|Home | Archives | About | Login | Submissions | Subscribe | Contact | Search|
Copyright © 2010 by the author(s). Published here under license by The Resilience Alliance.
Go to the pdf version of this article The following is the established format for referencing this article:
Caron, A., M. De Garine-Wichatitsky, N. Gaidet, N. Chiweshe, and G. S. Cumming. 2010. Estimating dynamic risk factors for pathogen transmission using community-level bird census data at the wildlife/domestic interface. Ecology and Society 15(3): 25. [online] URL: http://www.ecologyandsociety.org/vol15/iss3/art25/
Research, part of Special Feature on Risk mapping for avian influenza: a social-ecological problem Estimating Dynamic Risk Factors for Pathogen Transmission Using Community-Level Bird Census Data at the Wildlife/Domestic Interface
1UPR AGIRs, Department ES, Cirad, Harare Zimbabwe, 2UPR AGIRs, Department ES, Cirad, Montpellier France, 3Mammal Research Institute, Department of Entomology and Zoology, University of Pretoria, Pretoria, South Africa, 4UPR AGIRs, Deparmtent ES, Cirad, Harare, Zimbabwe, 5UPR AGIRs, Department ES, Cirad, Montpellier, France, 6UPR AGIRs, Department ES, Cirad, Harare, Zimbabwe, 7Percy Fitz-Patrick Institute, DST/NRF Center of Excellence, University of Cape Town
The ecology of host species is crucial for understanding the mechanisms of pathogen transmission and spread in complex multi-host systems. In this article, we use detailed observations of the host community to develop and apply a new approach to mapping temporal variation in risk for avian influenza. Working in an extensive wetland system near Harare, Zimbabwe, we use the overlap in space and time of highly variable bird communities, combined with ecological risk factors, to assess the risk of Avian Influenza viruses (AIV) maintenance and transmission between bird populations. The estimated introduction and maintenance risks associated with waterfowl populations at a given time are then multiplied by the level of interactions with neighboring domestic production systems during the same period. This approach is used to develop hypotheses for the dynamics of the introduction and circulation of AIV strains in waterfowl populations and as a way of understanding the potential role of “bridge” species at the wild/domestic interface. The novel approach presented here offers a potentially useful way to explore AIV risk, identify which wild bird species may be acting as reservoirs or vectors of pathogens at a local scale, and improve local surveillance.
Key words: Avian influenza; bridge species; community ecology; risk factor; wild/domestic interface
The success of multi-host infectious pathogens in ecosystems is heavily dependent on the composition of the community of organisms in which they occur (Ostfeld 2009). The species composition of the host community and the temporal dynamics of its constituent populations will influence pathogen success through variation in such parameters as host susceptibility, host abundance, host population turn-over, the presence and absence of reservoir species, and encounter rates between hosts and pathogens (Dwyer et al. 1997, Childs et al. 2007, Borer et al. 2009).
For pathogens that are transmissible either by direct contact or via the shared use of the same habitat at different times, transmission parameters often cannot be directly measured in the field. Doing so is particularly difficult for multi-host pathogens. Transmission is usually evaluated through host-pathogen models (Breban et al. 2009, Rohani et al. 2009) that lack direct measurements of actual interspecies contact. Epidemiological interactions (i.e., ecological interactions that may result in the transmission of a pathogen) between susceptible, infected, and recovered hosts can be used to define a network from which to explore transmission pathways and assess spatial and temporal variation in transmission risks (Takeuchi and Yamamoto 2006, Duerr et al. 2007, Kenah and Robins 2007). While graph theoretic methods for creating and analyzing networks from direct data on species interactions are fairly well established (Williams et al. 2002, Lafferty et al. 2008), the application of standard network methods in cases where interactions and mechanisms must be inferred from higher-level data on co-occurrences is poorly developed and computationally challenging (Bascompte and Melian 2005, Rabbat et al. 2008).
Here we consider the use of co-occurrence data to infer avian influenza virus (AIV) potential transmission pathways in communities of birds in Zimbabwe. The recent HPAI H5N1 (Highly Pathogenic AI) panzootic has spread across the world, exploiting avian communities and sporadically infecting humans (Webster et al. 2007). The mechanisms of AI spread across ecosystems are still unclear. The international poultry trade and waterfowl migration are the two most intensively tested hypotheses that have been proposed to explain patterns of HPAI spread (Olsen et al. 2006). However, existing information implies different roles for different modes of dispersal across regions, indicating a need for regional or subregional research frameworks (Kilpatrick et al. 2006). The epidemiology of LPAI (Low Pathogenic AI) is better understood than that of HPAI: waterfowl are considered to be the primary reservoirs of LPAI with spill-over to domestic poultry occurring periodically (Webster et al. 1992). These cross-species transmission events can lead to HPAI selection in domestic populations (Caron et al. 2009). We use AIV as a complex multi-host pathogen model with a potentially high impact on the socio-economic level for Africa and the world.
