of host ecology on the probability of blood parasite infection, suggesting an important role for microenvironment and behaviors that govern exposure to vectors as key drivers of infection. Species using open cup nests had higher prevalence of Haemoproteus infection than did those nesting in closed cups in both Africa (Lutz et al. 2015) and the Neotropics (Fecchio et al. 2011,

Barrow et al. 2019; but see Fecchio et al. 2013, González et al. 2014). Rates of

Haemoproteus infection have also been associated with the height at which species nest

or forage (Fecchio et al. 2011, Laurance et al. 2013, González et al. 2014, Lutz et al.

2015; but see Fecchio et al. 2013, Svensson-Coelho et al. 2013, Barrow et al. 2019).

Contrasting with these earlier findings, we did not detect a strong effect of host ecology

(foraging stratum or nest structure) on Haemoproteus infection in our sample.

However, this does not imply that host ecology is unimportant in determining patterns of infection. Rather, it may be that other, unobserved aspects of host ecology are key drivers of exposure and infection in the study system. For instance, as many vectors are crepuscular or nocturnal (Mullen and Durden 2019), diurnal foraging activity may not represent a major exposure hazard for potential hosts. Nocturnal roosting behavior might

provide a better perspective on behavioral exposures, as it appears to do for West Nile

virus infection (Janousek et al. 2014). Unfortunately, little is known about the blood

parasite vectors at this site and their behavioral ecology.

A principal objective of our study was to determine whether interspecific social

interactions influence the probability of blood parasite infections. We expected that host

species exhibiting more or stronger connections to other species might exhibit a higher

prevalence of infection in this multi-host, multi-parasite community. Sociality in general,

and the transition between solitary and communal lifestyles, are associated with increased

rates of blood parasitism among closely related host species (Tella 2002), but the impacts

of interspecific social behavior are unclear. Participation in mixed-species flocks was

associated with relatively high rates of Plasmodium infections among a diverse

assemblage of birds from southeast Africa (Lutz et al. 2015), but had no effect on

Plasmodium prevalence among Neotropical species (González et al. 2014, Fecchio et al.

2017, Barrow et al. 2019). In contrast, species participating in mixed-species flocks had

lower Haemoproteus prevalence than did solitary species in Africa (Lutz et al. 2015),

while in the Neotropics the pattern was reversed (González et al. 2014, but see Barrow et

al. 2019). Using centrality in a mixed-species foraging flock network as an index for social interactions, we did not find any effect of interspecific sociality on Haemoproteus

prevalence. This suggests that associations between species do not drive patterns of infection in this system. In one important respect, the observation makes sense for

Haemoproteus, as these parasites tend to show comparatively strong patterns of host–

parasite association (Beadell et al. 2004, Clark et al. 2014). Indeed, each Haemoproteus

lineage we identified was restricted to a single host family within our sample (see

Chapter 2). Specificity of host–parasite relationships implies that host associations across

deep phylogenetic divisions should not directly affect infection probability, which is

consistent with our findings.

While interspecific social interactions appear to be unimportant in driving

Haemoproteus infection in this community, we did find evidence that intraspecific

associations are related to the prevalence of infection. Contrary to our expectations,

prevalence was lowest among species with larger group sizes. In the two most highly- parasitized families in our sample (Meliphagidae and Maluridae), species with smaller conspecific group sizes had the highest rates of infection. We suspected that larger groups might promote transmission, hypothesizing that the proximity of larger contingents of individuals might facilitate host searching by vectors. This would be consistent with the correlation between sociality and blood parasitism identified by Tella (2002). Fecchio et al. (2011, 2013) also reported higher rates of Haemoproteus infection among group- living species than among those living in pairs (but see González et al. 2014, Barrow et

al. 2019). That larger groups tended to have lower prevalence in our study could instead suggest an important role of social antiparasite behaviors (e.g., allopreening) in reducing transmission by limiting contacts between hosts and vectors, or of social benefits in effecting improved resistance to parasite infection.

Another possible explanation for the negative effect of group size on

Haemoproteus prevalence is that encounter-dilution effects reduce individual infection risk among aggregated individuals (Mooring and Hart 1992). This pattern of per-capita risk reduction is expected when both (1) host clustering reduces the rate of encounter by host-seeking parasites (encounter reduction), and (2) parasites do not increase attack rates in proportion to host aggregation size (dilution; Mooring and Hart 1992). However, in the case of Culicoides midges (the presumptive vectors of avian Haemoproteus parasites of host species in our study; Valkiūnas 2005), vector abundance does appear to scale with host abundance for small host aggregations (at least for mammalian hosts; Garcia-Saenz et al. 2011), suggesting that encounter-reduction may not occur in this system.

In addition to effects of the social and physical environment, physiological influences play a key role in determining host infection status. When immune function is costly, life history theory predicts trade-offs between immunological investment and other aspects of life history such as reproduction and sexual ornamentation (Norris and

Evans 2000). Such a trade-off has been hypothesized to account for interspecific variation in blood parasite prevalence among avian hosts (Ricklefs 1992). In the years since this hypothesis was proposed, evidence has accumulated to suggest that higher immune function among bird species is associated with longer developmental periods and greater longevity (Tella et al. 2002). The length of development, in turn, has been linked

to blood parasite prevalence, with more slowly-developing species tending to have lower rates of infection than species with shorter developmental periods (Ricklefs 1992, Tella et al. 1999). Despite the link between immune function and longevity, no clear connection between the prevalence of blood parasites and longevity among avian host species has been found (Tella et al. 2002, Svensson-Coelho et al. 2013). We used an index of longevity, the maximum elapsed time between captures for a species, to test whether

longevity is associated with Haemoproteus infection. We predicted that longer-lived

species would have lower infection prevalences, which could support the existence of a

connection between longevity and immune function (Tella et al. 2002). Although we did

identify a small effect of longevity on the prevalence of Haemoproteus infections among

the host species sampled, the direction of this effect was opposite to what we expected,

and seems inconsistent with such a relationship. Rather, longer-lived species (i.e., those

thought to have higher immune function; Tella et al. 2002) had greater prevalences of

Haemoproteus infection than did those with shorter lifespans. This pattern could be due

in part to the chronic nature of most haemosporidian infections in avian hosts (Valkiūnas

2005). As individuals seldom completely clear blood parasite infections, older age classes

tend to have higher prevalences of infection (Wood et al. 2007, Knowles et al. 2011,

Eastwood et al. 2019). Thus, if the average age of individuals of longer-lived species is

relatively high compared to individuals of shorter-lived species, and if there is no strong

effect of longevity per se on the rate of parasitism (but there is an effect of age), then we

might expect to see the observed correlation between prevalence and longevity at the

species level. We were not able to make a robust test for age effects in our sample, as

individuals of most species could not be aged precisely.

This study joins a growing number of avian blood parasite studies from Australia, contributing important baseline infection data for this under-surveyed region and improving the global perspective on haemosporidian diversity and distribution.

Australian birds are ecologically diverse, and the regional avifaunal assemblage exhibits a high degree of endemism. Globally, the diversity of haemosporidian blood parasites of birds is correlated with the diversity of their avian hosts (Clark et al. 2014). Nevertheless, comparatively little recent effort has been made to systematically catalog and characterize haemosporidian infections in Australian avifauna despite the more or less cosmopolitan distribution of blood parasites in subpolar regions (Clark et al. 2014). Of the more than 400 studies that have deposited 4,519 unique haemosporidian cytochrome b sequences in the MalAvi database as of June 2021, just 13 included specimens isolated from Australian birds, representing 103 unique parasite lineages (Bensch et al. 2009).

Thus, there is a substantial gap in the knowledge of haemosporidian infections from the birds of Australia. For many of the host species investigated, this study represents the first examination, to our knowledge, for blood parasite infections.

We found that Haemoproteus infections were predicted by intraspecific sociality, but not by interspecific social behavior, in a socially and ecologically diverse community of birds. This finding is perhaps best understood in light of the relatively strong phylogenetic associations between Haemoproteus parasites and their hosts.

Haemoproteus lineages tend to infect only members of comparatively closely related clades, both in our study system (see Chapter 2) and more broadly (e.g., Beadell et al.

2004). Species with relatively larger group sizes tended to have lower prevalences of

Haemoproteus infection, suggesting a possible role for conspecific social interactions or

50 encounter–dilution effects in reducing the risk of blood parasitism in this system. Further study is needed to identify the mechanism by which this reduction is achieved. Additional information on vector distribution, competence, and host preference at this site is also needed to disentangle host and vector effects on patterns of haemosporidian prevalence in the avian community.

CHAPTER 2

BLOOD PARASITE DIVERSITY IN AN UNDER-SURVEYED AVIFAUNA

Avian blood parasites (Apicomplexa: Haemosporidia) have attracted substantial

research attention since their discovery in the latter part of the nineteenth century. Their

global distribution, the taxonomic breadth of their hosts, and their impacts on wild and

domestic birds have made them an important focus of study for ecologists, wildlife

epidemiologists, and conservationists alike. In recent decades, advances in molecular

biology have facilitated a resurgence of field studies of blood parasite infections by

permitting rapid, simple, and inexpensive diagnosis of blood parasite infections (Bensch

et al. 2009). At the same time, these tools have fostered a new appreciation for the

diversity of blood parasites (Bensch et al. 2009, Clark et al. 2014).

Both globally and regionally, the diversity of haemosporidian parasites reflects that of their avian hosts (Clark et al. 2014, McNew et al. 2021). However, broader sampling of bioregions worldwide is needed for unbiased estimation of global parasite diversity patterns. Extensive work has been done to document the diversity of blood parasites, especially in north temperate regions and, more recently, the Neotropics.

However, relatively few recent broad-scale survey efforts have been conducted in

Australia (Beadell et al. 2004, Zamora-Vilchis et al. 2012, Laurance et al. 2013, Clark et al. 2015b), despite the ecological diversity of its endemic avifauna (Clark et al. 2014).

To contribute to the understanding of global haemosporidian diversity, we surveyed avian blood parasites from a community of birds in this poorly-sampled region.

The survey was conducted as part of a larger study relating prevalence of infection in a

52 diverse host assemblage to host behavioral and ecological traits (see Chapter 1). Here, we describe the genetic (lineage) diversity of haemosporidian parasites identified in the survey, and relate it to regional and global patterns.

Materials and methods

We sampled wild birds during the 2019 breeding season (August–December) at

Brookfield Conservation Park, South Australia (−34.35°, 139.50°). Adult and juvenile birds were captured in mist nets using a combination of flushing and audio playback, and a small (5–25 µL) blood sample was taken by brachial venipuncture. Nestlings of two species of fairywren (Maluridae: Malurus spp.) were also sampled at 3–7 d as part of routine monitoring for those species. Samples were preserved as dried blood spots on

Whatman® FTA® classic cards (GE Healthcare, Chicago, IL, USA) for later analysis.

Birds were aged (and sexed, when possible) in the field using characteristics of plumage and soft parts as described by Hardy (2019). For all hatch-year birds and for adults of monomorphic species, we determined the molecular sex using standard PCR protocols

(Griffiths et al. 1998, Kahn et al. 1998, Lee et al. 2010; see Appendix D).

We extracted DNA from blood spots with DNeasy Blood & Tissue Kits (Cat. No.

69506; QIAGEN Inc., Germantown, MD, USA). To identify blood parasite infections, we used the haemosporidian screening and barcoding protocols described by Eastwood et al. (2019). This procedure amplifies a region of the mitochondrial cytochrome b gene used to identify haemosporidian lineages (Bensch et al. 2009). Briefly, we screened for apicomplexan DNA by PCR with the primers MalMitoF1 and MalMitoR1, then checked for amplification by electrophoresis on 2% agarose gels. Samples that tested positive in the initial screening were then barcoded for lineage identification by a nested PCR using

external primers Prim3_F2/Prim3_R1 and internal primers HaemNF/HaemNR2 (see

Appendix D for additional details regarding diagnostic procedures). Products from the

nested reaction were submitted to the University of Minnesota Genomics Center for

bidirectional sequencing.

Sequencing reads were trimmed and aligned in R using the packages msa,

DECIPHER, and sangerseqR (Hill et al. 2014, Bodenhofer et al. 2015, Wright 2016, R

Core Team 2020). We aligned forward and reverse reads using the Muscle algorithm,

then identified consensus sequences for each sample. We aligned the consensus

sequences with one another using the Muscle algorithm and merged this with the lineage

alignment from the MalAvi database, which we accessed with malaviR (Ellis et al. 2021).

The aligned sample sequences were trimmed to the 479 bp lineage barcoding region and

all base insertions were removed. Sequences with low coverage of the barcoding region

were excluded from further analyses. We considered those that were identical except for

a few ambiguous basecalls (e.g., a thymine residue in one sequence matched to an

uncertain pyrimidine in another) to represent the same lineage. To identify lineages, we

performed BLAST searches of the MalAvi and GenBank databases using the aligned and

trimmed consensus sequences. Only unique lineages were retained for subsequent

phylogenetic analyses. Sequences for novel lineages were submitted to both the MalAvi

and GenBank databases (GenBank accession numbers: MW888415–MW888421;

MalAvi lineage names given in Table 2.1).

We estimated relationships among haemosporidian lineages using a maximum

likelihood approach in R with the packages phangorn and ape (Schliep 2011, Paradis and

Schliep 2019). First, we estimated a local phylogeny using all lineages (Plasmodium and

Haemoproteus) identified in this study. We also estimated separate regional phylogenies for all Australian lineages of Plasmodium and Haemoproteus, respectively, that have

sequences in the MalAvi database. We used a Leucocytozoon lineage (PTIVIC03) as an outgroup for all three phylogenies. For each lineage subset, we used AICc-based selection to identify the best nucleotide substitution model and compared the resulting maximum likelihood estimates to 1000 bootstrap replicates. A generalized time-reversible (GTR) nucleotide substitution model was selected for all three phylogenies, with gamma- distributed rate variation among sites (i.e. GTR + Γ, all phylogenies) and invariant sites

(i.e. GTR + Γ + I, regional phylogenies only). When comparing individual lineages, we

computed the pairwise genetic distances (dTN) between sequences based on Tamura and

Nei’s (1993) model of base substitution.

