| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 US Geological Survey, National Wildlife Health Center, 6006 Schroeder Road, Madison, Wisconsin 53711, USA
2 US Geological Survey, Wisconsin Cooperative Wildlife Research Unit, Department of Wildlife Ecology, 204 Russell Labs, 1630 Linden Drive, Madison, Wisconsin 53706, USA
3 Corresponding author (email: dblehert{at}usgs.gov)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Pasteurella multocida isolates are classified based on serologic antigen presentation, and there are 16 somatic serotypes designated 1 though 16 (Rhoades and Rimler, 1991). Most pathogenic P. multocida isolates cultured from wild waterfowl in the Pacific, Central, and Mississippi flyways of North America are serotype 1 (Botzler, 1991). There have been significant annual and geographic fluctuations in the patterns of avian cholera mortality among wild bird species, and factors influencing the initiation and the course of an avian cholera outbreak are not well understood (Rosen, 1969; Wobeser, 1992; Blanchong et al., 2006; Samuel et al., 2007).
To gain further epidemiological insight into the dynamics of avian cholera transmission and disease spread among wild birds, it is necessary to develop techniques to differentiate P. multocida isolates of the same serotype. The use of DNA-based techniques provides this ability (Owen, 1989), but to date studies demonstrating the usefulness of these techniques for understanding the epidemiology of avian cholera in wild birds have been limited (Wilson et al., 1995a, b; Samuel et al., 2003b; Samuel et al., 2007).
A more recently developed genomic fingerprinting technique, fluorescent amplified fragment length polymorphism (AFLP) analysis, provides a greater capacity than previous techniques to identify polymorphic regions within a genome (Vos et al., 1995) and has been applied to pasteurellae epizootics in domestic birds (Amonsin et al., 2002). Amplified fragment length polymorphism analysis is based on the detection of polymorphic restriction fragments by selective polymerase chain reaction (PCR) and provides the capacity to examine in excess of 1,000 DNA markers per isolate (Vos et al., 1995). To conduct AFLP analysis, genomic DNA is digested with two restriction enzymes, followed by ligation of adapters of defined DNA sequence to the ends of the restriction fragments. Ligation products are reamplified by PCR using primers that include a selective nucleotide at their 3'-end to reduce the complexity of AFLP patterns (Geert et al., 1996). A fluorescent label is incorporated during selective PCR to facilitate analysis of AFLP reaction products using an automated DNA sequencing instrument. Amplified fragment length polymorphism analysis generates reproducible, complex patterns, referred to as genetic fingerprints, that are useful for distinguishing closely related organisms.
The objective of this study was to evaluate the utility of AFLP analysis as an epidemiological tool to distinguish P. multocida serotype 1 isolates based on regional and temporal genetic characteristics. Fifty-three serotype 1 isolates cultured during a laboratory challenge of Mallards (Anas platyrhynchos; Samuel et al., 2003a) and 120 serotype 1 isolates cultured from wild birds and environmental samples during avian cholera outbreaks were examined. Analysis of the AFLP data revealed that the isolates were distinguishable.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Amplified fragment length polymorphism analysis was applied to a collection of P. multocida serotype 1 isolates from the US Geological Survey–National Wildlife Health Center diagnostic culture collection and from field studies on avian cholera epidemiology (Samuel et al., 2003b). A collection of serotype 1 isolates collected during a 15-wk laboratory challenge study of captive-reared mallard ducks (Samuel et al., 2003a) was also analyzed. The challenge study was conducted as follows (for additional study details, see Samuel et al., 2003a): Six-to-eight-wk-old male Mallards were divided into groups consisting of 30 birds each, and each group was housed in a separate isolation room (designated 4, 5, and 6). Within each room birds were divided into three groups of 10; two groups were challenged with P. multocida inocula of different environmental origins, and the third group was unchallenged. Hence, six different P. multocida inoculae (designated 1, 2, 3, 4, 7, and 8) were used. The inocula, all originally isolated from environmental samples collected in central California (Samuel et al., 2003a), were utilized as follows: inocula 1 and 2 (room 6), inocula 3 and 4 (room 5), inoculae 7 and 8 (room 4). Pasteurella multocida was cultured from tissues and/or swabs collected periodically from live birds during the course of the study, from birds that died, and following euthanasia of birds at the end of the study. During the study some birds inoculated with P. multocida cleared the organism, while some uninoculated birds became infected.