The importance of the ecology of wild birds in the epidemiology of AIV strains has been underlined by numerous studies (Olsen et al. 2006, Stallknecht and Brown 2007, McCallum et al. 2008, Munster and Fouchier 2009), but the high diversity of potential host species and a lack of information on their susceptibilities to LPAI and HPAI makes the overall picture unclear (Perkins and Swayne 2002, 2003, Brown et al. 2006, Brown et al. 2007a, Pasick et al. 2007). Some key features of waterbird ecology are thought to strongly facilitate virus maintenance or spread. These features include: their relatively high degree of inter- and intra-specific mixing; their tendency to move long distances during annual migrations and/or broad-scale nomadic movements; their colonial feeding and roosting habits; and their use of water, which improves viral survival outside the host. Some studies have already used these criteria to estimate hotspots of potential virus infection, regional spread, or inter-continental contamination (Kilpatrick et al. 2006, Veen et al. 2007, Cumming et al. 2008). However, at a local level, most AI risk factors show seasonal variation as species breed and as they respond to variations in resource availability, rainfall, the presence or absence of other species (including pathogens), and seasonal changes in human behavior. The corresponding variation in AI risk has not been thoroughly analyzed in wild bird communities.
In addition to the many uncertainties regarding spatiotemporal variation in transmission pathways, it is worth noting that most current field research still follows traditional distinctions: veterinarians investigate the health of domestic species and ornithologists focus on wild birds, but the gap between these two approaches is poorly filled.
The classical one-pathogen approach aims at detecting (directly or not) the pathogen in different hosts and inferring transmission pathways that are specific to this pathogen (Plowright et al. 2008). In this article we present a novel approach to assessing transmission risks in a complex epidemiological network that consists of spatiotemporally variable bird communities (i.e., waterbirds, domestic birds, and bridge species that interact with both wild and domestic communities). Rather than attempting to develop a formal network-based model, we integrate data on the frequency and intensity of inter- and intraspecific co-occurrences, together with information about relevant aspects of species ecology and behavior, to obtain a risk score for each species in the community and to build an adapted risk assessment model. In conceptual terms, this approach offers a mid-point between data-intensive, mechanistic network analysis (Takeuchi and Yamamoto 2006) and looser, more subjective assessments of risk (Cumming et al. 2008, Peterson and Williams 2008). Our approach has the advantage that it incorporates aspects of fine-scale transmission mechanisms while not being excessively data-demanding; the analysis is undertaken using the kinds of survey data that standard ornithological censusing procedures typically yield. In addition to presenting a useful picture of seasonal variation in AI risk, our analysis demonstrates how dynamic aspects of risk can still be included into epidemiological risk assessment in the absence of detailed pair-by-pair interaction data.
We undertook this study in the Manyame catchment (30°30’30’’, 17°45‘45’’), located 35 km West of Harare, the capital city of Zimbabwe. Our primary study sites were two impoundments, Lake Chivero and Lake Manyame, both of which were created in 1952. Together they form a linked wetland system (connected by the Manyame River) that harbors a community of waterfowl species. Part of the shoreline of Lake Chivero is a protected area. In addition, several commercial farms are located in the Manyame catchment, including industrial poultry farms and semi-extensive ostrich farms. Farm employees living in compounds located on the farm estates also raise backyard chickens for domestic consumption.
We considered the different avian communities in the study area to be four ‘compartments’ as defined in Caron et al. 2009: (1) the waterfowl compartment, consisting of the community of wild waterbird species sharing the lake habitat through the year; (2) the industrial compartment, being the population of domestic chickens raised in buildings at high densities for a period of about 40 days; (3) the backyard chicken compartment, in which chicken populations are free-ranging during the day, using fields and human-modified natural habitats in the vicinity of compounds, and resting in chicken pens at night; and (4) the ostrich farm compartment, consisting of a few hundred birds kept in open paddocks (usually around a hundred birds per paddock) surrounded by wooden fences. These different management practices result in variable contacts between domestic poultry and the surrounding wild bird communities, and biosecurity measures are implemented in intensive poultry and ostrich farms.
It is important to note that the status of AIV in this ecosystem is unknown. No H5N1 outbreaks have been recorded south of the equator in Africa. Outbreaks of H5N2 in the southern part of Zimbabwe in ostrich farms occurred in 2005, which had a possible link with outbreaks of the same strain in South Africa in 2004 (Sinclair et al. 2005, Abolnik et al. 2006).
The methodology followed a six-step process: (1) identify the hazard in relation to the objective; (2) describe the waterfowl and domestic bird communities; (3) define dynamic and non-dynamic ecological risk factors (RFs) for the presence of AIV infection in the waterfowl community; (4) combine RFs for both the release assessment (introduction risk - IR) and the exposure assessment (maintenance risk - MR) in the waterfowl community; (5) identify epidemiological interactions between waterfowl and domestic compartments through direct contact, indirect contact via shared habitat or potential bridge species; and (6) estimate the release assessment for each of the domestic compartments (transmission from waterfowl to the domestic compartments) through a dynamic Domestic Risk variable (DR).
The risk of AIV introduction from the waterfowl community to the three domestic compartments is dependent on the introduction of strains in the waterfowl community, the ability of this community to maintain such strains and the potential for spill-over from the waterfowl compartment to the domestic compartments. This risk increases from non-H5 and H7 LPAI (which can still recombine with other strains to produce HP strains) to H5 and H7 LPAI (which are the most likely strains to evolve into HPAI) to already high pathogenicity (HP) strains (including HPAI H5N1). Because of the little epidemiological information available for African bird species and because any AIV strain could be involved in the creation of HP strains, we identified all AIV strains as hazardous for this risk assessment.