Results

We sampled 889 individuals of 23 species and found 77 cases of haemosporidian infection in 8 species (see Appendix A for a list of species sampled). We identified 2

Plasmodium and 7 Haemoproteus lineages (Table 2.1). Five of the Haemoproteus lineages were novel. No cases of Leucocytozoon infection were detected. The two

Plasmodium lineages were very similar to one another (dTN = 0.2%), and matched

lineages previously identified in New Zealand. Three honeyeaters (Meliphagidae spp.)

were infected by the MYZCAL02 Plasmodium lineage, which was previously found in two species of honeyeaters (Olsson-Pons et al. 2015), and a splendid fairywren

(Maluridae: Malurus splendens) was infected by a lineage previously isolated from

parakeets (CYNOV1; Ortiz-Catedral et al. 2019). While the two Plasmodium lineages we identified did separate at the level of host family (Meliphagidae and Maluridae), they

55 appear to be comparatively similar in the broader context of Australian Plasmodium lineages as a whole (Fig. 2.1).

Table 2.1 Haemosporidian lineages identified in birds sampled at Brookfield Conservation Park, Australia. Lineages CYNOV1 and MYZCAL02 correspond to Plasmodium; all others represent Haemoproteus infections. Lineage counts include four cases of multiple infection: three ACARUF01/CLIPIC01 co-infections and one GAVVIR01/GAVVIR02 co-infection. In addition, one Malurus assimilis and one Ptilotula ornata had Haemoproteus infections that could not be properly characterized due to poor sequence quality.

Species N Infected CYNOV1 MYZCAL02 POMSUP01 SMIBRE01 CLIPIC02 GAVVIR01 GAVVIR02 CLIPIC ACARUF01 Malurus assimilis 119 12 . . . . 11 . . . . Malurus splendens 135 15 1 . . . 14 . . . . Acanthagenys rufogularis 14 13 ...... 9 7 Gavicalis virescens 9 8 . . . . . 3 3 3 . Ptilotula ornata 38 23 . 1 . . . . . 21 . Melithreptus brevirostris 30 3 . 2 . . . . . 1 . Smicrornis brevirostris 101 1 . . . 1 . . . . . Pomatostomus superciliosus 39 2 . . 2 ......

Figure 2.1 Maximum likelihood phylogenies for (a) locally-identified haemosporidian lineages and for all (b) Haemoproteus and (c) Plasmodium lineages identified in Australia. A Leucocytozoon lineage (PTIVIC03) was used as an outgroup for each reconstruction. Colored points indicate host families in which lineages were detected in the current study. Each lineage was isolated from a single host family. Trees are based on a 479-bp region of the mitochondrial cytochrome b gene, and were estimated using a generalized time-reversible nucleotide substitution model with gamma-distributed rate variation among sites (a, b, c) and invariant sites (b, c). Annotations in (a) indicate branch support from 1000 bootstrap replicate trees.

The Haemoproteus lineages that we detected varied across host species (Table

2.1), and our phylogenetic analysis indicated that lineages clustered at the avian host

family level (Fig. 2.1). The two most commonly-detected lineages were known from previous studies in Australia, although in both cases the earlier sequences were incomplete in the lineage-barcoding region (and thus, polymorphisms may exist in the un-matched regions). The most common infection of honeyeaters in our system

(CLIPIC01) was isolated from pardalotes (Pardalotidae: Pardalotus spp.) and a brown

treecreeper (Climacteridae: Climacteris picumnus) in Victoria (Balasubramaniam et al.

2013). All Haemoproteus infections of fairywrens in our system matched a lineage

(CLIPIC02) previously identified in a brown treecreeper in Victoria (Balasubramaniam et

al. 2013) as well as two fairywrens (Malurus coronatus and M. melanocephalus) in

Western Australia (Eastwood et al. 2019). We could not confidently characterize

Haemoproteus infections from one purple-backed fairywren (M. assimilis) and one yellow-plumed honeyeater (Ptilotula ornata), as the sequences were of poor quality and had several gaps. The fairywren infection was most similar to the Haemoproteus lineage infecting fairywrens (CLIPIC02; dTN = 6.0%), while the honeyeater infection was similar

to other lineages infecting honeyeaters (ACARUF01, CLIPIC01; both dTN < 2.0%).

Haemoproteus infections were detected throughout the sampling period, and for the two lineages identified in more than 10 individuals, the prevalence of infection appeared relatively consistent over time (Fig. 2.2).

Figure 2.2 Time series of weekly sampling effort (gray bars) and Haemoproteus-positive birds (colored bars). Counts are given separately for each Haemoproteus lineage, and are made on the basis of the host family in which each was identified. Only samples from adults (n = 763) were included. In the two lineages identified in more than 10 individuals (CLIPIC01 and CLIPIC02), prevalence remained fairly consistent throughout the sampling period, considering variation in effort.

We found evidence for multiple Haemoproteus lineage infections in four birds.

Electropherograms from three spiny-cheeked honeyeaters (Acanthagenys rufogularis) and one singing honeyeater (Gavicalis virescens) showed double peaks in the lineage- barcoding region, suggesting co-infection. We were able to resolve the double peaks based on corresponding single infections in other individuals. All three spiny-cheeked honeyeaters were infected with CLIPIC01 and ACARUF01, whereas the singing honeyeater was infected with GAVVIR01 and GAVVIR02. We did not detect any mixed-genus (Haemoproteus/Plasmodium) infections.

Discussion

Lineage richness in our sample reflects regional and global patterns. To date, nearly 1,800 Haemoproteus lineages have been identified globally, compared to just over

1,400 Plasmodium lineages. In Australia, researchers have isolated 72 Haemoproteus lineages and 32 Plasmodium lineages (including new lineages identified in this study).

The difference in richness between the two parasite genera may in part reflect greater host-specificity among Haemoproteus spp. (Beadell et al. 2004), suggesting a pattern of host–parasite co-diversification. Whereas Haemoproteus lineages regularly seem to be restricted to taxonomic families or closely-related clades within a host community,

Plasmodium lineages appear less constrained by host phylogeny, more readily infecting species from different families. This pattern was evident in our sample for

Haemoproteus, with lineages clustering according to the taxonomic family of their hosts

(Fig. 2.1), consistent with the idea that these parasites tend to be host specific.

These results help explain the lack of effect of interspecific sociality on blood parasite prevalence (Chapter 1). That lineages cluster at the host family level implies that

60 phylogenetically-correlated barriers to transmission are important in shaping cross- community trends in Haemoproteus infection. Yet whether these barriers reflect host– parasite, host–vector, or parasite–vector interactions remains unclear (see Martínez-de la

Puente et al. 2011). Further study of vector distribution and ecology, host preference, and vector competence for diverse parasite lineages is needed to fully resolve this issue.

The extremely low prevalence of certain lineages also raises intriguing questions.

Despite sampling over 300 thornbills, weebills, and whitefaces (Acanthizidae spp.), we detected only one infected individual (a weebill, Smicrornis brevirostris). No other host family at the site was as well sampled, nor was any represented by as low a prevalence

(with the exception of families for which all individuals were negative). How this lineage is maintained in the population is a mystery, although there are several potential explanations for the low prevalence of SMIBRE01. First, it’s possible that parasitized individuals clear infection rapidly enough to avoid detection. However, this seems unlikely, as haemosporidioses of birds tend to manifest as lifelong infections that may exhibit periodic cycles of parasitemia but from which individuals seldom make full recoveries (Valkiūnas 2005, Eastwood et al. 2019, but see Knowles et al. 2011).

Moreover, both diagnosis and transmission rely on the presence of blood-stage parasites, such that infections which are cleared too rapidly to detect should also have a comparatively low probability of transmission. Second, it may be that most infections in the population were in a latent state at the time of sampling, with agamic parasite stages present in the fixed tissues but with little to no parasitemia. This too seems unlikely, since we sampled birds before and throughout the breeding season, and recrudescence of latent infections is frequently associated with the onset of reproduction (Valkiūnas 2005).

Third, SMIBRE01 may represent a novel lineage that is newly arrived in the community, or one that is in decline due to evolved host defenses or shifts in vector distribution or

behavior. If a novel lineage, it seems improbable that SMIBRE01 was introduced by the

host in which we detected it. Weebills are nonmigratory, and banded individuals typically

move no more than a few kilometers between captures (average distance = 1 km,

maximum distance = 9 km; ABBBS 2020, Billerman et al. 2020). Thus, the infected bird

probably originated in or near the study site. Finally, the infection may represent an

abortive infection (i.e., one incapable of effecting further transmission). Current PCR- based diagnostics cannot distinguish between parasite life stages, and are thus incapable of establishing whether an infection has the potential for transmission (Valkiūnas et al.

2014). However, two observations support the hypothesis that the infection is non- abortive. First, we found no other obvious cases of apparently abortive (i.e., non-specific) infections involving the other lineages in our sample. If such infections were common, and commonly detected, we might expect to see (for example) cases of common infections of honeyeaters (e.g., CLIPIC01, ACARUF01) infecting non-specific host species (e.g., fairywrens). Yet as noted previously, each lineage was found in a single host family. Second, SMIBRE01 was unique among all lineages detected in our sample

(Fig. 2.2), and the two Australian Haemoproteus lineages to which it is most closely related (SERCIT02 and GERPAL02, Fig. 2.2) were also isolated from members of the same host family (Sericornis spp. and Gerygone spp.).

Parasite interactions with host immunity can play an important role in the host’s susceptibility to secondary infections, and can also mediate the virulence of multiple infections (Palinauskas et al. 2018). Thus, associations between parasites in a host

62 community may be indications of facilitation or of competition between parasites

(Palinauskas et al. 2011, Clark et al. 2016). Positive associations, which could indicate parasite–parasite facilitation, manifest as a relatively high frequency of multiple simultaneous infections within individual hosts. Four honeyeaters in our sample were co- infected with two Haemoproteus lineages each, which could indicate such a facilitative interaction involving these lineages. However, a much larger sample is needed to establish this with certainty. Additional sampling in this system would help to address this question. The true prevalence of co-infections may be higher, as PCR-based detection methods underestimate mixed haemosporidian infections (Valkiūnas et al.

2006, 2014), an effect thought to occur due to unequal PCR primer-binding affinity among different parasites.

Our results provide a small but important piece of a global puzzle depicting the distribution and diversity of avian haemosporidian parasites. We contribute critical baseline information from a region that has been relatively under-surveyed for blood parasites, including the identification of several novel parasite lineages. This study also highlights the need for further information about vectors, their distribution, and their ecological and coevolutionary interactions with both hosts and haemosporidian parasites.

CONCLUSION

This study provides baseline data on haemosporidian infections from a

comparatively undersampled avifauna, and also provides insight into the drivers of

variation in haemosporidian prevalence among avian host species. In Chapter 1, we found

that species’ longevity and conspecific group size (a measure of intraspecific sociality),

but not centrality in a mixed-species flocking network (a measure of interspecific

sociality) predicted Haemoproteus prevalence in a diverse avian community. In Chapter

2, we identified a key reason for the absence of an effect of interspecific sociality on

Haemoproteus prevalence among species. Our analysis revealed strong patterns of host–

parasite association, suggesting that individual parasite lineages are not transmitted

across relatively deep (i.e., family-level) divisions in the host phylogeny. However, the

factors maintaining those associations remain unclear. We identified seven novel

Haemoproteus cytochrome b lineages, and also found previously undocumented host–

parasite associations involving two Plasmodium lineages (Chapter 2). These data contribute to a clearer understanding of global haemosporidian diversity and distribution, and can further improve our understanding of coevolutionary patterns between birds and their haemosporidian parasites.

Our study also lays the conceptual groundwork for a more nuanced approach to investigating relationships between haemosporidian infections and interspecific social interactions. Network analyses are becoming increasingly popular tools to understand complex patterns of parasite and pathogen transmission, especially as advances in animal-tracking technology yield more and higher-resolution data on animal contacts.

But comparatively few network-based studies directly address the impacts of interspecific associations on transmission, and we are aware of none involving avian blood parasites.

Although we found no evidence for a connection between interspecific foraging associations and Haemoproteus infection, investigations of the more host-generalist

Plasmodium parasites might be better suited to this approach. Additionally, other types of contacts than foraging associations may prove better at capturing variation in behaviors that are important in generating variation in transmission risk between host species. The approach presented here could be adapted to accommodate data on spatial overlap networks, roosting networks, dispersal networks, or others. Additionally, if sampled individuals are uniquely marked and followed, the approach could be used to infer effects of individual heterogeneity in social behavior and ecology.

Although the host specificity of Haemoproteus parasites in this system presents a challenge for investigating impacts of interspecific host interactions on infection risk, it opens the door for a more targeted approach to understanding other host effects on infection. We suggest that future research in this system could take advantage of this specificity by focusing sampling effort within a single host family to investigate ecological and behavioral drivers of Haemoproteus infection without need for concern about reservoirs or transmission from unsampled host species. This approach would further facilitate inferences about individual- and group-level effects on infection. The host families Maluridae and Meliphagidae are likely the best candidates for this approach in this system, as members of these families showed the highest prevalences of infection and (in the case of the Meliphagidae) the most variation between sampled species.

Meliphagidae is an ecologically rich family, with species exhibiting considerable

variation in social behavior, breeding strategies, diet, and morphology. Members of the

Maluridae are cooperative breeders, but show considerable variation both within and

between species with respect to the sizes of family groups and territories. Moreover, two

of the malurid host populations investigated here (splendid and purple-backed fairywrens,

Malurus splendens and M. assimilis, respectively) have been the focus of intensive research for over a decade. The ongoing, long-term study of social behavior and breeding biology in these species has yielded a wealth of both archived blood samples and data on group size, genetic relations between individuals, reproductive success, and territory size

and location. This information could be used to identify potential fitness consequences of

infection, to reveal temporal dynamics of infection in the Brookfield system, and to

further investigate the effects of intraspecific sociality on Haemoproteus infection.

A complete understanding of the ecology and epidemiology of any vector-borne parasite relies upon thorough knowledge of both host and vector ecology. We provide a critical first look at host–parasite relationships in this system, but future research would benefit from additional information about the identity and distribution of vectors involved in haemosporidian transmission at Brookfield Conservation Park, their general ecology, and their associations with avian hosts and haemosporidian parasites. Such knowledge would likely help to clarify the nature of the host–parasite relationships we identified in

this study. Furthermore, much remains to be learned about the interactions between hosts

and parasites in this system. Future studies could incorporate investigations of infection

intensity among hosts as a means of more precisely assessing transmission potential. The

archived fairywren blood samples from this system also provide a unique opportunity to

gain insight into the temporal dynamics of haemosporidian infections in these two species

66 and into infection histories in individual hosts. Australia in general, and the Brookfield system in particular, offer exciting avenues for building a better understanding of the complicated and dynamic relationships between hosts, vectors, parasites, and their environments, and we hope that this research is but the first in what promises to be a fruitful series of studies.

