Final Report Summary - PATHOGEN NETWORKS (The spread of pathogens in social networks: a field study)
Project context and objectives:
The majority of theoretical models of disease evolution assume that populations of host and pathogen are well mixed (“mean-field” models). In such populations all individuals have an equal likelihood of encountering infection. In many populations however, this assumption is unrealistic because the number of potential contacts of a given host is finite, which leads to heterogeneity in the risk of infection among individuals. Spatial heterogeneity and the local nature of social interactions can have profound effects on the transmission and persistence of diseases.
Studying how diseases spread within social networks in wild populations would be particularly informative because in these populations mortality is often high and prevalence is not affected by any artificial intervention (e.g. drugs, vaccines). In most avian species, small-scale local social interactions (e.g. territoriality, winter grouping) are combined with larger-scale mobility, which makes birds good models to study the influence of social network structure on disease transmission. Cavity-nesting birds have the additional advantage of being easily accessed during reproduction, so that experimentations and monitoring of breeding populations can be carried out on large sample sizes.
Our research project combines disease ecology, microbial ecology and the analysis of social networks to explore the transmission routes of pathogens in populations of great tits (Parus major) and blue tits (Cyanistes cæruleus). More specifically, this project has the double aim of testing the extent to which social interactions among individuals affect the transmission of pathogens, and of using social network data to acquire knowledge on when and where individuals get infected. We focused on two kinds of pathogens. First, using recent molecular techniques we applied these questions to whole bacterial communities living in/on birds, with the double objective of understanding the dynamics of transmission of bacteria in wild populations and of acquiring basic knowledge on the structure and effects of bacterial communities in bird populations. Second, we also focused on avian malarial parasites (genus Plasmodium), which have been extensively studied in recent years and are particularly well characterized at the study site.
Work performed and main results
To achieve our objectives, our work consisted of a combination of (i) fieldwork to sample bacterial communities from birds, tag them using PIT-tags in order to characterize their social interactions, and monitor breeding individuals; (ii) laboratory work to extract and genotype bacterial communities using the 16s-23s rDNA intergenic spacer region; and (iii) statistical analysis of data to investigate how infection patterns and bacterial community structure relates to social behaviour.
During the 22 months of this project, we collected bacterial communities from the mouth, wing feathers and feet of 600 individual birds, and contributed to the characterisation of both social interactions and breeding performance of birds in two large study populations located in Wytham and Bagley woods, near Oxford. This required four periods of fieldwork (two winter and two intensive spring periods). The study of bacterial communities carried by birds required a preliminary phase of optimisation of our DNA extraction and genotyping protocols, after which we started to genotype a first set of samples. However during the autumn/winter 2011 we were faced with an unexpected issue. Despite previous experience of this technique, and appropriate measures taken to ensure clean laboratory conditions, our negative PCR controls regularly ended up positive, which pointed to contamination of the PCR mix with bacterial DNA. In addition, amplification success was low and quite unrepeatable when using low amounts of template DNA, which was the case for quite a number of our DNA samples.
Routine genotyping was therefore temporarily stopped and priority was given to solving this issue. From the medical and forensics literature we learned that many commercially available PCR reagents (e.g. Taq polymerase) are likely to contain bacterial DNA, which could both contaminate PCR results and hamper PCR sensitivity. Based on this literature we developed and tested a new protocol including an extra ‘decontamination’ step using restriction enzymes prior to the start of the PCR. We collected human test samples in order to compare the efficiency of various enzymes, enzyme concentrations and enzyme inactivation methods in PCR decontamination. Given the time spent solving this issue, we judged it necessary to publish it, so that other scientists in this field who might face the same issue could use our approach and save time. We submitted a manuscript describing the approach for publication in Microbial Ecology, which is currently under revision.
Once this contamination issue was solved, lab work resumed and we genotyped bacterial communities from six different groups of eight blue tits caught in the wild before being placed in aviaries for a one-week period, during which their social interactions were measured. By comparing bacteria carried by those birds both before and after their stay in captivity, we are investigating how bacterial communities change when birds are in close contact with each other, and which bacteria (phylotypes) are transmitted during social interactions. Our ultimate aim is then to use breeding data to explore the effects of those bacteria on the breeding performance and / or survival of birds in the wild. This part of the study relies on extensive, time-consuming data analysis and is still ongoing.
From the analysis of bacterial communities carried by wild blue tits, we gained insight into some consistent patterns found across different groups of birds. Oral bacterial communities are by far the most diverse (up to 80 different phylotypes detected) and differ markedly from both wing feathers and feet communities, which are very similar to each other, and vary very little across various locations within populations. In contrast, bird oral communities differ spatially, and can therefore be used to discriminate different groups of birds.