Focal counts were undertaken to estimate species diversity and the abundance of waterfowl and domestic communities. Based on local knowledge of the field site, 15 shoreline sites were selected for their high diversity of waterfowl species and abundance of birds. From May 2007 to March 2009, bird community counts were carried out every two months at each of these sites. Four 30-minute counts, each at a different time of the day (06:00-09:00; 09:00-12:00; 12:00-15:00; and 15:00-18:00) were carried out in a random sequence at each site for each recording session. Prior to each count, the counter waited for 10 minutes to habituate the birds to the presence of the counter. During each count, the counter stood or sat at a distance of 30-50 m from the lake shore and recorded all birds in a 150-m-radius semicircle.
In a radius of ten kilometers from the lake shoreline, we selected 19 domestic compartment sites, located in (or in direct proximity to) production units (buildings, paddocks, or villages). Six sites in three different ostrich farms, seven sites in intensive poultry farms and six sites in villages with backyard poultry were selected. At each of these domestic sites, the same counting protocol (10-minute wait plus 30-minute count) was applied from June 2008 to April 2009 with both wild and domestic birds being counted.
Ecological Risk Factors (RFs)
The use of variables that capture ecologically relevant variation to build epidemiological RFs has been applied in different studies related to AIV maintenance and spread at regional or continental scales (Kilpatrick et al. 2006, Veen et al. 2007, Cumming et al. 2008). We were interested in describing, at the community level, the risk of introduction and maintenance of AIV within and between bird communities across seasons in response to variability in host ecology. We thus developed dynamic RFs based on seven ecological variables that were likely to influence the epidemiology of AIV, including two variables for the introduction risk (estimated local immigration and risk related to AIV strain in relation to the origin of the birds) and five variables for the maintenance risk (the overall abundance of birds, the gregariousness of the species, interspecific aggregation, percentage of juveniles and feeding habits) (Stallknecht et al. 1990b, Olsen et al. 2006). These risk factors were characterized as RFs 1 to 7 (Table 1). Susceptibility to AIV infection and the immunological status of the birds were not considered in this model because of a lack of information for African bird species.
Introduction and maintenance risk (IR and MR) in waterfowl community
For each bird community count, the species values of each RF were multiplied by 1 for species recorded at least once, and by 0 when the species was absent. IR and MR were calculated as described in Table 2. IR was calculated for any AIV strain and specifically for HPAI H5N1 in order to display the proportion of the relative risk for exposing the community to HPAI H5N1 introduction.
The standard deviation of each RF of the MR was calculated. A Spearman Rank Correlation test was performed for each RF in relation to the MR in order to assess their relative contributions.
Quantifying epidemiological interactions (domestic risk - DR) and their risks
Calculating the degree of ecological interaction between wild and domestic compartments
Each waterfowl count session was paired with count sessions carried out in domestic compartments (separated by a maximum of three weeks). The community composition of each domestic compartment was calculated using the same method as for the waterfowl compartment. For each domestic compartment, all species seen during the same session in the waterfowl compartment were identified as the shared community. We calculated the proportion of the shared community for each domestic compartment potentially in contact with the waterfowl compartment during the same period.
Calculating the interaction risk (DR) of the shared community
For each species recorded in the waterfowl and domestic compartments during the same period, the DR was calculated as described in Table 2. For each session, we estimated the DR (i.e., of AIV spreading from the waterfowl to the domestic compartments) by summing the DR of all species in the shared community.
Dynamics of waterfowl community
Variation in waterbird numbers observed across the two years was characterized by a peak during the end of the cold-dry season and running into the hot-dry season (July-September-November; Figure 1). This peak resulted from two general trends: (1) the concentration of nomadic sub-Saharan waterfowl (Dendrocygnidae and Anatidae) on larger bodies of water as seasonal wetlands within the subregion dried down; and (2) the return of paleartic migrants from Europe during the (European) fall migration. The paleartic migrants leave the ecosystem between March and April, at the end of the Zimbabwean rainy season. Species diversity was highest during the dry season and lowest during the rainy season in both years. Note that there are no paleartic migrant duck species in southern Africa (Cumming et al. 2008).
Dynamics of domestic communities
The birds observed per family and the species diversity of the three domestic compartments are shown in Figure 2. For each of the domestic compartments, domestic species dominated the counts. Intensive poultry represented 98% of all birds observed in the intensive poultry compartment, backyard chickens represented 25% of the backyard compartment, and ostrich represented 79% of the ostrich compartment. The remaining bird community in each of the domestic compartments was quite homogenous across the compartments, dominated by birds from the Ploceidae, Estrildidae, Ardeidae, Columbidae and Hirundinidae families, which represent between 59 and 67% of the birds observed. The maximum number of birds observed in these three communities was in April, mainly due to an increase in Ploceidae, particularly the red-billed queleas (Quelea quelea). This species is considered a pest species by local farmers and exhibits high variability in population dynamics. Species diversity varied between the three compartments as well as seasonally, particularly in the ostrich compartment.