REFERENCES

Altizer S, Bartel R, Han BA (2011) Animal migration and infectious disease risk. Science

331:296–302. doi: 10.1126/science.1194694

Ashford RW (1971) Blood parasites and migratory fat at Lake Chad. Ibis 113:100–1.

Atkinson CT, Forrester DJ (1987) Myopathy associated with megaloschizonts of

Haemoproteus meleagridis in a wild turkey from Florida. J Wildl Dis 23:495–8.

doi: 10.7589/0090-3558-23.3.495

Atkinson CT, van Riper III C (1991) Pathogenicity and epizootiology of avian

haematozoa: Plasmodium, Leucocytozoon, and Haemoproteus. Pages 19–48 in

Bird–parasite interactions: ecology, evolution, and behaviour (Loye JE, Zuk M,

Editors). Oxford University Press, Oxford, United Kingdom.

Atkinson CT, Thomas NJ, Hunter DB (Editors) (2008) Parasitic diseases of wild birds.

Wiley-Blackwell, Ames, Iowa, USA.

Australian Bird and Bat Banding Scheme [ABBBS] (2020) Bird and bat banding

database. Department of Agriculture, Water and the Environment, Government of

Australia, Canberra, Australia. https://www.environment.gov.au/topics/science-

and-research/bird-and-bat-banding/banding-data/search-abbbs-database

Balasubramaniam S, Mulder RA, Sunnucks P, Pavlova A, Amos JN, Melville J (2013)

Prevalence and diversity of avian haematozoa in three species of Australian

passerine. Emu 113:353–8. doi: 10.1071/MU13012

Barrow LN, McNew SM, Mitchell N, Galen SC, Lutz HL, Skeen H, Valqui T, Weckstein

JD, Witt CC (2019) Deeply-conserved susceptibility in a multi-host, multi-

parasite system. Ecol Lett 22:987–98. doi: 10.1111/ele.13263

Beadell JS, Gering E, Austin J, Dumbacher JP, Peirce MA, Pratt TK, Atkinson CT,

Fleischer RC (2004) Prevalence and differential host-specificity of two avian

blood parasite genera in the Australo-Papuan region. Mol Ecol 13:3829–44. doi:

10.1111/j.1365-294X.2004.02363.x

Becker DJ, Ketterson ED, Hall RJ (2020) Reactivation of latent infections with migration

shapes population-level disease dynamics. P Roy Soc Lond B Bio 287:20201829.

doi: 10.1098/rspb.2020.1829

Behdenna A, Lembo T, Calatayud O, Cleaveland S, Halliday JEB, Packer C, Lankester F,

Hampson K, Craft ME, Czupryna A, Dobson AP, Dubovi EJ, Ernest E,

Fyumagwa R, Hopcraft JGC, Mentzel C, Mzimbiri I, Sutton D, Willett B, Haydon

DT, Viana M (2019) Transmission ecology of canine parvovirus in a multi-host,

multi-pathogen system. P Roy Soc Lond B Bio 286:20182772. doi:

10.1098/rspb.2018.2772

Bennett GF, Fallis AM (1960) Blood parasites of birds in Algonquin Park, Canada, and a

discussion of their transmission. Can J Zool 38:261–73.

Bennett GF, Mead CJ, Barnett SF (1974) Blood parasites of birds handled for ringing in

England and Wales. Ibis 117:232–5.

Bennett GF, Caines JR, Bishop MA (1988) Influence of blood parasites on the body mass

of passeriform birds. J Wildl Dis 24:339–43.

Bennett GF, Peirce MA, Ashford RW (1993) Avian haematozoa: mortality and

pathogenicity. J Nat Hist 27:993–1001.

Bensch S, Hellgren O, Pérez-Tris J (2009) MalAvi: a public database of malaria parasites

and related haemosporidians in avian hosts based on mitochondrial cytochrome b

lineages. Mol Ecol Resources 9:1353–8. doi: 10.1111/j.1755-0998.2009.02692.x

Berger L, Speare R, Daszak P, Green DE, Cunningham AA, Goggin CL, Slocombe R,

Ragan MA, Hyatt AD, McDonald KR, Hines HB, Lips KR, Marantelli G, Parkes

H (1998) Chytridiomycosis causes amphibian mortality associated with

population declines in the rain forests of Australia and Central America. P Natl

Acad Sci USA 95:9031–6.

Billerman SM, Keeney BK, Rodewald PG, Schulenberg TS (Editors) (2020) Birds of the

world. Cornell Laboratory of Ornithology, Ithaca, New York, USA.

https://birdsoftheworld-org.proxy.birdsoftheworld.org/bow/home

Bodenhofer U, Bonatesta E, Horejš-Keinrath C, Hochreiter S (2015) msa: an R package

for multiple sequence alignment. Bioinformatics 31:3997–9. doi:

10.1093/bioinformatics.btv494

Brown CR, Bomberger Brown M (1986) Ectoparasitism as a cost of coloniality in cliff

swallows (Hirundo pyrrhonota). Ecol 67:1206–18.

Brown CR, Komar N, Quick SB, Sethi RA, Panella NA, Bomberger Brown M, Pfeffer M

(2001) Arbovirus infection increases with group size. P Roy Soc Lond B Bio

268:1833–40.

Bureau of Meteorology [BOM] (2005) Climate classification of Australia. Bureau of

Meteorology, Government of Australia. Melbourne, Victoria, Australia.

http://www.bom.gov.au/jsp/ncc/climate_averages/climate-

classifications/index.jsp?maptype=kpn#maps (accessed 15 May 2020).

Bürkner P-C (2018) Advanced Bayesian multilevel modeling with the R package brms. R

Journal 10:395–411. doi: 10.32614/RJ-2018-017

Clark NJ, Clegg SM (2015) The influence of vagrant hosts and weather patterns on the

colonization and persistence of blood parasites in an island bird. J Biogeogr

42:641–51. doi: 10.1111/jbi.12454

Clark NJ, Clegg SM, Lima MR (2014) A review of global diversity in avian

haemosporidians (Plasmodium and Haemoproteus: Haemosporida): new insights

from molecular data. Int J Parasitol 44:329–39. doi: 10.1016/j.ijpara.2014.01.004

Clark NJ, Adlard RD, Clegg SM (2015a) Molecular and morphological characterization

of Haemoproteus (Parahaemoproteus) ptilotis, a parasite infecting Australian

honeyeaters (Meliphagidae), with remarks on prevalence and potential cryptic

speciation. Parasitol Res 114:1921–8. doi: 10.1007/s00436-015-4380-8

Clark NJ, Olsson-Pons S, Ishtiaq F, Clegg SM (2015b) Specialist enemies, generalist

weapons and the potential spread of exotic pathogens: malaria parasites in a

highly invasive bird. Int J Parasitol 45:891–9. doi: 10.1016/j.ijpara.2015.08.008

Clark NJ, Wells K, Dmitrov D, Clegg SM (2016) Co-infections and environmental

conditions drive the distributions of blood parasites in birds. J Anim Ecol

85:1461–70. doi: 10.1111/1365-2656.12578

Clark NJ, Clegg SM, Sam K, Goulding W, Koane B, Wells K (2018) Climate, host

phylogeny and the connectivity of host communities govern regional parasite

assembly. Div Dist 24:13–23.

Cleland JB (1915) The haematozoa of Australian birds. Number 3. Trans Proc R Soc

South Australia 39:25–37.

Cleland JB (1922) The parasites of Australian birds. Trans Proc R Soc South Australia

46:85–118.

Cleland JB, Johnston TH (1909) Description of new haemoprotozoa from birds in New

South Wales, with a note on the resemblance between the spermatozoa of certain

honeyeaters (fam. Meliphagidae) and spirochaete-trypanosomes. J Proc R Soc

New South Wales 43:75–96.

Cleland JB, Johnston TH (1910) The haematozoa of Australian birds. Number 1. Trans

Proc Rep R Soc South Australia 34:100–14.

Cleland JB, Johnston TH (1911) The haematozoa of Australian birds. Number 2. J Proc

R Soc New South Wales 45:415–44.

Colombelli-Négrel D, Kleindorfer S (2008) In superb fairy wrens (Malurus cyaneus),

nuptial males have more blood parasites and higher haemoglobin concentration

than eclipsed males. Aust J Zool 56:117–21. doi: 10.1071/ZO07072

Craft ME, Hawthorne PL, Packer C, Dobson AP (2008) Dynamics of a multihost

pathogen in a carnivore community. J Anim Ecol 77:1257–64. doi:

10.1111/j.1365-2656.2008.01410.x

Craft ME, Volz E, Packer C, Meyers LA (2009) Distinguishing epidemic waves from

disease spillover in a wildlife population. P Roy Soc Lond B Bio 276:1777–85.

doi: 10.1098/rspb.2008.1636

Craft ME, Volz E, Packer C, Meyers LA (2011) Disease transmission in territorial

populations: the small-world network of Serengeti lions. J R Soc Interface 8:776–

86. doi: 10.1098/rsif.2010.0511

Csardi G, Nepusz T (2006) The igraph software package for complex network research.

InterJournal CX.18:1695.

Department of Environment, Water and Natural Resources [DEWNR] (2017) SA land

cover models (2010-2015). Government of South Australia, Adelaide, South

Australia. https://data.sa.gov.au/data/dataset/sa-land-cover (accessed 17 May

2020).

Eastwood JR, Peacock L, Hall ML, Roast M, Murphy SA, Gonçalves da Silva A, Peters

A (2019) Persistent low avian malaria in a tropical species despite high

community prevalence. Int J Parasitol: Parasites Wildl 8:88–93. doi:

10.1016/j.ijppaw.2019.01.001

Ellis VA, Bensch S, Canback B (2021) malaviR: an R interface to MalAvi. Version 0.2.0.

https://github.com/vincenzoaellis/malaviR

Epstein JH, Price JT (2009) The significant but understudied impact of pathogen

transmission from humans to animals. Mt Sinai J Med 76:448–55. doi:

10.1002/msj.20140

Fecchio A, Lima MR, Silveira P, Braga ÉM, Marini MÂ (2011) High prevalence of

blood parasites in social birds from a Neotropical savanna in Brazil. Emu

111:132–8. doi: 10.1071/MU10063

Fecchio A, Lima MR, Svensson-Coelho M, Marini MÂ, Ricklefs RE (2013) Structure

and organization of an avian haemosporidian assemblage in a Neotropical

savanna in Brazil. Parasitol 140:181–92. doi: 10.1017/S0031182012001412

Fecchio A, Ellis VA, Bell JA, Andretti CB, d’Horta FM, Silva AM, Tkach VV,

Weckstein JD (2017) Avian malaria, ecological host traits and mosquito

abundance in southeastern Amazonia. Parasitol 144:1117–32. doi:

10.1017/S003118201700035X

Franks DW, Ruxton GD, James R (2010) Sampling animal association networks with the

gambit of the group. Behav Ecol Sociobiol 64:493–503. doi: 10.1007/s00265-009-

0865-8

Gallana M, Ryser-Degiorgis M-P, Wahli T, Segner H (2013) Climate change and

infectious diseases of wildlife: altered interactions between pathogens, vectors

and hosts. Curr Zool 59:427–37. doi: 10.1093/czoolo.59.3.427

Garcia-Saenz A, McCarter P, Baylis M (2011) The influence of host number on the

attraction of biting midges, Culicoides spp., to light traps. Med Vet Entomol

25:113–5. doi: 10.1111/j.1365-2915.2010.00904.x

Garamszegi LZ (2011) Climate change increases the risk of malaria in birds. Global

Change Biol 17:1751–9.

Garnett ST, Duursma DE, Ehmke G, Guay P-J, Stewart A, Szabo JK, Weston MA,

Bennett S, Crowley GM, Drynan D, Dutson G, Fitzherbert K, Franklin DC (2015)

Biological, ecological, conservation and legal information for all species and

subspecies of Australian bird. Sci Data 2:150061. doi: 10.1038/sdata.2015.61

Garvin MC, Remsen Jr. JV (1997) An alternative hypothesis for heavier parasite loads of

brightly colored birds: exposure at the nest. Auk 114:179–91.

González AD, Matta NE, Ellis VA, Miller ET, Ricklefs RE, Gutiérrez HR (2014) Mixed

species flock, nest height, and elevation partially explain avian haemoparasite

prevalence in Colombia. PLoS ONE 9:e100695. doi:

10.1371/journal.pone.0100695

Griffiths R, Double MC, Orr K, Dawson RJG (1998) A DNA test to sex most birds. Mol

Ecol 7:1071–5.

Hamede RK, Bashford J, McCallum H, Jones M (2009) Contact networks in a wild

Tasmanian devil (Sarcophilus harrisii) population: using social network analysis

to reveal seasonal variability in social behaviour and its implications for

transmission of devil facial tumour disease. Ecol Lett 12:1147–57. doi:

10.1111/j.1461-0248.2009.01370.x

Hardy JW (Compiler) (2019) Bird in the hand. Second Edition. Australian Bird Study

Association Inc. absa.asn.au

Härkönen T, Dietz R, Reijnders P, Teilmann J, Harding K, Hall A, Brasseur S, Siebert U,

Goodman SJ, Jepson PD, Dau Rasmussen T, Thompson P (2006) The 1988 and

2002 phocine distemper virus epidemics in European harbour seals. Dis Aquat

Organ 68:115–30. doi: 10.3354/dao068115

Hellard E, Cumming GS, Caron A, Coe E, Peters JL (2016) Testing epidemiological

functional groups as predictors of avian haemosporidia patterns in southern

Africa. Ecosphere 7:e01225. doi: 10.1002/ecs2.1225

Hijmans RJ (2020) raster: geographic data analysis and modeling. Version 3.4-5.

https://CRAN.R-project.org/package=raster

Hill JT, Demarest BL, Bisgrove BW, Su Y-C, Smith M, Yost HJ (2014) Poly peak

parser: method and software for identification of unknown indels using Sanger

sequencing of polymerase chain reaction products. Dev Dyn 243:1632–6. doi:

10.1002/dvdy.24183

Ilgūnas M, Bukauskaitė D, Palinauskas V, Iezhova TA, Dinhopl N, Nedorost N,

Weissenbacher-Lang C, Weissenböck H, Valkiūnas G (2016) Mortality and

pathology in birds due to Plasmodium (Giovannolaia) homocircumflexum

infection, with emphasis on the exoerythrocytic development of avian malaria

parasites. Malar J 15:256. doi: 10.1186/s12936-016-1310-x

Ishtiaq F, Gering E, Rappole JH, Rahmani AR, Jhala YV, Dove CJ, Milensky C, Olson

SL, Peirce MA, Fleischer RC (2007) Prevalence and diversity of avian

hematozoan parasites in Asia: a regional survey. J Wildl Dis 43:382–98.

