|
nif:isString
|
-
On 25 September 2012, we downloaded 11,870 distinct avian records with subtype data (HA and NA gene segments) from the Influenza Virus Database (http://www.ncbi.nlm.nih.gov/genomes/FLU/Database/nph-select.cgi?go=database) in GenBank [8], date collected from 1902 through 2012. Records were classified into wild (including migratory birds), domestic (including poultry and farmed), feral, captive (including birds in trade, in zoos, and pets) or unknown, based on a variety of resources including GenBank records, GenBank linked publications, sampling location, and species distributions. Flyways were defined according to the North American Flyway Directory [9] and a country-based division into North America (USA and Canada only), Latin America, Europe, Africa, Asia, and Australasia. Maps were created in ArcGIS 9.3 (ESRI, Inc.) and Photofiltre 6.5.1. We used the GenBank data to examine the relationship between the detection of a particular AIV subtype from domestic birds and host range of that subtype in wild birds (host genus richness). We used a generalized linear model with a binomial distribution and a logit link function to calculate the odds ratio of isolating a particular subtype in domestic birds (presence in domestic birds) based on observed wild bird host genus richness. We adjusted for effort by including the number of GenBank records for each subtype (effort). Statistical analysis and figures were completed in R version 2.15.3 [10]. Thus, for each HA/NA subtype we modeled: f(presence in domestic birds) = α + β1 (wild bird host genus richness) + β2 (effort). b. We collected published and non-published surveillance efforts that non-discriminately tested for AIV subtype. Many surveillance programs and studies did not meet the criteria, such as those that screened samples by real-time reverse-transcriptase PCR (RT-PCR) and only subtyped those samples that tested positive for specific subtypes (e.g. H5 or H7). Full descriptive information (including associated references, location, prevalence, detected AIV richness, sampling and testing methodology, number of sampling periods, and sample years) for each of these studies can be found in the electronic supplementary material (Supplementary Tables S1 and S2). Location, year of sampling, and authorship were tracked to avoid duplicate reporting, resulting in non-overlapping studies from the Northern Hemisphere (n = 41) and the Southern Hemisphere (n = 9). These studies relied on virus isolation or RT-PCR methods for AIV detection in cloacal, fecal, or tracheal samples. Virus isolates were further characterized by HA or NA inhibition assays, subtype specific RT-PCR, or by sequencing the HA and NA gene segments of the virus isolate. Viruses partially subtyped (those with only HA or NA subtype) were not included in the analysis. Studies with at least five sampling periods and 5,000 birds tested overall were identified and nonparametric AIV subtype richness was predicted for each study (Table1). The baseline effort measure (5,000 birds) was established to focus on studies with the largest comparable sample sizes. We restricted the sampling periods to maintain consistency of effort among studies in addition to sample size, and to limit the analysis to a manageable number of major studies. We also predicted cumulative unique subtype richness from GenBank by year from 1959 to 2012.
Table data removed from full text. Table identifier and caption: 10.1371/journal.pone.0090826.t001 Overview of seven studies with at least 5,000 birds collected over at least five sampling events. 1Unpublished data provided by Martin Gilbert 10 February 2013.2Richness estimated based on data thru 2007, population prevalence based on data through 2009.3Sampling method was not reported but based on historical sampling patterns suggest it was cloacal.4Provided by Vincent Munster 21 December 2012. These data excluded Ottenby Mallard data reported under Sweden.5Extended data provided by Meng-Chu Cheng 12 November 2012. Bird families based on published data from Cheng et al. 2010.Positive samples used to calculate prevalence (positive/total) may not all have been fully subtyped.
We used EstimateS v 8.2.0 [11] to generate a presence-absence accumulation function of subtypes and calculated the nonparametric estimate of subtype richness with 95% confidence intervals using the Chao2 estimate and 50 randomizations with replacement [6]. Bias-corrected Chao2 was calculated unless the coefficient of variation for the incidence distribution was less than 0.5, in which case Classic Chao2 was calculated. We applied Chao's nonparametric estimator of sufficient sampling to calculate the minimum number of birds necessary to detect 75% of the estimated asymptotic subtype richness [12]. A 75% target was selected because reaching the asymptote is problematic [13]. Datasets are staged on the Knowledge Network for Biocomplexity repository (http://doi.org/10.5063/F1HT2M7Q).
ii. We examined the attributes of the 50 studies and identified variables associated with richness that could be reliably extracted and analyzed as covariates. Our attempts to isolate measures of host diversity (percentages of Anseriformes and Charadriiformes) were hampered by data availability reducing the number of studies to 41, but we were able to extract AIV prevalence and duration of study (years). We used linear mixed models (R library lme4, function lmer) to examine the effects of interactions and to estimate the variance of subtype richness associated with region and selected the best model based on Bayesian Information Criterion (Supplementary Table 3).
c. Sampling methods and ethic statements for data provided by co-authors: Samples from Sweden were collected from wild ducks at an important stopover site in the island of Öland (56°12′N 16°24′E) located in the Northwest European flyway [14]. Breeding grounds of the duck populations using the site are Baltic countries and Northwestern Russia [15]. Ducks were caught using a live-duck trap and all handling of birds was performed by trained ornithologists from Ottenby Bird Observatory. Samples were collected in transport media [16] and kept frozen at −70°C until analysis [17]. The sampling protocol was approved by Linköping Animal Research Ethics Board (permit numbers 8–06, 34–06, 80–07, 111–11, 112–11) in accordance with national legislation. In the European Union, expert ornithologists trapped birds using duck decoys, duck traps, wader funnel traps, mist nets, clap nets, cannon nets, or Helgoland traps. The majority of samples were obtained from migratory birds during fall migration at long-term sampling sites in the Netherlands [18]. Cloacal swabs were collected using sterile cotton swabs and stored in transport medium [16] and shipped to the laboratory where they were stored at −80°C for analysis. The handling of birds within the European Union study was in accordance with national and international guidelines that were approved by an independent Animal Ethics Committee of the Erasmus Medical Center (Stichting DEC Consult) under permit number 122-09-20. Mongolian sample collections focused on environmental fecal samples, negating the need for national permits for the capture and handling of wild birds. Sampling took place on state-owned land in 27 locations in the East region of Mongolia, four in the North-Central region and three in the West region, based on the nomenclature used in Gilbert et al. 2012 [19]. The project was approved by the University of Minnesota, Institutional Animal Care and Use Committee (Protocol 1006A84613). Work was completed under the authorization of the Mongolian State Central Veterinary Laboratory.
|
![[This page as RDF]](http://data.gesis.org/somesci/static/rdf-icon.gif){kind=icon}