Patterns of IR and MR in the waterfowl compartment and relation to the RFs
IR peaked in September 2008 and July 2009 (Figure 3). MR peaked in November in both years (Figure 4). For both risks, there was a difference in the intensity of the peak between the two years, which correlated with the variability in waterfowl abundance (Figure 1). The trends of all five RFs followed the MR with one major peak per year during or slightly before the dry season (Figure 4). The “Feeding” RFs had a slightly advanced peak in July or September depending on the year. Some RFs, such as “Mixing” and “Juvenile,” had a higher variability than others (Table 3). Three RFs had a significant correlation with the MR curve (Gregariousness, Abundance and Feeding, in decreasing order). The results are consistent with a higher risk of the presence of AIV strains in the waterfowl community at the end of the dry season. Waterfowl species contributing the most to IR and MR are presented respectively in Tables 4 and 5. IR for AIV and HPAI H5N1 was dominated by Charadriiformes (35.8% and 47.0%, respectively) and Anseriformes (37.1% and 33.7%, respectively). MR was largely dominated by Anseriformes.
Variation in DR for the three domestic compartments
The intensive poultry and ostrich farm DR curves were similar (Figure 5), with two peaks of similar amplitude: one in November, the other in March. For the backyard poultry curve, only the peak in March was evident. The DR curves for the intensive poultry and ostrich compartments, and the MR curve for the waterfowl, showed the highest risk during the month of November (end of the hot-dry season). In our model, this month had the highest risk for transmission of AIV strains from the waterfowl to the domestic compartment. The second peak observed in the intensive poultry and ostrich farm DR curves was not related to a peak in the waterfowl MR. There was consistency in the most represented families for each of the three domestic compartments (Table 6).
Our results provide a clear illustration of the ways in which community-level risk varies over time, both within and between years. IR peaked during the early hot-dry season, when regional waterbirds were concentrating on larger water bodies and migrants began to arrive from Europe. By contrast, MR peaked in November at the end of the dry season when the largest waterbird concentrations were observed. A number of bridge species were shared between different epidemiological compartments, suggesting a strong potential for interactions between domestic and wild birds in this system.
On the use of dynamic risk factors
We are not aware of any previous studies that have attempted to track variations in community-level risk factors through time. Although we worked primarily with indicators rather than with empirical proof of pathogen transmission, it is important to remember that community ecology and epidemiology have been used in combination for the last 25 years to explore and understand the behavior of multi-host and/or multi-pathogen systems (Holt and Pickering 1985, Hudson and Greenman 1998). A solid body of empirical evidence suggests that the availability of hosts, their movements, and their interactions with other hosts will influence pathogen transmission (Morgan et al. 2006, Bordes et al. 2009). Intra- and inter-species mixing, the presence or absence of particular species, and the proportion of juveniles in the population vary seasonally for waterfowl and are important influences on the ecology of infectious diseases (Wallensten et al. 2007). There is therefore a lot of evidence-based support for the a priori definition of RFs that take into account the ecology of hosts and the ways in which host ecology may influence the behavior of pathogens in a system. At the same time, it is important to note that these RFs remain hypotheses until such time as further data on influenza occurrence within the system become available.
The development of dynamic RFs in previous studies has primarily focused on differences between summer and winter bird communities. Bimonthly risk mapping presents a finer-scale and considerably more informative pattern. Despite the high quality of our count data, however, a number of parameters used in this analysis remain difficult to estimate. For example, the immigration RF assumes that the arrival in the counts of new birds represents a risk for AIV introduction; in reality, numbers could stay constant while individuals change, and a proportion of the birds arriving in the system may be coming from nearby areas. For some bird species (e.g., red-billed teal Anas erythroryncha and white-faced duck Dendrocygna viduata), movement patterns are estimated from scarce ring recovery data. Often, the proportion of the population undertaking nomadic vs. trans-equatorial movements is unknown (Underhill et al. 1999). This information is important for estimating a risk of introduction (according to different AIV strains) but cannot be taken into account in our model (Cumming et al. 2008). Dispersal is particularly crucial for the two species mentioned above because they constitute some key species identified by the IR. Environmental RFs could have been taken into account in this model. In the Manyame catchment, measurements of water temperature at various seasons averaged 21.08°C (n=70; min 14.85°C; max 25.4°C, Caron, unpublished data); this supports the idea that the environment may be a potential reservoir throughout the year (with better conditions for virus survival during May-August) using data from recent studies (Brown et al. 2007b, Brown et al. 2008, Weber and Stilianakis 2008).
MR is calculated without weighting the RFs because there is no empirical evidence from which to argue that one RF is more important than another. With suitable data collection and sampling for influenza viruses, it may eventually be possible to use linear models to weight different risk factors. Another important assumption used in this analysis is that birds seen within the counting area are potentially in contact. This assumption may not truly reflect fine-scale non-randomness in interaction networks.
Anseriformes and Charadriiformes represent the main families identified for IR, the first mainly as a function of their numbers and the second by their potential risk in introducing dangerous strains. Charadriiformes, mainly paleartic waders, but also Anseriformes crossing the equator are identified by the model as potential introducers of HPAI H5N1. Interestingly, when waterfowl are ranked for each of the five RFs and the ranks are summed across the two years, the species contributing the most to the MR (Table 5) belong to the bird orders known to be reservoirs for LPAI strains (Anseriformes and Charadriiformes) with the two most influential species in the model, the white-faced duck and the red-billed teal, being the most abundant ducks in the system. The only other orders present in the 20 most important species were Gruiformes (Coot sp.) and Ciconiiformes (Egret and Ibis spp.). These orders and families have been found with, or dead of, LPAI or HPAI strains (Gauthier-Clerc et al. 2007, Hars et al. 2008, Stoops et al. 2009). Additionally, the MR curve (Figure 4) was consistent across the two years and indicated a maximum risk of AIV presence in the waterfowl community during the hot-dry season, when migratory and paleartic waterfowl are present in the system, coming from areas where AIV strains circulate. This result is consistent with a basic epidemiological model for AIV in Africa (Gaidet et al. 2006) that assumes a strong likelihood of introduction of strains during the paleartic migration. The fact that most of the RFs follow the MR trends reflects some consistency in the model: the high risk season for AIV presence in the waterfowl community derives from a convergence of peaks of RFs during this season. The “Gregariousness” and “Abundance” RFs have a high correlation with MR and an increase in the weight of these factors would accentuate the current trend in MR (Table 3).