Janousek WM, Marra PP, Kilpatrick AM (2014) Avian roosting behavior influences

vector–host interactions for West Nile virus hosts. Parasite Vector 7:399. doi:

10.1186/1756-3305-7-399

Jetz W, Thomas GH, Joy JB, Hartmann K, Mooers AO (2012) The global diversity of

birds in space and time. Nature 491:444–8. doi: 10.1038/nature11631

Johnston TH (1910) On Australian avian entozoa. J Proc R Soc New South Wales 44:84–

122.

Johnston TH (1912) Internal parasites recorded from Australian birds. Emu 12:105–12.

Johnston TH, Cleland JB (1909) Notes on some parasitic protozoa. Proc Linn Soc New

South Wales 34:501–13.

Johnston E, Weinstein P, Slaney D, Flies AS, Fricker S, Williams C (2014) Mosquito

communities with trap height and urban-rural gradient in Adelaide, South

Australia: implications for disease vector surveillance. J Vector Ecol 39:48–55.

Kahn NW, St. John J, Quinn TW (1998) Chromosome-specific intron size differences in

the avian CHD gene provide an efficient method for sex identification in birds.

Auk 115:1074–8.

Kappeler PM, Cremer S, Nunn CL (2015) Sociality and health: impacts of sociality on

disease susceptibility and transmission in animal and human societies. Philos T R

Soc B 370:20140116. doi:10.1098/rstb.2014.0116

Keesing F, Holt RD, Ostfeld RS (2006) Effects of species diversity on disease risk. Ecol

Lett 9:485–98. doi: 10.1111/j.1461-0248.2006.00885.x

Kelly TR, MacGillivray HL, Sarquis-Adamson Y, Watson MJ, Hobson KA,

MacDougall-Shackleton EA (2016) Seasonal migration distance varies with natal

dispersal and predicts parasitic infection in song sparrows. Behav Ecol Sociobiol

70:1857–66. doi: 10.1007/s00265-016-2191-2

Kleindorfer S, Lambert S, Paton DC (2006) Ticks (Ixodes sp.) and blood parasites

(Haemoproteus spp.) in New Holland honeyeaters (Phylidonyris

novaehollandiae): evidence for site specificity and fitness costs. Emu 106:113–8.

doi: 10.1071/MU05055

Knowles SCL, Palinauskas V, Sheldon BC (2010) Chronic malaria infections increase

family inequalities and reduce parental fitness: experimental evidence from a wild

bird population. J Evol Biol 23:557–69. doi: 10.1111/j.1420-9101.2009.01920.x

Knowles SCL, Wood MJ, Alves R, Wilkin TA, Bensch S, Sheldon BC (2011) Molecular

epidemiology of malaria prevalence and parasitaemia in a wild bird population.

Mol Ecol 20:1062–76. doi: 10.1111/j.1365-294X.2010.04909.x

Krams I, Suraka V, Cīrule D, Hukkanen M, Tummeleht L, Mierauskas P, Rytkönen S,

Rantala MJ, Vrublevska J, Orell M, Krama T (2012) A comparison of microscopy

and PCR diagnostics for low intensity infections of haemosporidian parasites in

the Siberian tit Poecile cinctus. Ann Zool Fennici 49:331–40. doi:

10.5735/086.049.0506

LaPointe DA, Atkinson CT, Samuel MD (2012) Ecology and conservation biology of

avian malaria. Ann N Y Acad Sci 1249:211–26.

Laurance SGW, Jones D, Westcott D, Mckeown A, Harrington G, Hilbert DW (2013)

Habitat fragmentation and ecological traits influence the prevalence of avian

blood parasites in a tropical rainforest landscape. PLoS ONE 8:e76227. doi:

10.1371/journal.pone.0076227

Lee JC-I, Tsai L-C, Hwa P-Y, Chan C-L, Huang A, Chin S-C, Wang L-C, Lin J-T,

Linacre A, Hsieh H-M (2010) A novel strategy for avian species and gender

identification using the CHD gene. Mol Cell Probes 24:27–31. doi:

10.1016/j.mcp.2009.08.003

Leu ST, Kappeler PM, Bull CM (2010) Refuge sharing network predicts ectoparasite

load in a lizard. Behav Ecol Sociobiol 64:1495–503. doi: 10.1007/s00265-010-

0964-6

Loehle C (1995) Social barriers to pathogen transmission in wild animal populations.

Ecol 76:326–35.

Loiseau C, Harrigan RJ, Robert A, Bowie RCK, Thomassen HA, Smith TB, Sehgal RNM

(2012) Host and habitat specialization of avian malaria in Africa. Mol Ecol

21:431–41. doi: 10.1111/j.1365-294X.2011.05341.x

Lutz HL, Hochachka WM, Engel JI, Bell JA, Tkach VV, Bates JM, Hackett SJ,

Weckstein JD (2015) Parasite prevalence corresponds to host life history in a

diverse assemblage of Afrotropical birds and haemosporidian parasites. PLoS

ONE 10:e0121254. doi: 10.1371/journal.pone.0121254

Mackerras MJ, Mackerras IM (1960) The haematozoa of Australian birds. Aust J Zool

8:226–60.

Makowski D, Ben-Shachar M, Lüdecke D (2019a) bayestestR: describing effects and

their uncertainty, existence and significance within the Bayesian framework. J

Open Source Softw 4:1541. doi: 10.21105/joss.01541

Makowski D, Ben-Shachar MS, Chen SHA, Lüdecke D (2019b) Indices of effect

existence and significance in the Bayesian framework. Front Psychol 10:2767.

doi: 10.3389/fpsyg.2019.02767

Markus MB, Oosthuizen JH (1971) Pathogenicity of Haemoproteus columbae. Trans R

Soc Trop Med Hyg 66:186–7. doi: 10.1016/0035-9203(72)90072-7

Martínez-de la Puente J, Martínez J, Rivero-de Aguilar J, Herrero J, Merino S (2011) On

the specificity of avian blood parasites: revealing specific and generalist

relationships between haemosporidians and biting midges. Mol Ecol 20:3275–87.

doi: 10.1111/j.1365-294X.2011.05136.x

McNew SM, Barrow LN, Williamson JL, Galen SC, Skeen HR, DuBay SG, Gaffney

AM, Johnson AB, Bautista E, Ordoñez P, Schmitt CJ, Smiley A, Valqui T, Bates

JM, Hackett SJ, Witt CJ (2021) Contrasting drivers of diversity in hosts and

parasites across the tropical Andes. P Natl Acad Sci USA 118:e2010714118. doi:

10.1073/pnas.2010714118

Merino S, Moreno J, Sanz JJ, Arriero E (2000) Are avian blood parasites pathogenic in

the wild? A medication experiment in blue tits (Parus caeruleus). P Roy Soc Lond

B Bio 267:2507–10. doi: 10.1098/rspb.2000.1312

Mooring MS, Hart BL (1992) Animal grouping for protection from parasites: selfish herd

and encounter-dilution effects. Behaviour 123:173–93.

Mordecai EA, Paaijmans KP, Johnson LR, Balzer C, Ben-Horin T, Moor E, McNally A,

Pawar S, Ryan SJ, Smith TC, Lafferty KD (2013) Optimal temperature for

malaria transmission is dramatically lower than previously thought. Ecol Lett

16:22–30. doi: 10.1111/ele.12015

Mullen GR, Durden LA (Editors) (2019) Medical and veterinary entomology. Third

Edition. Academic Press, London, United Kingdom.

Murdock CC, Paaijmans KP, Cox-Foster D, Read AF, Thomas MB (2012) Rethinking

vector immunology: the role of environmental temperature in shaping resistance.

Nat Rev Microbiol 10:869–76. doi: 10:1038/nrmicro2900

Norris K, Evans MR (2000) Ecological immunology: life history trade-offs and immune

defense in birds. Behav Ecol 11:19–26.

Olsson-Pons S, Clark NJ, Ishtiaq F, Clegg SM (2015) Differences in host species

relationships and biogeographic influences produce contrasting patterns of

prevalence, community composition and genetic structure in two genera of avian

malaria parasites in southern Melanesia. J Anim Ecol 84:985–98. doi:

10.1111/1365-2656.12354

Ortiz-Catedral L, Brunton D, Stidworthy MF, Elsheikha HM, Pennycott T, Schulze C,

Braun M, Wink M, Gerlach H, Pendl H, Gruber AD, Ewen J, Pérez-Tris J,

Valkiūnas G, Olias P (2019) Haemoproteus minutus is highly virulent for

Australasian and South American parrots. Parasite Vector 12:40. doi:

10.1186/s13071-018-3255-0

Paaijmans KP, Blanford S, Chan BHK, Thomas MB (2012) Warmer temperatures reduce

the vectorial capacity of malaria mosquitoes. Biol Lett 8:465–8. doi:

10.1098/rsbl.2011.1075

Palinauskas V, Valkiūnas G, Bolshakov CV, Bensch S (2011) Plasmodium relictum

(lineage SGS1) and Plasmodium ashfordi (lineage GRW2): the effects of the co-

infection on experimentally infected passerine birds. Exp Parasitol 127:527–33.

Palinauskas V, Žiegytė R, Šengaut J, Bernotienė R (2018) Different paths – the same

virulence: experimental study on avian single and co-infections with Plasmodium

relictum and Plasmodium elongatum. Int J Parasitol 48:1089–96.

Paradis E, Schliep K (2019) ape 5.0: an environment for modern phylogenetics and

evolutionary analyses in R. Bioinformatics 35:526–8.

Pauli JN, Buskirk SW, Williams ES, Edwards WH (2006) A plague epizootic in the

black-tailed prairie dog (Cynomys ludovicianus). J Wildl Dis 42:74–80. doi:

10.7589/0090-3558-42.1.74

Pedersen TL (2020) tidygraph: a tidy API for graph manipulation. Version 1.2.0.

Peirce MA (1984) Haematozoa of Zambian birds. I. General survey. J Nat Hist 18:105–

22.

Peirce MA, Mead CJ (1978) Haematozoa of British birds: III. Spring incidence of blood

parasites of birds from Hertfordshire, especially returning migrants. J Nat Hist

12:337–40.

R Core Team (2020) R: a language and environment for statistical computing. R

Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org

Raffel TR, Romansic JM, Halstead NT, McMahon TA, Venesky MD, Rohr JR (2013)

Disease and acclimation in a more variable and unpredictable climate. Nat Clim

Chang 3:146–51. doi: 10.1038/nclimate1659

Rawsthorne J, Watson DM, Roshier DA (2011) Implications of movement patterns of a

dietary generalist for mistletoe seed dispersal. Austral Ecol 36:650–5. doi:

10.1111/j.1442-9993.2010.02200.x

Reid N (1984) The role of birds in the reproduction of an arid zone population of grey

mistletoe Amyema quandang (Loranthaceae). PhD Dissertation. University of

Adelaide, Adelaide, Australia.

Richard FA, Sehgal RNM, Jones HI, Smith TB (2002) A comparative analysis of PCR-

based detection methods for avian malaria. J Parasitol 88:819–22.

Ricklefs RE (1992) Embryonic development period and the prevalence of avian blood

parasites. P Natl Acad Sci USA 89:4722–5.

Schliep KP (2011) phangorn: phylogenetic analysis in R. Version 2.5.5. Bioinformatics

27:592–3.

Sehgal RNM, Jones HI, Smith TB (2005) Blood parasites of some West African

rainforest birds. J Vet Med Sci 67:295–301.

Slowinski SP, Fudickar AM, Hughes AM, Mettler RD, Gorbatenko OV, Spellman GM,

Ketterson ED, Atwell JW (2018) Sedentary songbirds maintain higher prevalence

of haemosporidian parasite infections than migratory conspecifics during seasonal

sympatry. PLoS ONE 13:e0201563. doi: 10.1371/journal.pone.0201563

Smith VW, Cox FEG (1972) Blood parasites and the weights of Palaearctic migrants in

central Nigeria. Ibis 114:105–6.

Spencer KA, Buchanan KL, Leitner S, Goldsmith AR, Catchpole CK (2005) Parasites

affect song complexity and neural development in a songbird. P Roy Soc Lond B

Bio 272:2037–43. doi: 10.1098/rspb.2005.3188

Stan Development Team (2020) RStan: the R interface to Stan. Version 2.21.2. http://mc-

stan.org

Svensson-Coelho M, Blake JG, Loiselle BA, Penrose AS, Parker PG, Ricklefs RE (2013)

Diversity, prevalence and host specificity of avian Plasmodium and

Haemoproteus in a western Amazon assemblage. Ornithol Monogr 76:1–47. doi:

10.1525/om.2013.76.1.1

Swanson DA, Adler PH (2010) Vertical distribution of haematophagous Diptera in

temperate forests of the southeastern U.S.A. Med Vet Entomol 24:182–8. doi:

10.1111/j.1365-2915.2010.00862.x

Tamura K, Nei M (1993) Estimation of the number of nucleotide substitutions in the

control region of mitochondrial DNA in humans and chimpanzees. Mol Biol Evol

10:512–26.

Tella JL (2002) The evolutionary transition to coloniality promotes higher blood

parasitism in birds. J Evol Biol 15:32–41.

Tella JL, Blanco G, Forero MG, Gajón Á, Donázar JA, Hiraldo F (1999) Habitat, world

geographic range, and embryonic development of hosts explain the prevalence of

avian hematozoa at small spatial and phylogenetic scales. P Natl Acad Sci USA

96:1785–9.

Tella JL, Scheuerlein A, Ricklefs RE (2002) Is cell-mediated immunity related to the

evolution of life-history strategies in birds? P Roy Soc Lond B Bio 269:1059–66.

Valkiūnas G (2005) Avian malaria parasites and other haemosporidia. CRC Press, Boca

Raton, Florida, USA.

Valkiūnas G, Bensch S, Iezhova TA, Križanauskienė A, Hellgren O, Bolshakov CV

(2006) Nested cytochrome b polymerase chain reaction diagnostics underestimate

mixed infections of avian blood haemosporidian parasites: microscopy is still

essential. J Parasitol 92:418–22.