In parallel with the study of bacterial communities, we also investigated how infection by two species of malarial parasites (Plasmodium circumflexum and Plasmodium relictum) relates to the social behaviour of great tits and their location within the study populations. A striking fact in the Wytham study population is that, although these parasites are vector-borne, infection patterns (type and prevalence) are strongly and consistently structured at a small spatial scale. Indirect evidence suggests that infection occurs at some point between fledging and first breeding attempt the following spring but there is no clear evidence as to when and where birds get infected.
Our results show contrasting, opposite relationships between the two malarial parasites and the feeding behaviour of great tits. Irrespective of their survival probability (all birds included in our dataset survived during the whole duration of the study), birds infected with P. circumflexum subsequently visited feeders less often, and visited a lower number of feeding locations than uninfected birds, while birds infected with P. relictum visited feeders more often, and a higher number of feeders than uninfected birds. These results are in line with those of recent studies demonstrating species-specific, contrasted fitness effects of those two parasites in the ecologically similar blue tit. They emphasize the need for considering parasite communities rather than single species when studying disease ecology in wild populations.
We also investigated the links between winter-feeding behaviour and infection patterns during the following spring. Here again, we found contrasting results: great tits infected with P. circumflexum had visited feeders more often, while those infected with P. relictum had visited feeders less often over the previous autumn/winter than uninfected individuals. This suggests a potential link between the presence of the birds at common feeders and the probability of P. circumflexum infection. We are now exploring this link further and incorporating more detailed analysis of social interactions between birds, as well as refined, month-per-month analysis to better understand the spatiotemporal effects of Plasmodium infection in great tits.
Expected final results
The analysis of the data collected during the duration of the project is still in progress. Ultimately we will be able to determine how similarity in bacterial communities relates to social interactions among birds, and better understand which bacterial types are transmissible via direct contact. Using detailed data on winter location and behaviour of the birds, as well as data on their breeding performance, we will explore what fitness effects those transmissible bacteria may have on wild blue tits and great tits. This might open the way to further studies aiming, for example, at identifying with more accuracy (e.g. via sequencing) those bacteria that might display pathogenic effects, thereby providing insights into the complex field of microbial disease ecology in wild populations.
The results obtained so far also suggest links between the behaviour of wild great tits and their infection by malarial parasites of the genus Plasmodium. We will continue the analysis on a more refined basis in order to better understand why two close species of Plasmodium seem to affect their hosts’ behaviour in opposite ways, what effects this might have on their probability of infection and, more generally, on the evolution of host-parasite interactions when more than a single host – and parasite species are involved.
The majority of theoretical models of disease evolution assume that populations of host and pathogen are well mixed (“mean-field” models). In such populations all individuals have an equal likelihood of encountering infection. In many populations however, this assumption is unrealistic because the number of potential contacts of a given host is finite, which leads to heterogeneity in the risk of infection among individuals. Spatial heterogeneity and the local nature of social interactions can have profound effects on the transmission and persistence of diseases.
Studying how diseases spread within social networks in wild populations would be particularly informative because in these populations mortality is often high and prevalence is not affected by any artificial intervention (e.g. drugs, vaccines). In most avian species, small-scale local social interactions (e.g. territoriality, winter grouping) are combined with larger-scale mobility, which makes birds good models to study the influence of social network structure on disease transmission. Cavity-nesting birds have the additional advantage of being easily accessed during reproduction, so that experimentations and monitoring of breeding populations can be carried out on large sample sizes.
Our research project combines disease ecology, microbial ecology and the analysis of social networks to explore the transmission routes of pathogens in populations of great tits (Parus major) and blue tits (Cyanistes cæruleus). More specifically, this project has the double aim of testing the extent to which social interactions among individuals affect the transmission of pathogens, and of using social network data to acquire knowledge on when and where individuals get infected. We focused on two kinds of pathogens. First, using recent molecular techniques we applied these questions to whole bacterial communities living in/on birds, with the double objective of understanding the dynamics of transmission of bacteria in wild populations and of acquiring basic knowledge on the structure and effects of bacterial communities in bird populations. Second, we also focused on avian malarial parasites (genus Plasmodium), which have been extensively studied in recent years and are particularly well characterized at the study site.
Work performed and main results
To achieve our objectives, our work consisted of a combination of (i) fieldwork to sample bacterial communities from birds, tag them using PIT-tags in order to characterize their social interactions, and monitor breeding individuals; (ii) laboratory work to extract and genotype bacterial communities using the 16s-23s rDNA intergenic spacer region; and (iii) statistical analysis of data to investigate how infection patterns and bacterial community structure relates to social behaviour.