The difference in MR between the two years reflects the differences in bird abundance. There is a relationship between lake level (determined by the rainfall in the previous year and human management) and bird abundance; the lakes dry down during the hot-dry season and exposed shorelines offer a muddy, vegetated, resource-rich habitat for dabbling ducks and waders. MR defined here could be predicted in advance with rainfall from the previous years, offering the potential for disease forecasting in this system. The use of environmental data to predict epidemiological patterns through an ecological (host or vector) link have already been demonstrated (Harvell et al. 2002).
Domestic Risk (DR) between waterfowl and domestic compartments
The trends in the DR curves for the three domestic compartments were different. The 19 domestic sites chosen varied between zero and ten kilometers from the lake shore, and this distance could have influenced the observed wild bird community. However, although the ostrich farm sites were the farthest from the lake shore, their DR curve followed the intensive poultry DR curve. There may be other factors besides distance to the lake that influence the wild bird community, including variation in artificial resource availability in the production buildings, farms and villages; natural resource availability; breeding sites; predation; and so on. The most likely explanation for the similar trends between intensive poultry and ostrich farm DRs is that they both used artificial feed, attracting specific bird communities, while backyard chickens forage for their own food like wild birds.
IR was not related to any peak of the DR. However, according to our model, there are always interactions between the waterfowl and domestic compartments. In a specific epidemiological situation (e.g., regional spread of a HP strain threatening the ecosystem), this IR could help to target surveillance and control measures during high interaction seasons. The fact that the highest DR curve for two domestic compartments coincided with the highest waterfowl MR is of interest (Figure 5). The end of the hot-dry season is a high risk period for these two domestic compartments, not only because the waterfowl community has the highest risk of harboring AIV strains but also because the epidemiological interactions between the compartments are at their highest. We can hypothesize that this period represents a hotspot for pathogen circulation and transmission between compartments (Jones et al. 2008). The second peak after the end of the rainy season (in March) was consistent for the three domestic compartments but was not linked with a peak in risk associated with the waterfowl community. However, the shared community of wild birds between the waterfowl community and the three domestic compartments was always high (Table 6) suggesting a year-long risk of pathogen transmission from the waterfowl compartment. The validity of the DR estimate is limited by its population-level approach; birds of the same species observed in two different compartments were assumed to belong to the same population. However, we cannot prove that they were indeed the same individuals beyond the fact that the study site is fairly small.
Validating the model and testing the bridge species hypothesis
In order to validate the global approach and the RFs used, long-term and intensive monitoring of waterfowl will be necessary. Community analyses based on bird census data, as presented here, can contribute to the development of specific hypotheses relating to AIV maintenance and spread in the system. The community level perspective is often missing in multi-host wild population studies (Yasue et al. 2006). Usually, access to wild individuals is difficult, technically biased or limiting, and for most capture protocols it is not possible to choose precisely the epidemiological sample composition and size (Wobeser 2002). By contrast, as this study demonstrates, bird count data can drive the sampling design and/or provide an indication of the representativeness of the samples obtained from the system.
To test hypotheses concerning the role of bridge species between waterfowl and the domestic compartments usually requires selective sampling among a broad range of avian diversity. More than 100 species in 25 families of birds have been detected dead or alive with AIV strains (Olsen et al. 2006). Some terrestrial birds have been found to harbor AIV strains and even HPAI H5N1 strains (Nestorowicz et al. 1987, Boon et al. 2007, Brown et al. 2009). We thus assumed that any wild bird species could be capable of harboring and transmitting AIV strains. Consideration of the families and species contributing the most to the peak DR for each domestic compartment (Table 6) shows that the first four families represent between 58% and 72% of the total of birds involved in the DRs. For each of these families, there is one species that represents between 50 and 99% of the birds observed. This unexpected result means that only a few species represent the bulk of the DR and that a targeted sampling focusing on these species will achieve not only a surveillance of the species most at risk of transmitting AIV but also an extensive coverage of the overall DR. Sampling protocols targeting these species should cast light on the role of potential bridge species between the waterfowl and domestic compartments. To our knowledge, there has not been sufficient local-scale testing of potential bridge species to characterize a bridge species community, despite some published suggestions (Veen et al. 2007) and an obvious missing link in HPAI outbreaks that have involved spatially segregated poultry and waterfowl.
The ultimate goal of this study was to integrate ecological and epidemiological data in a risk-mapping context (as discussed by Caron et al. 2009). The main outputs are a set of hypotheses that describe the mechanisms that generate patterns of AIV circulation in the waterfowl community and the role of bridge species between the waterfowl and the domestic compartments. Although we have focused on a one-way analysis (from the waterfowl compartment to each of the domestic compartments), the same analysis could be conducted for transmission between the four compartments in both directions.