Valkiūnas G, Palinauskas V, Ilgūnas M, Bukauskaitė D, Dimitrov D, Bernotienė R,

Zehtindjiev P, Ilieva M, Iezhova TA (2014) Molecular characterization of five

widespread avian haemosporidian parasites (Haemosporida) with perspectives on

the PCR-based detection of haemosporidians in wildlife. Parasitol Res 113:2251–

63. doi: 10.1007/s00436-014-3880-2

Van Bael S, Pruett-Jones S (2000) Breeding biology and social behaviour of the eastern

race of the splendid fairy-wren Malurus splendens melanotus. Emu 100:95–108.

doi: 10.1071/MU9831 van Riper III C, van Riper SG, Goff ML, Laird M (1986) The epizootiology and

ecological significance of malaria in Hawaiian land birds. Ecol Monogr 56:327–

44.

Vredenburg VT, Knapp RA, Tunstall TS, Briggs CJ (2010) Dynamics of an emerging

disease drive large-scale amphibian population extinctions. P Natl Acad Sci USA

107:9689–94. doi: 10.1073/pnas.0914111107

Waldenström J, Bensch S, Kiboi S, Hasselquist D, Ottosson U (2002) Cross-species

infection of blood parasites between resident and migratory songbirds in Africa.

Mol Ecol 11:1545–54.

Waldenström J, Bensch S, Hasselquist D, Östman Ö (2004) A new nested polymerase

chain reaction method very efficient in detecting Plasmodium and Haemoproteus

infections from avian blood. J Parasitol 90:191–4.

Warner RE (1968) The role of introduced diseases in the extinction of the endemic

Hawaiian avifauna. Condor 70:101–20.

Whitehead H (2008) Analyzing animal societies: quantitative methods for vertebrate

social analysis. University of Chicago Press, Chicago, Illinois, USA.

Wilman H, Belmaker J, Simpson J, de la Rosa C, Rivadeneira MM, Jetz W (2014)

EltonTraits 1.0: species-level foraging attributes of the world’s birds and

mammals. Ecol 95:2027. doi: 10.1890/13-1917.1

Wood MJ, Cosgrove CL, Wilkin TA, Knowles SCL, Day KP, Sheldon BC (2007)

Within-population variation in prevalence and lineage distribution of avian

malaria in blue tits, Cyanistes caeruleus. Mol Ecol 16:3263–73. doi:

10.1111/j.1365-294X.2007.03362.x

Wright ES (2016) Using DECIPHER v2.0 to analyze big biological sequence data in R. R

Journal 8:352–9.

Zamora-Vilchis I, Williams SE, Johnson CN (2012) Environmental temperature affects

prevalence of blood parasites of birds on an elevation gradient: implications for

disease in a warming climate. PLoS ONE 7:e39208. doi:

10.1371/journal.pone.0039208

APPENDIX A

LIST OF SAMPLED HOST SPECIES

Table A.1 Species sampled at Brookfield Conservation Park, Australia. Average (conspecific) group size, foraging stratum, and weighted eigenvector centrality for each species were derived from direct observations of the focal community as described in the text (see Chapter 1). Nest structure was compiled from Birds of the World species accounts (Billerman et al. 2020), movement patterns from the Australian Bird Database (Garnett et al. 2015), and longevity from the Australian Bird and Bat Banding Scheme database (ABBBS 2020).

Average Foraging Nest Migratory Longevity Eigenvector Species N Haemoproteus Plasmodium Leucocytozoon group size stratum structure / nomadic (years) centrality

Climacteridae

Climacteris picumnus 2 . . . 1.82 low canopy hollow 0 13.92 0.272

Maluridae

Malurus assimilis 119 12 . . 2.69 bush closed 0 10.97 0.869

Malurus splendens 135 14 1 . 2.43 ground closed 0 11.01 0.959

Malurus leucopterus 1 . . . — bush closed 0 5.97 —

Meliphagidae

Purnella albifrons 26 . . . 2.5 low canopy open 1 4.11 0.272

Acanthagenys rufogularis 14 13 . . 1.4 low canopy open 1 13.57 0.348

Gavicalis virescens 9 8 . . 1 bush open 1 13.13 0.091

Ptilotula ornata 38 22 1 . 2.5 high canopy open 1 9.12 0.696

Nesoptilotis leucotis 3 . . . 1.4 low canopy open 1 12.78 0.218

Melithreptus brevirostris 30 1 2 . 3.71 high canopy open 1 15.71 0.545

Pardalotidae

Pardalotus punctatus 7 . . . 2 high canopy hollow 1 6.5 0.368

Pardalotus striatus 6 . . . 2.31 high canopy hollow 1 6.02 0.660

Average Foraging Nest Migratory Longevity Eigenvector Species N Haemoproteus Plasmodium Leucocytozoon group size stratum structure / nomadic (years) centrality

Acanthizidae Acanthiza apicalis 34 . . . 2.23 low canopy closed 0 16.89 0.575

Acanthiza chrysorrhoa 3 . . . 2.29 ground closed 0 13.1 0.272

Acanthiza uropygialis 130 . . . 3.01 low canopy hollow 0 12.63 1.000

Smicrornis brevirostris 101 1 . . 1.98 low canopy closed 0 7.39 0.789

Aphelocephala leucopsis 100 . . . 2.2 ground hollow 0 8.58 0.552

Pomatostomidae

Pomatostomus superciliosus 39 2 . . 4 bush closed 0 14.42 0.119

Pomatostomus ruficeps 52 . . . 3.75 ground closed 0 8.72 0.093

Neosittidae

Daphoenositta chrysoptera 2 . . . 3.12 low canopy open 0 6.92 0.389

Rhipiduridae

Rhipidura leucophrys 2 . . . 1.25 low canopy open 1 9.57 0.099

Petroicidae

Petroica goodenovii 35 . . . 1.69 ground open 1 5.46 0.504

Melanodryas cucullata 1 . . . 2 ground open 0 9.62 0.256

TOTAL 889 73 4 0

APPENDIX B

BANDING DATA SHEET

APPENDIX C

BANDING REFERENCE GUIDE

APPENDIX D

MOLECULAR ASSAYS

Sexing. For sampled individuals for which sex could not be determined in the field (i.e. nestlings and adults of monomorphic species), we determined the genetic sex

by PCR. PCR amplification of the avian chromodomain helicase DNA binding (CHD)

gene is a fast, accurate, and reliable means of determining an individual’s genetic sex,

and several assays have been developed to detect different sex-varying portions of the gene. The efficacy of individual assays to determine sex varies between species (Çakmak et al. 2017). I used three different assays, as no test was effective at distinguishing between the sexes for all species (based on individuals of known sex, e.g., females with brood patches). Table D.1 provides a listing of the species sexed using each method. A few honeyeaters showed ambiguous results with the CHD1F/CHD1R primers. For those

birds, sex was confirmed using the primers 1237L/1272H (Kahn et al. 1998).

All sexing reactions were carried out in volumes of 15 µl containing 1× PCR

master mix (M7822; GoTaq® G2 Green, Promega Corp., Madison, WI, USA), 0.2 µM of each primer, and 25 ng of DNA template. Reaction conditions are provided in Table D.2.

Amplification products were separated in 3% agarose gel under 90 V after 75 (P2/P8 and

1237L/1272H) or 45 (CHD1F/CHD1R) min.

Table D.1 Primers used in sexing PCRs. Each primer pair was used for only a subset of sampled species.

Primer Primer sequence (5' → 3') Reference Species P2 TCTGCATCGCTAAATCCTTT Maluridae spp., Acanthizidae spp., Griffiths et al. 1998 P8 CTCCCAAGGATGAGRAAYTG Petroicidae spp. CHD1F TATCGTCAGTTTCCTTTTCAGGT Climacteridae sp., Pardalotidae spp., Lee et al. 2010 CHD1R CCTTTTATTGATCCATCAAGCCT Pomatostomidae spp., Meliphagidae spp. 1237L GAGAAACTGTGCAAAACAG Kahn et al. 1998 Meliphagidae spp. 1272H TCCAGAATATCTTCTGCTCC

Table D.2 Cycling conditions for sexing PCRs.

Primers: P2/P8 CHD1F/CHD1R 1237L/1272H Initial heat activation 3 min @ 94°C 4 min @ 94°C 9 min @ 94°C Number of cycles (N): 30 7 + 301 35 Denaturation 60 s @ 94°C 30 s @ 94°C 45 s @ 94°C N × Annealing 60 s @ 50°C 45 s @ 57–50°C1 45 s @ 56°C Extension 45 s @ 72°C 45 s @ 72°C 45 s @ 72°S Final extension 5 min @ 72°C 5 min @ 72°C 6 min @ 72°C Reference — Çakmak et al. 2017 Myers et al. 2012

1The CHD1F/CHD1R protocol uses a touchdown scheme. The annealing temperature begins at 57°C and drops by 1°C/cycle until reaching 50°C. This temperature is maintained for 30 cycles. Diagnostic procedure. We performed an initial nonspecific screening for

infection using the primers MalMitoF1 and MalMitoR1 (Table D.3). This assay is sensitive to low template concentrations and detects a broad range of haemosporidian parasites (Eastwood et al. 2019). We performed the screening reactions in volumes of 10

µL containing 1× PCR master mix (M7422; GoTaq® G2 Hot Start Green, Promega Corp.,

Madison, WI, USA), 0.1 µM of each primer, and 40 ng of DNA template. We included

one negative control (nuclease-free water) and one positive control for every 12 samples.

As a positive control, we used extracted DNA from a bird previously identified as

positive for haemosporidian infection. Cycling conditions are given in Table D.4.

Products of the screening PCR were resolved in 2% agarose gels under 90 V after 45

min.

Table D.3 Primers used in diagnostic PCRs.

Primer Primer sequence (5' → 3') Reference MalMitoF1 AGCCAAAAGAATAGAAACAGATGCCAGGCCAA Eastwood et al. 2019 MalMitoR1 AGCGATRCGTGAGCTGGGTTAAGAACGTCTTGAG Prim3_F2 ACTGGTGTATTATTAGCAACTTGTTATACT Eastwood et al. 2019 Prim3_R1 GCTTGGGAGCTGTAATCATAATGT HaemNF CATATATTAAGAGAATTATGGAG Waldenström et al. 2004 HaemNR2 AGAGGTGTAGCATATCTATCTAC

Haemosporidian genetic lineages are defined as variants of a ~480 bp segment of

the mitochondrial cytochrome b gene (Waldenström et al. 2004). To identify parasite

lineages present in the sample, we used the DNA barcoding method described by

Eastwood et al. (2019). This protocol is based on the nested PCR approach of

Waldenström et al. (2004), with modifications designed to enhance detection of

Australasian haemosporidians. Thus, samples that tested positive in the initial

(MalMitoF1/MalMitoR1) screening were further assayed using the nested PCR to amplify a region that includes the ~480 bp barcoding segment of the parasite mtDNA.

For the external reaction, we prepared 25 µL reaction volumes containing 1× PCR master mix (M7432; GoTaq® G2 Hot Start Colorless, Promega Corp., Madison, WI, USA),1.25

µM of each primer (Prim3_F2 and Prim3_R1; see Table D.3), and 30 ng template DNA.

Reaction conditions are indicated in Table D.4. We confirmed amplification by running

the products in 1% agarose gels under 90 V for 45 min.

As the template for the internal reaction, we used either diluted product from the

external reaction (5 µL PCR product in 95 µL nuclease-free water; when the external

reaction yielded a strong positive result) or diluted genomic DNA (5µL extracted DNA in

45 µL nuclease-free water; when the external reaction yielded a weak positive result or

failed to amplify the target). The internal reaction mixture was prepared similarly to that

for the external reaction, using HaemNF and HaemNR2 as primers and including 3 µL of

the diluted template. Reaction conditions are given in Table D.4. Amplified products

were resolved in 1% agarose gels under 90 V for 45 min.

Table D.4 Cycling conditions for screening and barcoding PCRs.

Primers: MalMitoF1/MalMitoR1 Prim3_F2/Prim3_R1 HaemNF/HaemNR2 Initial heat activation 15 min @ 95°C 15 min @ 95°C 15 min @ 95°C Number of cycles (N): 5 + 10 + 301 14 + 302 5 + 353 Denaturation 30 s @ 94°C 30 s @ 94°C 30 s @ 94°C N × Annealing 30 s @ 65–55°C1 60 s @ 70–56°C2 45 s @ 55–50°C3 Extension 30 s @ 72°C 70 s @ 70°C 70 s @ 70°S Final extension 5 min @ 72°C 5 min @ 70°C 10 min @ 70°C

1 An annealing temperature of 65°C is used for the first 5 cycles, after which the annealing temperature drops by 1°C/cycle until reaching 55°C (10 cycles). This temperature is maintained for 30 cycles. 2 Annealing temperature begins at 70°C and drops by 1°C/cycle until reaching 56°C (14 cycles). This temperature is maintained for the remaining 30 cycles. 3 Annealing temperature begins at 55°C and drops by 1°C/cycle until reaching 50°C (5 cycles). This temperature is maintained for the remaining 35 cycles. We diluted PCR products from the internal reaction with nuclease-free water and

eliminated residual PCR reagents and primers using exonuclease I and shrimp alkaline

phosphatase (70995.1.KT; PCR Product Pre-Sequencing Kit, Thermo Fisher Scientific,

Waltham, MA, USA). We submitted the cleaned, diluted products to the University of

Minnesota Genomics Center (Minneapolis, MN, USA) for bidirectional sequencing using

the HaemNF/HaemNR2 primers. Consensus sequences were submitted to the MalAvi

and GenBank (accession numbers MW888415–MW888421) databases.

References

Çakmak E, Pekşen ÇA, Bilgin CC (2017) Comparison of three different primer sets for

sexing. J Vet Diagn Invest 29:59–63. doi: 10.1177/1040638716675197

Eastwood JR, Peacock L, Hall ML, Roast M, Murphy SA, Gonçalves da Silva A, Peters

A (2019) Persistent low avian malaria in a tropical species despite high

community prevalence. Int J Parasitol: Parasites Wildl 8:88–93. doi:

10.1016/j.ijppaw.2019.01.001

Griffiths R, Double MC, Orr K, Dawson RJG (1998) A DNA test to sex most birds. Mol

Ecol 7:1071–5.

Kahn NW, St. John J, Quinn TW (1998) Chromosome-specific intron size differences in

the avian CHD gene provide an efficient method for sex identification in birds.

Auk 115:1074–8.

Lee JC-I, Tsai L-C, Hwa P-Y, Chan C-L, Huang A, Chin S-C, Wang L-C, Lin J-T,

Linacre A, Hsieh H-M (2010) A novel strategy for avian species and gender

identification using the CHD gene. Mol Cell Probes 24:27–31. doi:

10.1016/j.mcp.2009.08.003

Myers S, Paton DC, Kleindorfer S (2012) Sex determination by morphology in New

Holland honeyeaters, Phylidonyris novaehollandiae: contrasting two popular

techniques across regions. S Aust Ornithologist 38:1–11.