AFLP reaction preparation
Genomic DNA was extracted from P. multocida isolates using the Puregene DNA purification kit (Gentra Systems Inc., Minneapolis, Minnesota, USA) according to the manufacturer’s instructions. Genomic DNA concentration and purity were evaluated by UV spectroscopy and agarose gel electrophoresis. Amplified fragment length polymorphism reactions were prepared as described by Vos et al. (1995). Genomic DNA (500 ng) was digested with the restriction enzymes EcoRI and HpyCH4 IV for 3 hr and ligated, using T4 ligase (New England Biolabs, Beverly, Massachusetts, USA), to EcoRI (5'-CTC GTA GCT GCG TAC C-3' plus 3'-CAT CTG ACG CAT GGT TAA-5') and HpyCH4 IV (5'-GAC GAT GAG TCC TGA G-3' plus 3'-TAC TCA GGA CTC GC-5') double-stranded adapters. The digested and ligated DNA was then diluted fivefold in a 1x solution of Taq DNA polymerase reaction buffer without MgCl2 (Promega, Madison, Wisconsin, USA) prior to amplification by preselective PCR using the EcoRI and HpyCH4 IV primers 5'-GAC TGC GTA CCA ATT C-3' and 5'-GAT GAG TCC TGA GCG T-3', respectively, in a final volume of 50 µl. PCR reactions contained 10 µl diluted ligation mixture, 15 pmol each EcoRI and HpyCH4 IV primers, 4 µl 40 mM dNTP mix, and 2.5 U Taq polymerase in storage buffer B (Promega). Preselective PCR amplification conditions were an initial extension cycle at 72 C for 60 sec, followed by 35 cycles of denaturation at 94 C for 50 sec, annealing at 56 C for 1 min, and extension at 72 C for 2 min. Preselective PCR products were then diluted 10-fold with double-deionized water, and 3 µl of the diluted preselective mixture were used as template for selective amplification. Selective PCR reactions contained 15 pmol fluorescein phosphoramidite (FAM)–labeled EcoRI+G selective primer (5'-FAM-GAC TGC GTA CCA ATT CG-3'), 25 pmol unlabeled HpyCH4 IV+A selective primer (5'-GAT GAG TCC TGA GCG TA-3'), 3 µl 40 mM dNTP mix, and 1.25 U Taq polymerase in storage buffer B (Promega) in a final reaction volume of 25 µl. Selective PCR amplification conditions were 35 cycles of denaturation at 94 C for 50 sec, annealing at 56 C for 1 min, and extension at 72 C for 2 min, with a final extension at 72 C for 10 min. Selective PCR reaction products were diluted fivefold with double-deionized water, and fragment analysis was conducted at the University of Wisconsin–Madison Biotechnology Center using an Applied Biosystems (Foster City, California, USA) 3730 automated capillary DNA sequencing instrument.
AFLP data analysis
Amplified fragment length polymorphism profiles were normalized and aligned to one another with respect to an internal sizing standard using BioNumerics 3.5 (Applied Maths Inc., Austin, Texas, USA) software’s proprietary algorithm. Continuous AFLP fingerprint data, in the form of densitometry distributions, were then analyzed directly without attempting to identify discrete bands because discretization in a nonarbitrary manner is not possible. Once AFLP densitometry patterns were aligned, a Pearson product moment correlation matrix was computed for all pairwise combinations, exported as a text file, and imported into SAS statistical software (SAS Institute Inc., Cary, North Carolina, USA). The SAS PROC CLUSTER was used to produce the dendrograms, using the average linkage (also known as the unweighted pair group method with arithmetic mean [UP-GMA]; Sokal and Michener, 1958) and WARD (Ward, 1963) options for comparison. The UPGMA and WARD dendrogram outputs were virtually the same, and only UPGMA dendrograms are reported. SAS PROC PRINCOMP was used to conduct a principal components analysis (PCA; Johnson and Wichern, 1982) using the Pearson product moment correlation matrix. Principal components analysis uses a different algorithm than dendrogram analysis to evaluate the data, providing a complementary assessment. The second and third principal components were used for the PCA. Although the first principal component explained the majority of the variation in the data, the first component is essentially the mean axis and typically reflects the average signal strength of the fingerprint. Signal strength largely reflects sample preparation attributions and therefore is not usually an informative biologic axis on which to align samples. Fisher’s 2-tailed exact test was used to determine whether specific groups of samples fell disproportionately on different branches of the dendrograms. Analysis of variance (ANOVA) followed by pairwise comparisons of the means was used to test whether the principal components scores differed between groups of interest. Values of P
0.05 were considered significant.