An important advantage of our sampling protocol is that it provides the information that is needed to assess the adequacy of epidemiological sampling. This step is often missing in wildlife surveillance and decreases the validity of results. The next step will be to add to this data set an AIV prevalence layer (i.e., of wild and domestic compartments) to test the model and the bridge species hypotheses. The protocol described here is intensive but feasible. Its approach could easily be simplified and reproduced. In the context of AIV surveillance, a series of counts by ornithologists during suspected high-risk seasons would prepare the ground for targeted sampling. In some countries, this type of data is regularly collected by ornithological organizations and is therefore already available.
The strength of this research relative to traditional epidemiological analyses lies in its ecological dimensions. Although our model was designed with the ecology of AIV in mind, most pathogens with direct transmission will be dependent on the ecological traits estimated by the RFs (with some adjustments; e.g., “Feeding” RF). Can this risk factor analysis be extended to other pathogens to develop more ‘ecological’ predictions of disease risk? Such approaches may ultimately provide useful guidelines for surveillance in hotspots of disease emergence at the wildlife/domestic interface (Jones et al. 2008).
Responses to this article are invited. If accepted for publication, your response will be hyperlinked to the article. To submit a response, follow this link. To read responses already accepted, follow this link.
ACKNOWLEDGMENTSWe are grateful to the many people who assisted with the bird counts, particularly Fadzai Matzvimbo, Innocent Magunje, and Josphine Mundava. The Zimbabwe Wildlife Authority and Gary Stafford kindly gave us permission to work in areas under their jurisdiction. This work was conducted within the framework of the “Mesures d’Urgence” and GRIPAVI projects and the Research Platform Producion and Conservion in Partnership (RP-PCP). It benefited from funds from the French Ministry of Foreign Affairs. Additional funding support was provided by the USAID through the Wildlife Conservation Society’s GAINS (Global Avian Influenza Network for Surveillance) programme. We thank two referees for their comments and suggestions.
Abolnik, C., E. Cornelius, S. P. R. Bisschop, M. Romito, and D. Verwoerd. 2006. Phylogenetic analyses of genes from South Africa LPAI viruses isolated in 2004 from wild aquatic birds suggests introduction by Eurasian migrants. Pages 189-199 in OIE/FAO International Scientific Conference on Avian Influenza, Basel, Karger, Switzerland.
Bascompte, J., and C. J. Melian. 2005. Simple trophic modules for complex food webs. Ecology 86:2868-2873.
Boon, A. C. M., M. R. Sandbulte, P. Seiler, R. J. Webby, T. Songserm, Y. Guan, and R. G. Webster. 2007. Role of terrestrial wild birds in ecology of influenza A virus (H5N1). Emerging Infectious Diseases 13:1720-1724.
Bordes, F., S. Morand, D. A. Kelt, and D. H. Van Vuren. 2009. Home range and parasite diversity in mammals. American Naturalist 173:467-474.
Borer, E. T., C. E. Mitchell, A. G. Power, and E. W. Seabloom. 2009. Consumers indirectly increase infection risk in grassland food webs. Proceedings of the National Academy of Sciences of the United States of America 106:503-506.
Breban, R., J. M. Drake, D. E. Stallknecht, and P. Rohani. 2009. The role of environmental transmission in recurrent avian influenza epidemics. PLoS Computational Biology 5:e1000346.
Brown, J. D., D. E. Stallknecht, J. R. Beck, D. L. Suarez, and D. Swayne. 2006. Susceptibility of North American ducks and gulls to H5N1 highly pathogenic avian influenza viruses. Emerging Infectious Diseases 12:1663-1670.
Brown, J. D., D. E. Stallknecht, R. D. Berghaus, and D. E. Swayne. 2009. Infectious and lethal doses of H5N1 highly pathogenic avian influenza virus for house sparrows (Passer domesticus) and rock pigeons (Columbia livia). Journal of Veterinary Diagnostic and Investigation 21:437-445.
Brown, J. D., D. E. Stallknecht, S. Valeika, and D. E. Swayne. 2007a. Susceptibility of Wood Ducks to H5N1 highly pathogenic avian influenza virus. Journal of Wildlife Diseases 43:660-667.
Brown, J. D., D. E. Swayne, R. J. Cooper, R. E. Burns, and D. E. Stallknecht. 2007b. Persistence of H5 and H7 avian influenza viruses in water. Avian Dis 50:285-289.
Caron, A., N. Gaidet, M. de Garine-Wichatitsky, S. Morand, and E. Z. Cameron. 2009. Evolutionary biology, community ecology and avian influenza research. Infection Genetics and Evolution 9:298-303.
Childs, J. E., J. A. Richt, and J. S. Mackenzie. 2007. Introduction: conceptualizing and partitioning the emergence process of zoonotic viruses from wildlife to humans. Pages 1-31 in J. E. Childs, J. S. Mackenzie, and J. A. Richt, editors. Wildlife and emerging zoonotic diseases: the biology, circumstances and consequences of cross-species transmission. Springer, Heidelberg, Germany.