Waldenström J, Bensch S, Hasselquist D, Östman Ö (2004) A new nested polymerase

chain reaction method very efficient in detecting Plasmodium and Haemoproteus

infections from avian blood. J Parasitol 90:191–4.

APPENDIX E

PRIORS, CONVERGENCE, AND POSTERIOR PREDICTIVE CHECKS

Priors. Prior distributions describe the support and density naïvely expected for a

particular parameter—that is, before the data are taken into consideration. Thus, the

vaguest possible prior is a flat (uniform) distribution across all real numbers. Such priors

may be desirable to analysts seeking to minimize their influence on the results of an

analysis. However, the use of extremely vague priors can at times be a hindrance, as their use can cause Markov-chain Monte Carlo (MCMC) samplers to spend considerable time

exploring unrealistic parameter space. Fortunately, the application of even modest

amounts of domain knowledge can vastly improve sampling efficiency and model

convergence without unduly constraining the posterior. Prior specifications can also

improve the identifiability of individual parameters in the presence of collinearity.

Moreover, a series of simple prior and posterior predictive checks can help guide prior

selection and ensure that the influence of the prior on the posterior is minimized. Logistic

regression offers an excellent example of how limited prior knowledge can vastly

improve the efficiency of the MCMC sampling procedure.

Logistic regression takes place in real parameter space—that is, parameters can in

theory take any positive or negative real value. However, these parameters are

functionally stand-ins representing changes observed in probability space—the infinite

set of real numbers between 0 and 1. Moreover, the inverse logit transformation, which conveniently maps real (parameter) space onto probability (data) space, does so in such a

way as to result in redundancy among large positive and negative real values for most

practical applications (Fig. E.1).

Figure E.1 The inverse logit transformation is commonly applied as a link function in logistic regression to map real space (x axis) onto probability space (y axis). Given the nature of this mapping, there is considerable redundancy among absolutely large values in the domain, so that small changes to very large positive or negative values in real space are essentially meaningless in probability space. Much probability resolution is achieved at x values close to 0 (±5), whereas comparatively little resolution of probability values occurs among large real values. (a) A unit change in real space corresponds to a 23.1% probability change when the real change is near the origin. (b) But for unit changes away from the origin in real space, the corresponding probability change is drastically reduced. As a consequence of this redundancy, the effect on probability estimates of unit

changes to real-space parameter estimates varies depending on where in real space the

change occurs. For instance, parameter estimates of 0 and 1 correspond to a 23%

difference in probability, whereas estimates of 5 and 6 map to a probability change of

<0.5% (Fig. E.1). Thus, the nature of the inverse logit transformation means that many

logistic-regression parameter estimates can be expected to be relatively close to 0,

particularly when covariates are centered prior to fitting the model (as is performed

internally by brms). As such, when fitting Bayesian regression models it is often prudent

to restrict the MCMC sampler’s consideration of large values by placing weakly

informative priors on the parameter estimates.

The need to consider more informative priors is further illustrated by examining

the implications of the extremely vague flat priors which are the defaults for coefficient

and intercept estimates in brms. Priors that are flat in parameter (real) space may not be

flat in data (probability) space. Sampling from a uniform distribution with limits at

±10,000 places much prior density close 0 and 1 on the probability scale (Fig. E.2). On

the other hand, a more restrictive prior can reduce the prior density in these regions while

providing significant gains in terms of sampling efficiency.

Figure E.2 Density functions for (a) flat (Uniform(–1E4,1E4)) and (b) weakly informative (Normal(0,1.5)) priors after mapping to the probability scale. The vague flat prior distribution places most prior density near 0 and 1 in probability space, whereas the more informative prior distributes prior weight more evenly across this space. We initially performed MCMC sampling using the default (flat) priors for all

coefficient and intercept estimates. This resulted in long sampling times and convergence

warnings. Essentially, the MCMC sampler was unable to efficiently sample the posterior

because it was spending time considering values for the parameters that (as described

above) were not meaningful in the context of the data. We therefore placed weakly

informative Gaussian priors on the intercept and slope parameters. These priors were

located at 0 (corresponding to a probability of 0.5), with standard deviation 1.5. This

distribution yields the density function presented in Figure E.2 (b) when mapped to

probability space. This transformed prior has some support for all possible values in the

set (0,1) (i.e., the probability domain) without placing undue weight on 0 and 1 as is done

by the default priors.

The default priors used by brms for the scale parameters in a multilevel regression

are half-t, with 3 degrees of freedom and a standard deviation of 2.5. This is a relatively

fat-tailed distribution, and for a logistic regression gives the group effects freedom to

vary comparatively widely about the origin. With these default priors, we found that the

model suffered from a lack of identifiability between the slope coefficients and the group

intercepts. This lack of identifiability manifested as very broad credible intervals for all

slope parameters, and a large estimate for the scale of the group effects. It was

compounded by the fact that most of the coefficients of interest separate at the group

level (as species, rather than individuals, comprise the primary target of inference). Thus,

the covariates themselves were able to be perfectly predicted by the grouping factor (i.e.,

species). In short, it seemed that all of the variation in the response was being explained

by the group (species) effects, and none was allotted to the covariates. In this case, we

were able to improve the identifiability of the parameter estimates by specifying a more

restrictive prior for the scale parameter. After trying several variations from the t family

of distributions, we ultimately used a half-normal with scale 0.5. The selected prior balances parameter identifiability against explanatory power. t priors with fewer degrees

of freedom or higher variance still allowed the group effects to account for all variance in

the response (yielding non-identifiable slope parameters). On the other hand, more

restrictive priors (i.e., those with smaller variance) improved the identifiability of the

coefficient estimates, but lacked predictive power. These models showed signs of

100

underfitting, consistently under-predicting prevalence for species with high observed

prevalences and over-predicting for those with low observed prevalence.

In summary, by specifying a weakly-informative prior for the coefficient

estimates, we were able to increase sampling efficiency substantially. The domain-

specific prior information allowed the model to achieve convergence (see Convergence diagnostics, below) without encountering the same difficulties as when the default priors were used. Using a more restrictive prior for the scale on the group (“random”) intercepts allowed us to strike a balance between having too much phylogenetic information explaining the response (leaving nothing for the covariates to explain) and too little

(yielding an underfitted model). Moreover, posterior predictive checks (see below) suggest that the choice of prior was not too restrictive, but that the likelihood (informed by the data) was able to dominate the posterior distribution.

Convergence diagnostics. In the context of MCMC sampling, convergence describes the degree to which each Markov chain settles on (converges to) a particular value for an estimate over the course of sampling the posterior for that parameter, as well as the degree to which the independent chains settle on (converge to) the same estimate for a parameter. There are a number of numerical and graphical tools that can be used to diagnose convergence pathologies, and in general these should always be evaluated after sampling. Models that have not yet converged have failed to adequately sample the posterior. Estimates (and inferences) from models with serious convergence issues thus cannot be relied upon as accurate representations of the full posterior.

To fit the full model with weakly-informative priors, we ran four Markov chains for 2,000 iterations each, discarding the first 1,000 iterations from each chain. We fit the

101

model using 100 phylogenetic covariance matrices sampled from a Bayesian

phylogenetic posterior, for a total of 400,000 posterior samples. A few sampling runs

encountered divergent transitions after the warm-up period. When that occurred, we reduced the step size and resampled the posterior. Doing so eliminated all divergent transitions, and we encountered no other convergence issues. The Gelman-Rubin diagnostic statistics (R̂ ) were close to 1 (±0.05) for all parameters, suggesting that convergence was achieved both within and among chain. Examination of the trace plots

(Fig. E.3) further supports this conclusion. The chains show little separation, and each appears to have achieved stationarity (i.e., a generally horizontal trajectory) during the warm-up or burn-in period. Furthermore, the marginal posterior densities are not overly diffuse, but indicate convergence to a reasonable range, with broad agreement between chains (Fig. E.4).

102

Figure E.3 Trace plots for all population (fixed effect) and scale parameters. All four chains show evidence of convergence for each parameter, having achieved stationarity (a generally horizontal trajectory) and little separation between chains.

103

Figure E.4 Posterior densities for all population (fixed effect) and scale parameters. The density of the estimates derived from each of the four chains is shown as a separate curve. Convergence is evident both from the shape or scale of the individual curves and the agreement between chains.

104

Posterior predictive checks. The posterior predictive check offers a means of

model validation. Provided that a model has converged, these tests permit assessment of

the quality of the model structure (i.e., the likelihood) in light of the available data.

Bayesian models are both evaluative and generative, allowing for the generation of new

“observations” by drawing samples from the posterior distribution. Posterior predictive

checks thus make use of replicated datasets sampled from the posterior distribution.

Conceptually, if a model is a good fit for the data, it should be able to generate new data

that is similar to the original data. That is, the posterior distribution of a well-fitting model ought to make “good” predictions.

Figure E.5 Graphical posterior predictive checks, showing the distribution of outcomes across 40,000 replicated datasets. y is a Bernoulli-distributed outcome, individual infection status. Each panel provides a slightly different illustration of the distribution of model-predicted prevalence and its relation to the observed prevalence. (a) Observed prevalence (dark vertical line) and a histogram showing the distribution of posterior prevalences. (b) Observed counts for negative (0) and positive (1) infection status (bars), together with the predicted counts (points, with error bars illustrating 90% credible intervals). Note that both the posterior predictions and the observed data show a preponderance of infection-negative outcomes, corresponding to a low prevalence (~10%, (a)). To perform a posterior predictive check, replicated observations are generated at

each iteration in the sample. Thus, for each observation, a matched replicate (assumed to

have the same covariate values as the original) is drawn from the posterior at every

105

iteration in every chain. It’s then possible to compare the original data (or summaries

thereof) to the distribution of replicated datasets (or summaries thereof). For example, the panels in Figure E.5 illustrate that the model described above reliably predicts prevalence at the population level, while Figure E.6 demonstrates that the same is true at the group

(species) level.

106

Figure E.6 Histograms of posterior group (species) prevalences. Observed prevalence for each species is visible as a dark vertical line, while the distributions of predicted species-specific prevalences are shown by the histograms. Note that the sample size for each species constrains the set of possible realizations for prevalence for that species. For instance, a species having only two samples has only three possible prevalence outcomes: 0, 0.5, or 1. It can also be informative to compare the posterior predictions to prior predictions. This comparison should demonstrate that the likelihood, rather than the prior, dominates the posterior distribution by drawing the posterior density towards the

107 observed data. That the posterior densities more closely reflect the data (Fig. E.5) than they do the prior (Fig. E.7) suggests that the priors were not too restrictive, and that they did not dominate the posterior.

Figure E.7 Graphical prior predictive checks, showing the distribution of outcomes across 100,000 replicated datasets. y is a Bernoulli-distributed outcome, individual infection status. Each panel provides a slightly different illustration of the distribution of model-predicted prevalence and its relation to the observed prevalence. (a) Observed prevalence (dark vertical line) and a histogram showing the distribution of prior prevalences. (b) Observed counts for negative (0) and positive (1) infection status (bars), together with the naïvely-predicted counts (points, with error bars illustrating 90% credible intervals). Note that in both cases the prior distribution places at least some density across all possible values for the response.

108

APPENDIX F

LINEAGE-SPECIFIC INFECTION ANALYSES

When parasite lineages have different vector and host associations, they may be

more or less sensitive to environmental changes or differences in host ecology and

behavior. This heterogeneity, in turn, can obscure unique patterns of occurrence for

specific parasites when we pool infection data across lineages for analysis (Wood et al.

2007). Moreover, knowledge about the occurrence of individual parasite lineages can provide additional insight into the proximate drivers of transmission. The genetic data we presented in Chapter 2 shed light on potential transmission pathways in the study system

by revealing stark patterns of host–parasite association, with parasite lineages clustering

at the host family level. To test whether our initial hypotheses about host ecology and

social behavior might hold for infection with individual parasite lineages, we modelled

infection by the two most prevalent lineages separately within their respective host families. In this section, we describe these two lineage-specific models and their results.

The predictions of both models are largely similar to those of the general model of

Haemoproteus infection.

Models. We created separate Bayesian phylogenetic multilevel regression models for infection with Haemoproteus lineages CLIPIC01 and CLIPIC02. In each case, we

restricted the dataset to include only samples from the host family in which we identified

the respective lineage—Meliphagidae for CLIPIC01, and Maluridae for CLIPIC02 (see

Chapter 2). As in the general model of Haemoproteus infection, we only considered

samples from adults for the reduced models. We used the same basic model structure as

109

for the general Haemoproteus model (Chapter 1), but removed terms for nest structure

and migratory status, as these did not vary within either family. The two fairywren species also had nearly identical longevity (purple-backed fairywren [Malurus assimilis]:

10.97 years; splendid fairywren [M. splendens]: 11.01 years), so we excluded the effect

of longevity from the CLIPIC02 model as well. Thus, the model included population

(“fixed”) effects of sex, habitat (proportion of mallee cover), average conspecific group size, longevity (CLIPIC01 model only), modal foraging stratum, nest structure, and eigenvector centrality in the mixed-species flock network. We included phylogenetic

group (“random”) intercepts for species as described in Chapter 1. Prior and posterior

predictive checks and convergence tests were performed for both models as described for

the general Haemoproteus model in Chapter 1 and Appendix E. Prevalences are reported

with 95% highest density intervals computed using a Jeffreys prior (Beta(0.5, 0.5)).

Results.

110

Figure F.1 Posterior densities for effects on Haemoproteus sp. lineage CLIPIC01 infection probability among honeyeaters (Meliphagidae spp.). The tails of the distributions indicate the ranges of parameter estimates across 400,000 MCMC iterations. Points indicate the posterior median for each distribution, with error bars showing the 95% highest posterior density intervals. The region of practical equivalence to zero (ROPE) for logistic regression coefficients is delimited by dashed lines (0 ± 0.18). We sampled 120 adult honeyeaters (Meliphagidae spp.) from 6 species, and

identified 34 individuals of 4 species infected by CLIPIC01 (28.3% [20.6%, 36.6%]).