During the 22 months of this project, we collected bacterial communities from the mouth, wing feathers and feet of 600 individual birds, and contributed to the characterisation of both social interactions and breeding performance of birds in two large study populations located in Wytham and Bagley woods, near Oxford. This required four periods of fieldwork (two winter and two intensive spring periods). The study of bacterial communities carried by birds required a preliminary phase of optimisation of our DNA extraction and genotyping protocols, after which we started to genotype a first set of samples. However during the autumn/winter 2011 we were faced with an unexpected issue. Despite previous experience of this technique, and appropriate measures taken to ensure clean laboratory conditions, our negative PCR controls regularly ended up positive, which pointed to contamination of the PCR mix with bacterial DNA. In addition, amplification success was low and quite unrepeatable when using low amounts of template DNA, which was the case for quite a number of our DNA samples.
Routine genotyping was therefore temporarily stopped and priority was given to solving this issue. From the medical and forensics literature we learned that many commercially available PCR reagents (e.g. Taq polymerase) are likely to contain bacterial DNA, which could both contaminate PCR results and hamper PCR sensitivity. Based on this literature we developed and tested a new protocol including an extra ‘decontamination’ step using restriction enzymes prior to the start of the PCR. We collected human test samples in order to compare the efficiency of various enzymes, enzyme concentrations and enzyme inactivation methods in PCR decontamination. Given the time spent solving this issue, we judged it necessary to publish it, so that other scientists in this field who might face the same issue could use our approach and save time. We submitted a manuscript describing the approach for publication in Microbial Ecology, which is currently under revision.
Once this contamination issue was solved, lab work resumed and we genotyped bacterial communities from six different groups of eight blue tits caught in the wild before being placed in aviaries for a one-week period, during which their social interactions were measured. By comparing bacteria carried by those birds both before and after their stay in captivity, we are investigating how bacterial communities change when birds are in close contact with each other, and which bacteria (phylotypes) are transmitted during social interactions. Our ultimate aim is then to use breeding data to explore the effects of those bacteria on the breeding performance and / or survival of birds in the wild. This part of the study relies on extensive, time-consuming data analysis and is still ongoing.
From the analysis of bacterial communities carried by wild blue tits, we gained insight into some consistent patterns found across different groups of birds. Oral bacterial communities are by far the most diverse (up to 80 different phylotypes detected) and differ markedly from both wing feathers and feet communities, which are very similar to each other, and vary very little across various locations within populations. In contrast, bird oral communities differ spatially, and can therefore be used to discriminate different groups of birds.
In parallel with the study of bacterial communities, we also investigated how infection by two species of malarial parasites (Plasmodium circumflexum and Plasmodium relictum) relates to the social behaviour of great tits and their location within the study populations. A striking fact in the Wytham study population is that, although these parasites are vector-borne, infection patterns (type and prevalence) are strongly and consistently structured at a small spatial scale. Indirect evidence suggests that infection occurs at some point between fledging and first breeding attempt the following spring but there is no clear evidence as to when and where birds get infected.
Our results show contrasting, opposite relationships between the two malarial parasites and the feeding behaviour of great tits. Irrespective of their survival probability (all birds included in our dataset survived during the whole duration of the study), birds infected with P. circumflexum subsequently visited feeders less often, and visited a lower number of feeding locations than uninfected birds, while birds infected with P. relictum visited feeders more often, and a higher number of feeders than uninfected birds. These results are in line with those of recent studies demonstrating species-specific, contrasted fitness effects of those two parasites in the ecologically similar blue tit. They emphasize the need for considering parasite communities rather than single species when studying disease ecology in wild populations.
We also investigated the links between winter-feeding behaviour and infection patterns during the following spring. Here again, we found contrasting results: great tits infected with P. circumflexum had visited feeders more often, while those infected with P. relictum had visited feeders less often over the previous autumn/winter than uninfected individuals. This suggests a potential link between the presence of the birds at common feeders and the probability of P. circumflexum infection. We are now exploring this link further and incorporating more detailed analysis of social interactions between birds, as well as refined, month-per-month analysis to better understand the spatiotemporal effects of Plasmodium infection in great tits.
Expected final results
The analysis of the data collected during the duration of the project is still in progress. Ultimately we will be able to determine how similarity in bacterial communities relates to social interactions among birds, and better understand which bacterial types are transmissible via direct contact. Using detailed data on winter location and behaviour of the birds, as well as data on their breeding performance, we will explore what fitness effects those transmissible bacteria may have on wild blue tits and great tits. This might open the way to further studies aiming, for example, at identifying with more accuracy (e.g. via sequencing) those bacteria that might display pathogenic effects, thereby providing insights into the complex field of microbial disease ecology in wild populations.
The results obtained so far also suggest links between the behaviour of wild great tits and their infection by malarial parasites of the genus Plasmodium. We will continue the analysis on a more refined basis in order to better understand why two close species of Plasmodium seem to affect their hosts’ behaviour in opposite ways, what effects this might have on their probability of infection and, more generally, on the evolution of host-parasite interactions when more than a single host – and parasite species are involved.