Cumming, G. S., P. A. R. Hockey, L. W. Bruinzeel, and M. A. Du Plessis. 2008. Wild bird movements and avian influenza risk mapping in southern Africa. Ecology and Society 13(2): 26. [online] URL: http://www.ecologyandsociety.org/vol13/iss2/art26/.
Duerr, H. P., M. Schwehm, C. C. Leary, S. J. De Vlas, and M. Eichner. 2007. The impact of contact structure on infectious disease control: influenza and antiviral agents. Epidemiology and Infection 135:1124-1132.
Dwyer, G., J. S. Elkinton, and J. P. Buonaccorsi. 1997. Host heterogeneity in susceptibility and disease dynamics: tests of a mathematical model. American Naturalist 150:685-707.
Gaidet, N., T. Dodman, A. Caron, G. Balança, S. Desvaux, F. Goutard, G. Cattoli, W. Hagemeijer, and F. Monicat. 2006. Influenza A viruses in waterbirds in Africa. Emerging Infectious Diseases 13:626-629.
Gauthier-Clerc, M., C. Lebarbenchon, and F. Thomas. 2007. Recent expansion of highly pathogenic avain influenza H5N1: a critical review. Ibis 149:202-214.
Hars, J., S. Ruette, M. Benmergui, C. Fouque, J. Y. Fournier, A. Legouge, M. Cherbonnel, B. Daniel, C. Dupuy, and V. Jestin. 2008. The epidemiology of the highly pathogenic H5N1 avian influenza in Mute Swan (Cygnus olor) and other Anatidae in the Dombes region (France), 2006. Journal of Wildlife Diseases 44:811-823.
Harvell, C. D., C. E. Mitchell, J. R. Ward, S. Altizer, A. P. Dobson, R. S. Ostfeld, and M. D. Samuel. 2002. Climate warming and disease risks for terrestrial and marine biota. Science 296:2158-2162.
Hockey, P. A. R., W. R. J. Dean, and P. G. Ryan. 2005. Roberts - Birds of Southern Africa. John Voelcker Bird Book Fund, Cape Town, South Africa.
Holt, R. D., and J. Pickering. 1985. Infectious disease and species coexistence: a model of Lotka-Voltera form. American Naturalist 126:196-211.
Hudson, P., and J. Greenman. 1998. Competition mediated by parasites: biological and theorotical progress. Trends in Ecology and Evolution 13:387-390.
Jones, K. E., N. G. Patel, M. A. Levy, A. Storeygard, D. Balk, J. L. Gittleman, and P. Daszak. 2008. Global trends in emerging infectious diseases. Nature 451:990-994.
Kenah, E., and J. M. Robins. 2007. Network-based analysis of stochastic SIR epidemic models with random and proportionate mixing. Journal of Theoretical Biology 249:706-722.
Kilpatrick, A. M., A. A. Chmura, D. W. Gibbons, R. C. Fleischer, P. P. Marra, and P. Daszak. 2006. Predicting the global spread of H5N1 avian influenza. Proceedings of the National Academy of Sciences of the United States of America 103:19368-19373.
Lafferty, K. D., S. Allesina, M. Arim, C. J. Briggs, G. De Leo, A. P. Dobson, J. A. Dunne, P. T. J. Johnson, A. M. Kuris, D. J. Marcogliese, N. D. Martinez, J. Memmott, P. A. Marquet, J. P. McLaughlin, E. A. Mordecai, M. Pascual, R. Poulin, and D. W. Thieltges. 2008. Parasites in food webs: the ultimate missing links. Ecology Letters 11:533-546.
McCallum, H. I., D. A. Roshier, J. P. Tracey, L. Joseph, and R. Heinsohn. 2008. Will Wallace’s Line save Australia from avian influenza? Ecology and Society 13(2): 41. [online] URL: http://www.ecologyandsociety.org/vol13/iss2/art41/.
Morgan, E. R., M. Lundervoldb, G. F. Medleyb, B. S. Shaikenovc, P. R. Torgersond, and E. J. Milner-Gullande. 2006. Assessing risks of disease transmission between wildlife and livestock: the saiga antelope as a case study. Biological Conservation 131:244-254.
Munster, V. J., and R. A. Fouchier. 2009. Avian influenza virus: of virus and bird ecology. Vaccine 27:6340-6344.
Nestorowicz, A., Y. Kawaoka, W. J. Bean, and R. G. Webster. 1987. Molecular analysis of the hemagglutinin genes of Australian H7N7 influenza viruses: role of passerine birds in maintenance or transmission? Virology 160:411-418.
World Organisation for Animal Health (OIE). 2009. Update on avian influenza in animals (type H5). http://www.oie.int/downld/AVIAN%20INFLUENZA/A_AI-Asia.htm.
Olsen, B., V. J. Munster, A. Wallensten, J. Waldenstrom, A. D. Osterhaus, and R. A. Fouchier. 2006. Global patterns of influenza A virus in wild birds. Science 312:384-388.
Ostfeld, R. S. 2009. Biodiversity loss and the rise of zoonotic pathogens. Clinical Microbiology and Infection 15:40-43.
Pasick, J., Y. Berhane, C. Embury-Hyatt, J. Copps, H. Kehler, K. Handel, S. Babiuk, K. Hooper-McGrevy, Y. Li, M. Q. Le, and L. S. Phuong. 2007. Susceptibility of Canada Geese (Branta canadensis) to highly pathogenic avian influenza virus (H5N1). Emerging Infectious Diseases 13:1821-1827.