Two species (white-eared honeyeater [Nesoptilotis leucotis], n = 3; white-fronted honeyeater [Purnella albifrons], n = 26) had no CLIPIC01 infections. Among species with CLIPIC01 infections, prevalence ranged from 3.3% [0.0%, 12.3%] (brown-headed

honeyeater [Melithreptus brevirostris], n = 30) to 64.3% [40.0%, 85.9%] (spiny-cheeked honeyeater [Acanthagenys rufogularis], n = 14). As was the case in the general model of

Haemoproteus infection, the estimates for most effects were small and had large

111 uncertainty (Fig. F.1). Only the effect of group size had a high probability of existence

(Fig. F.1, Table F.1); all others had both 95% highest density credible intervals overlapping zero and pd < 0.975. The negative effect of group size on the probability of

CLIPIC01 infection in honeyeaters was stronger than that reported for Haemoproteus infection in the community as a whole (Fig. F.2), and can be considered significant (0.0% in ROPE; Table F.1).

Table F.1 Posterior summary for CLIPIC01 infection status in 6 species of honeyeaters. For each population (“fixed”) effect parameter, median posterior estimates are reported with 95% highest density credible intervals, the probability of direction pd, and the percent of the full posterior overlapping the region of practical equivalence (ROPE) to the null value of zero. Effects with a high probability of existence (95% CrI not overlapping zero and pd > 0.975) are shown in bold. Estimated effects for the reference categories for factor variables (Sex: female and Stratum: bush) are captured in the estimate for the intercept term.

Posterior Effect median 95% HD CrI pd % in ROPE Intercept 0.232 −2.20, 2.26 0.575 11.5% Sex: male −0.342 −1.31, 0.61 0.760 22.8% Habitat (mallee coverage) 0.397 −1.29, 2.08 0.679 15.1% Longevity 0.110 −0.06, 0.30 0.901 77.6% Stratum: low canopy 0.105 −1.52, 1.71 0.551 17.4% Stratum: high canopy 1.998 −0.25, 4.27 0.959 2.7% Group size −2.110 −3.24, −0.97 1.000 0.0% Flock centrality 2.061 −0.68, 4.23 0.930 3.4%

112

Figure F.2 Decrease in probability of Haemoproteus sp. lineage CLIPIC01 infection with average conspecific group size among 6 species of honeyeaters (Meliphagidae) at Brookfield Conservation Park, Australia. We sampled 167 adult fairywrens (Maluridae: Malurus spp.) of 2 species and identified 25 individuals infected by CLIPIC02, for an overall prevalence of 15.0%

[9.9%, 20.7%]. CLIPIC02 prevalence was similar in the two species (purple-backed fairywren [M. assimilis]: 13.5% [6.7%, 20.8%], n = 84; splendid fairywren [M. splendens]: 16.9% [9.5%, 25.4%], n = 83). The marginal posterior densities for all model parameters suggest that none of the predictors investigated is likely to have a major effect on CLIPIC02 infection (all pd < 0.975; Table F.2). There was particularly large

uncertainty associated with the intercept and effects for interspecific sociality (as flock

centrality) and group size (Fig. F.3). The strongest apparent effect was that of habitat,

with birds from more mallee-dominated areas expected to have a higher probability of

113

CLIPIC02 infection. However, the significance of this effect is uncertain (8.8% in

ROPE).

Table F.2 Posterior summary for CLIPIC02 infection status in 2 species of Malurus fairywrens. For each population (“fixed”) effect parameter, median posterior estimates are reported with 95% highest density credible intervals, the probability of direction pd, and the percent of the full posterior overlapping the region of practical equivalence (ROPE) to the null value of zero. None of the effects had a high probability of existence (i.e., all 95% CrI overlapped zero and all pd < 0.975). Estimated effects for the reference categories of factor variables (Sex: female and Stratum: ground) are captured in the estimate for the intercept term.

Posterior Effect median 95% HD CrI pd % in ROPE Intercept −0.315 −3.06, 2.44 0.589 10.0% Sex: male 0.483 −0.36, 1.35 0.870 18.0% Habitat (mallee coverage) 0.913 −0.63, 2.54 0.877 9.4% Stratum: bush −0.199 −1.36, 0.98 0.635 23.2% Group size −0.816 −2.33, 0.70 0.855 10.7% Flock centrality −0.283 −3.01, 2.46 0.580 10.0%

Figure F.3 Posterior densities for effects on Haemoproteus sp. lineage CLIPIC02 infection probability among fairywrens (Maluridae spp.). The tails of the distributions indicate the ranges of parameter estimates across 400,000 MCMC iterations. Points indicate the posterior median for each distribution, with error bars showing the 95% highest posterior density intervals. The region of practical equivalence to zero (ROPE) for logistic regression coefficients is delimited by dashed lines (0 ± 0.18).

114

References

Wood MJ, Cosgrove CL, Wilkin TA, Knowles SC, Day KP, Sheldon BC (2007) Within-

population variation in prevalence and lineage distribution of avian malaria in

blue tits, Cyanistes caeruleus. Mol Ecol 16:3263–73. doi: 10.1111/j.1365-

294X.03362.x

115

APPENDIX G

BIOLOGY OF HAEMOSPORIDIAN PARASITES

Haemosporidians (Apicomplexa: Haemosporida) are obligate vector-borne parasites. Comprising four families, this group of intracellular parasites includes the causative agents for malaria (Plasmodium spp.), one of the most important sources of human morbidity and mortality throughout history. Other haemosporidians are responsible for similar diseases in their respective hosts. The parasite group as a whole is broadly distributed, both geographically and phylogenetically among tetrapod hosts. The majority of the group’s diversity, including many of the species collectively referred to as avian blood parasites, belong to the genera Plasmodium, Haemoproteus, and

Leucocytozoon.

The evolutionary history of the Haemosporida is poorly understood, and the majority of the group’s diversity is thought to be as-yet undescribed (Morrison 2009).

Even the monophyly of the familiar and well-studied human Plasmodium parasites is poorly supported (Lutz et al. 2016). Resolving the phylogeny of known haemosporidian lineages is further complicated by a complex mosaic of host–parasite coevolution and host-switching events that involve multiple levels of host taxonomy (Ricklefs and Fallon

2002, Duval et al. 2007, Lutz et al. 2016). Despite the phylogenetic richness of the clade, its members share a number of broad similarities of form and function, perhaps speaking to the success of their particular strategy for existence.

Parasite life cycle. The ontogeny of avian blood parasites has been studied in detail for only a minority of species, yet general patterns of development appear to be

116

conserved throughout the broader haemosporidian parasite group (Atkinson and van

Riper 1991, Valkiūnas 2005, Atkinson et al. 2008). Avian haemosporidians exhibit

complex heteroxenous life cycles, with sexual reproduction occurring in a definitive

haematophagous invertebrate host (insects in the Order Diptera), and several rounds of

asexual reproduction occurring in an intermediate host. Infected insects transmit blood

parasites to susceptible birds (intermediate hosts) by taking a blood meal. For simplicity,

and as the present discussion pertains chiefly to haemosporidian parasites of birds,

hereafter I will refer to the definitive host as the “vector” and the intermediate host as the

“host” (or simply “bird”).

Valkiūnas (2005) distinguishes five broad phases of infection that reflect major trends in agamic parasite reproduction and development occurring within the host: (1) prepatent, (2) acute, (3) crisis, (4) chronic, and (5) latent. The duration of each stage varies among parasite taxa, host taxa, and individual hosts, but in general each of the first three periods (during which most major clinical manifestations of disease are apparent)

elapses over a matter of days or weeks. In contrast, many (if not most) individuals are

thought to maintain chronic or latent infections for years following initial infection—

even for the remainder of their lives. These two stages are distinguished from one another

by the presence of parasites in the host bloodstream (i.e. parasitemia). Birds with chronic

infections have low levels of parasitemia, whereas latent infections are maintained only

in the fixed tissues such as the liver. Such later-stage infections often involve periodic

relapse to acute or crisis states. These relapses are generally coincident with the onset of

intense physiological stresses such as migration or breeding (Valkiūnas 2005, Atkinson et

al. 2008).

117

The prepatent (or incubation) phase begins when an infected vector takes a blood

meal, inoculating the bird with saliva laced with infective sporozoites (for definitions of

terms related to the haemosporidian life cycle, refer to Appendix H; Valkiūnas 2005,

Atkinson et al. 2008). Sporozoites migrate to and penetrate the cells of the lungs, liver,

spleen, and other organs. Once inside, they initiate several rounds of exoerythrocytic

merogony. During this stage, agamic reproduction by multiple fission results in

exponential growth in the number of parasites in the host. However, the infection remains

largely subclinical, and parasitemia is low or undetectable. The prepatent stage varies in

duration from five days (Plasmodium, Leucocytozoon) to three weeks or more in some

Haemoproteus species (Valkiūnas 2005, Atkinson et al. 2008).

Acute infection begins when merozoites resulting from exoerythrocytic merogony

flood the host bloodstream to invade erythrocytes. When parasitemia reaches a

maximum, the infection is said to be at its crisis stage. The most overt host pathologies

characteristic of haemosporidian infections are associated with acute- and crisis-stage infections, when the increase in blood concentrations of parasitic cells can lead to capillary blockages, and infected erythrocytes are actively removed from circulation by the host immune system, causing anemia (Valkiūnas 2005, Atkinson et al. 2008, Ilgūnas et al. 2016, Verwey et al. 2018). Late-stage exoerythrocytic meronts can also induce pathology in the tissues of the brain, liver, lungs, and other tissues, leading to necrosis as well as pneumonia.

In the Plasmodiidae, some merozoites that penetrate host erythrocytes develop into erythrocytic meronts, which give rise to further agamic division occurring in the bloodstream (erythrocytic merogony; Valkiūnas 2005). The products of erythrocytic

118

merogony (second-generation merozoites) may then induce additional rounds of

erythrocytic and/or exoerythrocytic merogony. Erythrocytic merogony is peculiar to

plasmodiids. In all other haemosporidians (as well as in the Plasmodiidae), merozoites in

host erythrocytes give rise to gametocytes. Gametogony, the development of mature male

(microgametocytes) and female (macrogametocytes) gametocytes, occurs within the

erythrocyte and marks the final stage of the parasite life cycle occurring in the

intermediate host. In Haemoproteus, viable gametocytes appear in the blood two to six

days after merozoites first enter the circulation (i.e. after the onset of acute infection;

Valkiūnas 2005).

Soon after infected erythrocytes are ingested by a vector during a blood meal, gametocytes exit the host cell and give rise to gametes (Valkiūnas 2005). Fertilization

occurs extracellularly in the lumen of the vector’s midgut, and the zygote develops into a

motile ookinete, which penetrates the epithelial lining of the midgut, forming an

encapsulated oocyst in the basal lamina. Sporogony (the development of sporozoites)

takes place within the oocyst over a period of five to ten days (or possibly longer for

some parasites). Upon completion, sporozoites migrate via hemolymph to the salivary

glands, from whence they are delivered into the host bloodstream when the vector takes a

blood meal.

The duration of the acute and crisis stages of infection in the vertebrate host can

vary between host and parasite species as well as with host condition (Valkiūnas 2005).

In general, parasitemia subsides within two to four months postinfection. The onset of

acute infection marks the completion of the first cycle of exoerythryocytic merogony, but

this event does not indicate the termination of that process. Agamic stages remain present

119

in various fixed tissues, and the cycle of growth, division by multiple fission, and

distribution may continue to occur, at varying levels, even after the initiation of

gametogony and the development of mature gametocytes. Indeed, the persistence of

meronts capable of continuing the cycle of asexual reproduction within the cells of the

spleen, liver, and other tissues is responsible for the producing the low-intensity infections characteristic of late-stage, chronic haemosporidian infections.

As with acute parasitemia, the duration of chronic (i.e., persistent and low- intensity) infections is highly variable. It is thought that total clearance of infection from all tissues is extremely rare (Valkiūnas 2005), and chronic infections regularly give rise to periodic relapses to acute parasitemia. Nonetheless, some individuals may succeed in clearing low-intensity infections from circulation, although agamic stages are frequently still present in the spleen, liver, and other tissues. Individuals in this condition are described as having latent infections.

The ability to diagnose haemosporidian infection in birds depends on the stage of the infection. The standard molecular and microscopic diagnostic methods used in most field surveys of haemosporidian infection involve samples of the peripheral blood. As such samples do not directly assess the presence of meronts in the fixed tissues of the host, they are incapable of diagnosing prepatent and latent-stage infections, which are characterized in large part by the absence of parasites circulating in the bloodstream.

Rather, these infection stages can only be diagnosed by performing a necropsy and examining the fixed tissues directly. Thus, those infections that are detected in the bloodstream must be in either the acute, crisis, or chronic stage. Given the relative durations of each of these stages, most natural infections diagnosed in wild birds are

120 likely to be in the chronic stage. Although the intensity of parasitemia probably does influence the detectability of haemosporidian infections, standard molecular diagnostics are generally considered sufficiently sensitive to detect the low parasitemias typical of chronic-stage infections (Richard et al. 2002, Krams et al. 2012).

Vectors. Many parasite species can be transmitted by multiple vector species, and some vectors can transmit multiple parasites. However, individual vector species do vary in their ability to transmit specific parasites, and the identification of competent vectors has not been completed for most species of haemosporidian blood parasites (Valkiūnas

2005). In general, Plasmodium spp. are transmitted by mosquitoes (Culicidae: esp. Culex,

Culiseta, Aedes, Anopheles, and Mansonia spp.), whereas most Leucocytozoon spp. are transmitted by blackflies (Simuliidae: esp. Simulium, Prosimulium, and Cnephia spp.), with at least one species transmitted by a ceratopogonid biting midge (Culicoides sp.).

Haemoproteus (Parahaemoproteus) spp. are also transmitted by biting midges

(Ceratopogonidae: Culicoides spp.), although transmission of the nominate subgenus

Haemoproteus (Haemoproteus) is performed by louse flies (Hippoboscidae: esp.

Pseudolynchia spp.; Valkiūnas 2005, Atkinson et al. 2008).

Part of the variation in the epizootiology of haemosporidian blood parasites can be explained by differences in the biology of their dipteran vectors. Each of the three families of vectors demonstrates a distinct life history, with unique life cycle, host- seeking behavior, and habitat requirements (Mullen and Durden 2019). These distinctions may, in turn, contribute to variation in parasite incidence and distribution in the affected hosts.

121

Mosquitoes, which are responsible for the transmission of plasmodiid parasites,

are largely crepuscular or nocturnal, though certain species are more active during

daylight hours (Marquardt 2005). Haematophagy in mosquitoes is performed only by

females, which require blood for the development of ovarian tissues prior to oviposition.