To understand the resolution AFLP analysis in distinguishing closely related bacteria, a single genomic DNA preparation from a P. multocida serotype 1 isolate was used to conduct 10 replicate AFLP analyses. All aspects of AFLP reaction processes were conducted independently. A Pearson product moment correlation matrix comparing the resulting AFLP patterns indicated an average correlation of 0.92 (SD=0.05) among the replicate samples.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Amplified fragment length polymorphism analysis was completed for the six P. multocida inocula and for 53 isolates cultured from Mallards during the infection trial. Of the 53 isolates examined, 21, 12, and 20 were cultured from birds housed in rooms 4, 5, and 6, respectively. Amplified fragment length polymorphism profiles for each P. multocida isolate were used to construct a dendrogram, consisting of four major branches, illustrating genetic relationships among the isolates (Fig. 1
). Isolates cultured from birds that died (filled symbols) allocated disproportionately to dendrogram branch 1, while isolates obtained from live, apparently healthy birds allocated disproportionately to branches 2 and 4 (open symbols). The distribution of P. multocida isolates cultured from dead birds was distinct from those cultured from apparently healthy live birds (P<0.001). Analysis of AFLP data using PCA (components 2 and 3; Fig. 2) also supported the division of bacterial isolates into the two major groups observed on the dendrogram, one consisting of isolates from dead birds (filled symbols; quadrant IV), and the other consisting of isolates from live, apparently healthy birds (open symbols; quadrant I; P<0.001 for principal component 2). Further, PCA indicated that room 6 bacterial isolates (circles) allocated disproportionately to the right-hand quadrants of the plot (Fig. 2) compared to room 4 and room 5 isolates (diamonds and squares, respectively; P<0.001 for principal component 3).
|
|
Analysis of P. multocida serotype 1 isolates from wild birds and environmental samples
One-hundred-twenty serotype 1 isolates from a variety of wild bird species (72 total; Table 1) and environmental samples (48 total; Table 2) were examined using the AFLP technique. Isolate collection locations included California (89 isolates), Hawaii (one isolate), Iowa (two isolates), Missouri (one isolate), Nebraska (14 isolates), Utah (12 isolates), and Wisconsin (one isolate). The predominant bird species from which P. multocida was cultured included Eared Grebes (Podiceps nigricollis; 20 isolates), Snow Geese (Chen caerulescens; 13 isolates), and Ruddy Ducks (Oxyura jamaicensis; 12 isolates). To facilitate the comparison of bacterial isolates by collection date, the following categories were established: January 1981 to March 1997 (20 isolates), November 1997 to April 1998 (62 isolates), October 1998 to March 1999 (20 isolates), and January 2000 to October 2004 (18 isolates). These categories were determined based on the availability of isolates surrounding significant avian cholera mortality events from 1997 to 1999. The largest group of bird and environmental samples was collected in central California (61 isolates) during the winter of 1997/1998 (54 of the 61 isolates).
|
|
), the isolates segregated into four major clusters (labeled 1–4) consisting of 12 branches (labeled A–L). Cluster 1 (branches A and B) contained six isolates (three bird and three environmental) from California (three isolates) and Nebraska (three isolates; Fig. 3a). Cluster 2 (branches C–F) contained 47 isolates (38 bird and nine environmental) from California (27 isolates), Iowa (one isolate), Missouri (one isolate), Nebraska (nine isolates), Utah (eight isolates), and Wisconsin (one isolate; Figs. 3a and 3b). Cluster 3 (branches G–K) contained 66 serotype 1 isolates (31 bird and 35 environmental) from California (59 isolates), Iowa (one isolate), Nebraska (two isolates), and Utah (four isolates; Figs. 3a and 3c). A serotype 3 type strain was located on divergent branch K within cluster 3. A single serotype 1 isolate cultured from a Hawaiian Coot (Fulica americana alai) was the only member of cluster 4 (branch L).
|
). Collection location categories analyzed included central California, the Salton Sea in southern California, the Midwest (Iowa, Missouri, Nebraska, and Wisconsin, collectively), and the Great Salt Lake in Utah. Central California isolates (51/61) allocated disproportionately to cluster 3 (Figs. 3a and 3c; P<0.001). Salton Sea (19/25) and Midwestern isolates (12/18) allocated disproportionately to cluster 2 (Figs. 3a and 3b; P<0.001). The majority of Great Salt Lake isolates (8/12) allocated to cluster 2, but this result was not determined to be significant (P=0.22). Analysis of isolates based on collection dates indicated that isolates collected between January 1981 and March 1997 (14/20) allocated disproportionately to cluster 2 (Figs. 3a and 3b; P<0.001). Isolates collected between November 1997 and April 1998 (50/62) allocated disproportionately to cluster 3 (Figs. 3a and 3c; P<0.001). Although the majority of isolates collected between October 1998 and March 1999 (12/20) and between January 2000 and October 2004 (11/18) allocated within cluster 2, these groupings were not determined to be significant (P=0.16 and 0.21, respectively).