Perkins, L. E., and D. E. Swayne. 2002. Pathogenicity of a Hong Kong-origin H5N1 highly pathogenic avian influenza virus for emus, geese, ducks, and pigeons. Avian Diseases 46:53-63.
Perkins, L. E., and D. E. Swayne. 2003. Comparative susceptibility of selected avian and mammalian species to a Hong Kong-origin H5N1 high-pathogenicity avian influenza virus. Avian Diseases 47:956-967.
Peterson, A. T., and R. A. J. Williams. 2008. Risk mapping of highly pathogenic avian influenza distribution and spread. Ecology and Society 13(2): 15. [online] URL: http://www.ecologyandsociety.org/vol13/iss2/art15/.
Plowright, R. K., S. H. Sokolow, M. E. Gorman, P. Daszak, and J. E. Foley. 2008. Causal inference in disease ecology: investigating ecological drivers of disease emergence. Frontiers in Ecology and the Environment 6:420-429.
Rabbat, M. G., M. A. T. Figueiredo, and R. D. Nowak. 2008. Network inference from co-occurrences. Ieee Transactions on Information Theory 54:4053-4068.
Rohani, P., R. Breban, D. E. Stallknecht, and J. M. Drake. 2009. Environmental transmission of low pathogenicity avian influenza viruses and its implications for pathogen invasion. Proceedings of the National Academy of Sciences of the United States of America 106:10365-10369.
Sinclair, M., G. K. Bruckner, and J. J. Kotze. 2005. Avian Influenza in ostriches: epidemiological investigation in the Western Cape Province of South Africa. Elsenburg Journal 2:2-4.
Stallknecht, D. E., and J. D. Brown. 2007. Wild birds and the epidemiology of avian influenza. Journal of Wildlife Diseases 43:S15-S20.
Stallknecht, D. E., S. M. Shane, M. T. Kearney, and P. J. Zwank. 1990a. Persistence of avian influenza viruses in water. Avian Diseases 34:406-411.
Stallknecht, D. E., S. M. Shane, P. J. Zwank, D. A. Senne, and M. T. Kearney. 1990b. Avian influenza viruses from migratory and resident ducks of coastal Louisiana. Avian Diseases 34:398-405.
Stoops, A. C., K. A. Barbara, M. Indrawan, I. N. Ibrahim, W. B. Petrus, S. Wijaya, A. Farzeli, U. Antonjaya, L. W. Sin, N. Hidayatullah, I. Kristanto, A. M. Tampubolon, S. Purnama, A. Supriatna, T. H. Burgess, M. Williams, S. D. Putnam, S. Tobias, and P. J. Blair. 2009. H5N1 surveillance in migratory birds in Java, Indonesia. Vector Borne Zoonotic Diseases 9:695-702.
Takeuchi, F., and K. Yamamoto. 2006. Effectiveness of realistic vaccination strategies for contact networks of various degree distributions. Journal of Theoretical Biology 243:39-47.
Underhill, L. G., A. J. Tree, H. D. Oschadleus, and V. Parker. 1999. Review of ring recoveries of waterbirds in Southern Africa. University of Cape Town, Cape Town, South Africa.
Veen, J., J. Brouwer, P. Atkinson, C. Bilgin, J. Blew, S. Eksioglu, M. Hoffmann, R. Nardelli, F. Spina, C. Tendi, and S. Delany. 2007. Ornithological data relevant to the spread of avian influenza in Europe (phase2): further identification and first field assessment of higher risk species. Wetlands International, Wageningen, The Netherlands.
Wallensten, A., V. J. Munster, N. Latorre-Margalef, M. Brytting, J. Elmberg, R. A. Fouchier, T. Fransson, P. D. Haemig, M. Karlsson, A. Lundkvist, A. D. M. E. Osterhaus, M. Stervander, J. Waldenstrom, and B. Olsen. 2007. Surveillance of influenza A virus in migratory waterfowls in northern Europe. Emerging Infectious Diseases 13:404-411.
Weber, T. B., N. I. Stilianakis. 2008. Inactivation of influenza A viruses in the environment and modes of transmission: a critical review. Journal of Infection 57:361-373
Webster, R. G., W. J. Bean, O. T. Gorman, T. M. Chambers, and Y. Kawaoka. 1992. Evolution and ecology of influenza A viruses. Microbiological Reviews 56:152-179.
Webster, R. G., D. J. Hulse-Post, K. M. Sturm-Ramirez, Y. Guan, M. Peiris, G. Smith, and H. Chen. 2007. Changing epidemiology and ecology of highly pathogenic avian H5N1 influenza viruses. Avian Diseases 50:269-272.
Williams, R. J., E. L. Berlow, J. A. Dunne, A. L. Barabasi, and N. D. Martinez. 2002. Two degrees of separation in complex food webs. Proceedings of the National Academy of Sciences of the United States of America 99:12913-12916.
Wobeser, G. 2002. New and emerging diseases—the wildlife interface. Canadian Veterinary Journal 43:798.
Yasue, M., C. J. Feare, L. Bennun, and W. Fiedler. 2006. The epidemiology of H5N1 avian influenza in wild birds: why we need better ecological data. BioScience 56:923-929.
|Home | Archives | About | Login | Submissions | Subscribe | Contact | Search|