Female mosquitoes seeking a blood meal locate hosts using a combination of olfactory,

visual, and thermal cues. Important olfactory signals include carbon dioxide, lactic acid,

and other volatile organic compounds (Marquardt 2005). Individuals can cover distances

of several kilometers while host searching (Marquardt 2005), but most individuals probably remain in a much smaller area (Mullen and Durden 2019). Like all true flies, mosquitoes undergo a complete metamorphosis, transitioning between larval and adult forms that are both physically and functionally distinct. The larvae are aquatic, and most require standing fresh water to complete development. Many species readily lay eggs in small, ephemeral pools in the soil, on exposed bedrock, or in tree hollows (Marquardt

2005). These environments provide suitable habitat for the larvae to develop before

completing their metamorphosis into a flight-capable adult. Depending on the location

and time of year, mosquitoes generally complete their life cycle over a period of a few

weeks to months.

Haemoproteus (Parahaemoproteus) parasites and at least one leucocytozoid are

transmitted by ceratopogonid biting midges of the genus Culicoides. The vector biology

of Culicoides is broadly similar to that of mosquitoes: only female midges take blood meals, and most members of the genus Culicoides seek blood meals during twilight or at night (although a few species are active diurnally; Marquardt 2005, Mullen and Durden

2019). Host-seeking behavior in midges is mediated in part by olfaction, but visual and

122

thermal cues also play important roles (Logan et al. 2010). Like mosquitoes, female

midges may fly distances up to 4 km in search of hosts (though shorter distances are more

typical; Marquardt 2005). In contrast to mosquitoes, however, ceratopogonid midges

exhibit a much broader range of breeding habitats (though all do require some amount of

moisture). Larval midges can be found in fully aquatic environments such as rivers, lakes,

and ponds, as well as in puddles, damp soil, or rotting vegetation (Marquardt 2005). The greater breadth of suitable breeding habitats (relative to those of mosquitoes and black flies) may have major consequences for the distribution of midges, and could influence the spatial distribution of their haemosporidian parasites. In the wild, adult midges live only a few weeks, during which time most will likely take only 1–2 complete blood meals (Mullen and Durden 2019).

As with mosquitoes and biting midges, only female black flies take blood meals, and locate hosts using olfactory, visual, and thermal cues (Marquardt 2005, Mullen and

Durden 2019). Species demonstrate a range of host specificities, with some black flies feeding on only a single host species and others on a diverse array of hosts. Unlike many mosquitoes and biting midges, black flies are diurnal. Black flies are generally strong fliers, and females may disperse up to 15 km from the location of emergence (though distances of over 500 km have been recorded; in those cases, dispersal is likely wind- assisted; Marquardt 2005). Larval black flies are filter feeders, and hence require flowing water for development (Mullen and Durden 2019). Although many species are associated with large rivers or streams with swift currents, several species, including some vectors of avian leucocytozoonosis, routinely use small seeps or trickles (Marquardt 2005). The

123

typical lifespan of an adult black fly is one to three months, during which time a single

female may lay two to four clutches of eggs (Marquardt 2005).

Louse flies are a group of relatively long-lived obligate parasites of birds and

mammals. Most species in the family are ornithophilic (with the balance comprising

parasites of mammals, especially ungulates), and some of these are vectors for

Haemoproteus (Haemoproteus) (Valkiūnas 2005, Mullen and Durden 2019). Contrasting

with the ephemeral relationships characterizing the host–ectoparasite interactions for the

insect vectors described previously, louse flies engage in long-term parasitic relationships

with their vertebrate hosts, remaining in close proximity to hosts for most (or all) of their

development (Hutson 1984). Although strictly holometabolous, louse flies are

larviparous. Females deposit individual larvae either directly onto the host, around the

nest, or (for flight-capable species) away from the host (Hutson 1984). Species are

variously winged, wingless, or with vestigial wings. Louse flies exhibit an array of host

specificities, but movement between hosts is probably less frequent than for other

dipteran ectoparasites (Corbet 1956, Mullen and Durden 2019). The pattern of close association between individual hosts and vectors could lead to unique transmission dynamics for the parasites transmitted by hippoboscid vectors, and (together with relatively strong host–vector specificity) might also promote more specific coevolutionary linkages between hosts and parasites.

References

Atkinson CT, van Riper III C (1991) Pathogenicity and epizootiology of avian

haematozoa: Plasmodium, Leucocytozoon, and Haemoproteus. Pages 19–48 in

124

Bird–parasite interactions: ecology, evolution, and behaviour (Loye JE, Zuk M,

Editors). Oxford University Press, Oxford, United Kingdom.

Atkinson CT, Thomas NJ, Hunter DB (Editors) (2008) Parasitic diseases of wild birds.

Wiley-Blackwell, Ames, Iowa, USA.

Corbet GB (1956) The life-history and host-relations of a hippoboscid fly Ornithomyia

fringillina Curtis. J Anim Ecol 25:403–20.

Duval L, Robert V, Csorba G, Hassanin A, Randrianarivelojosia M, Walston J, Nhim T,

Goodman SM, Ariey F (2007) Multiple host-switching of Haemosporidia

parasites in bats. Malaria J 6:157. doi: 10.1186/1475-28756-157

Hutson AM (1984) Keds, flat-flies and bat-flies: Diptera, Hippoboscidae and

Nycteribiidae. Handbooks for the Identification of British Insects. Volume 10.

Part 7. Royal Entomological Society of London, London, United Kingdom.

Ilgūnas M, Bukauskaitė D, Palinauskas V, Iezhova TA, Dinhopl N, Nedorost N,

Weissenbacher-Lang C, Weissenböck H, Valkiūnas G (2016) Mortality and

pathology in birds due to Plasmodium (Giovannolaia) homocircumflexum

infection, with emphasis on the exoerythrocytic development of avian malaria

parasites. Malar J 15:256. doi: 10.1186/s12936-016-1310-x

Krams I, Suraka V, Cīrule D, Hukkanen M, Tummeleht L, Mierauskas P, Rytkönen S,

Rantala MJ, Vrublevska J, Orell M, Krama T (2012) A comparison of microscopy

and PCR diagnostics for low intensity infections of haemosporidian parasites in

the Siberian tit Poecile cinctus. Ann Zool Fennici 49:331–40. doi:

10.5735/086.049.0506

125

Logan JG, Cook JI, Mordue (Luntz) J, Kline DL (2010) Understanding and exploiting

olfaction for the surveillance and control of Culicoides biting midges. Pages 217–

46 in Ecology and control of vector-borne diseases. Volume 2. Olfaction in

vector–host interactions (Takken W, Knols BGJ, Editors). Wageningen Academic

Publishers, Wageningen, the Netherlands.

Lutz HL, Patterson BD, Peterhans JCK, Stanley WT, Webala PW, Gnoske TP, Hackett

SJ, Stanhope MJ (2016) Diverse sampling of East African haemosporidians

reveals chiropteran origin of malaria parasites in primates and rodents. Mol

Phylogenet Evol 99:7–15. doi: 10.1016/j.ympev.2016.03.004

Marquardt WC (Editor) (2005) Biology of disease vectors. Second Edition. Elsevier

Academic Press, Burlington, Massachusetts, USA.

Mullen GR, Durden LA (Editors) (2019) Medical and veterinary entomology. Third

Edition. Academic Press, London, United Kingdom.

Morrison DA (2009) Evolution of the Apicomplexa: where are we now? Trends Parasitol

25:375–82. doi: 10.1016/j.pt.2009.05.010

Richard FA, Sehgal RNM, Jones HI, Smith TB (2002) A comparative analysis of PCR-

based detection methods for avian malaria. J Parasitol 88:819–22.

Ricklefs RE, Fallon SM (2002) Diversification and host switching in avian malaria

parasites. P Roy Soc Lond B Bio 269:885–92. doi: 10.1092/rspb.2001.1940

Valkiūnas G (2005) Avian malaria parasites and other haemosporidia. CRC Press, Boca

Raton, Florida, USA.

126

Verwey JK, Peters A, Monks D, Raidal SR (2018) Spillover of avian haemosporidian

parasites (Haemosporidia: Plasmodium) and death of captive psittacine species.

Aust Vet J 96:93–7. doi: 10.1111/avj.12671

127

APPENDIX H

GLOSSARY

cryptozoite. first-generation meront appearing in the primary exoerythrocytic merogony series of the Plasmodiidae, typically developing in the reticuloendothelial cells of the spleen, liver, and skin. Cryptozoites characteristically release comparatively small numbers of merozoites that are not capable of penetrating erythrocytes. Instead, the merozoites produced from mature cryptozoites infect cells of the lymphoid macrophage system, giving rise to metacryptozoites. exflagellation. final stage of microgamete formation, whereby the microgametocyte cell divides into eight small, motile, threadlike gametes in the lumen of the vector midgut gametocyte. also (rarely) gamont; haploid germ cell formed by gametogony in host erythrocytes. Mature gametocytes within erythrocytes that are ingested by vectors exit the host cells in the vector midgut and undergo gametogenesis to form micro- and macrogametes gametogenesis. process by which mature gametocytes produce gametes by mitotic cell division. For haemosporidians, this is accomplished in the lumen of the vector midgut, and is initiated in part by changes in the concentrations of respiratory gases in the blood. gametogony. form of schizogony (agamic division) by which merozoites in host erythrocytes develop into mature gametocytes gamont. see gametocyte macrogamete. female gamete; formed by mitotic cell division (gametogenesis) from a macrogametocyte. Fertilization by the motile microgamete results in the development of a diploid zygote macrogametocyte. female gametocyte; formed by gametogony in host erythrocytes. Each mature macrogametocyte is capable of forming a single macrogamete by gametogenesis megalomeront. also megaloschizont; second generation of exoerythrocytic meront in the asexual development of some leucocytozoid haemosporidians; characterized by extremely large size and close association with host cells, forming host cell– parasite complexes. Megalomeronts arise when syncytia resulting from merogony of hepatic meronts enter the blood stream and penetrate macrophages and other reticuloendothelial cells. Megalomeronts frequently accumulate in the host spleen and lymph nodes, where they generate additional merozoites, some of which invade leukocytes, there to develop into gametocytes. Other merozoites produced

128

by megalomeronts perpetuate the cycle of exoerythrocytic merogony in the spleen, lymphatic system, and liver. Particularly large exoerythrocytic meronts characteristic of certain species of haemoproteid haemosporidians, and arising in tissues outside the lungs, are also referred to as megalomeronts. megaloschizont. see megalomeront merogony. form of schizogony (agamic division) undergone by meronts in host cells and resulting in the production of numerous merozoites. In the Plasmodiidae, merogony can be classified as erythrocytic or exoerythrocytic. Exoerythrocytic merogony is further categorized as primary (or preerythrocytic; involving cryptozoites and metacryptozoites) or secondary (or posterythrocytic; involving phanerozoites). meront. also schizont; agamic, multinuclear cellular form of Apicomplexan parasites; formed by maturation of sporozoites after initial host infection or of merozoites after merogony. Rupturing of the meront results in the release of many merozoites. See also cryptozoite, megalomeront, metacryptozoite, trophozoite, and phanerozoite. merozoite. motile distributional stage of Apicomplexan asexual life stages; released by meronts following merogony taking place in the host cells. Merozoites circulate to other fixed tissues or to blood cells, where they are capable of inducing additional rounds of agamic growth and division or initiating gametogony. metacryptozoite. second-generation meront appearing in the primary exoerythrocytic merogony series of the Plasmodiidae; typically develop in the cells of the lymphoid macrophage system. Metacryptozoites contain large numbers of merozoites, which are released when the cell reaches maturity. Merozoites produced by metacryptozoites are capable of inducing additional rounds of primary exoerythrocytic merogony in the macrophage system (resulting in additional metacryptozoites), secondary exoerythrocytic merogony in the endothelial cells of capillaries (resulting in the formation of phanerozoites), or erythrocytic merogony in cells of the erythrocytic series (resulting in the formation of trophozoites and erythrocytic meronts). microgamete. male gamete; formed by mitotic cell division (gametogenesis) from a microgametocyte. Fertilization of the macrogamete results in the development of a diploid zygote. microgametocyte. male gametocyte; formed by gametogony in host erythrocytes. Each mature microgametocyte is capable of forming eight thin, motile microgametes by gametogenesis and subsequent exflagellation oocyst. final stage of zygote development; form from ookinetes in the epithelial layer lining the vector gut. The final stages of oocyst maturation complete the process of sporogony, resulting in the rupturing of the cell and the release of many sporozoites.

129 ookinete. spindle-shaped motile stage of the Apicomplexan zygote developing shortly after fertilization of the macrogamete by the microgamete; migrates from the midgut into the epithelial lining of the digestive tract, where it encapsulates itself in the basal lamina and undergoes additional development as an oocyst phanerozoite. form of meront involved in secondary exoerythrocytic merogony of the Plasmodiidae; typically develop in capillary endothelia of various organs, including the brain. Merozoites resulting from schizogony of phanerozoites are partially responsible for maintaining chronic parasitemia, as well as for generating the increase in parasitemia associated with recrudescence. prevalence. the proportion of individuals within a population that are infected; the apparent prevalence of an infection may differ from the true prevalence due to sampling bias or poor specificity or incomplete sensitivity of the diagnostic schizont. see meront schizogony. general term referring to the process of multiple fission typical of agamic reproduction in the Apicomplexa; a form of cell division whereby a single parent cell produces multiple uninuclear daughter cells. In the Apicomplexa, schizogony is further classified as sporogony (producing sporozoites), gametogony (producing gametocytes), or merogony (producing merozoites). The term schizogony is sometimes also used in specific reference to the latter process. sporogony. a form of schizogony (agamic division) by which the oocyst, while encapsulated in the lining of the vector midgut, develops and divides into numerous sporozoites sporozoite. motile, uninuclear cellular form typical of the infective stage of Apicomplexan parasites; formed by sporogony occurring in the oocyst. After formation, sporozoites migrate to the vector’s salivary glands, from whence they are introduced into the host bloodstream when the vector takes a blood meal. In the host, sporozoites migrate via the circulatory system to penetrate hepatocytes and other cells of the fixed tissues, where they give rise to schizonts. syncytia. multinuclear parasitic cell fragments resulting from merogony of hepatic meronts in the Leucocytozoidae. Phagocytosis of syncytia by host macrophages leads to the development of megalomeronts. trophozoite. nonfissionable, growth-stage precursor to erythrocytic meronts leading up to erythrocytic merogony. Trophozoites arise in host erythrocytes following invasion by merozoites from the second generation of exoerythrocytic merogony. After completing growth, trophozoites transition to erythrocytic meronts, which produce merozoites by erythrocytic merogony.