Principal components analysis of the wild bird and environmental isolate AFLP data using components 2 and 3 supported the allocation of isolates into the same branch-based clusters observed on the dendrogram (data not shown). When examined based on collection location, PCA supported the existence of a visually distinct cluster of isolates from central California (Fig. 4, open circles; P<0.001 for component 2). This same cluster, consisting primarily of isolates cultured from samples collected between November 1997 and April 1998, was also apparent by PCA when samples were identified based on collection date (P<0.001 for component 2; data not shown). Statistical comparison of the contributions of regional (P<0.001 for component 2) and temporal (P=0.026 for component 2) genetic characteristics of the isolates to the observed clusters indicated that both exerted significant contributions.
|
). Examination of the central California isolates based on specific collection locations (16 sites distributed among the Sacramento and San Joaquin Valley Refuge Complexes) indicated that although branch H isolates were cultured from samples collected at both the Sacramento and San Joaquin Valley Refuge Complex sites, branch I isolates were only cultured from samples collected at San Joaquin Valley Refuge Complex sites (Fig. 5). Furthermore, branch I isolates were exclusively cultured during the 1997/1998 season (Tables 1 and 2). In contrast, consistent with their broader local distribution within central California, branch H isolates were also identified at collection sites outside of central California, including the Salton Sea, the Great Salt Lake, and the Mid-western United States (Tables 1 and 2 and Fig. 3c). Collection dates for all branch H isolates ranged from 1996 to 2004.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
A pattern revealed by both dendrogram (Fig. 1) and principal components (Fig. 2) analyses of the infection trial AFLP data was the presence of a cluster of P. multocida isolates cultured from live birds distinct from those cultured from dead birds. We hypothesize that this pattern resulted from genetic drift leading to divergence of the P. multocida isolates cultured from live birds later in the trial compared to isolates cultured from dead birds during the first 2 wk. Additionally, segregation of all trial 6 isolates along with inoculum 2 from the other inocula and isolates was consistent with the determination that 19 of 20 trial 6 isolates likely descended from inoculum 2. Thus, the ability to distinguish identified descendants of inoculum 2 throughout the infection trial demonstrated the utility of AFLP analysis to track bacteria with distinct genetic characteristics.
The capacity of AFLP analysis to distinguish P. multocida serotype 1 isolates during epidemiologic investigations of avian cholera outbreaks was demonstrated by using this technique to characterize 120 isolates cultured from wild birds (Table 1) and environmental samples (Table 2). The most distinct epidemiological cluster of P. multocida isolates examined originated from samples collected in central California (Figs. 4 and 5), and the majority of isolates from this region, regardless of whether they were collected from birds or from environmental samples, allocated to branches H and I of the dendrogram (Figs. 3a and 3c). The inability to distinguish P. multocida isolates originating from birds or the environment by AFLP analysis is consistent with the transmission of bacterial isolates between birds and their environment within an outbreak area (Samuel et al., 2004). Within central California, branch H isolates exhibited a wider distribution pattern than branch I isolates (Fig. 5). The wider spatial and temporal distribution of branch H isolates in central California was also consistent with a more diverse national distribution pattern for branch H isolates. Samuel et al. (2005) demonstrated that healthy wild waterfowl have the potential to transmit P. multocida to other birds and locations. Accordingly, the wider spatial and temporal distribution of branch H isolates, both locally and nationally, may have resulted from the movements of carrier birds. Identification of a group of bacterial isolates primarily associated with a distinct geographic region and a single season indicates that AFLP analysis provides a potentially useful tool to further elucidate the transmission of P. multocida between bird species, across the landscape, and over time.
Both dendrogram (Figs. 3a and 3b) and principal components (Fig. 4) analyses revealed that Great Salt Lake and Salton Sea P. multocida isolates clustered together. The majority of Great Salt Lake isolates analyzed for this study were cultured from Eared Grebes, while Salton Sea isolates were primarily cultured from Eared Grebes and Ruddy Ducks (Table 1). Large numbers of Eared Grebes move between the Great Salt Lake and the Salton Sea each year (Jehl, 1993; Jehl et al., 1999). Thus, interactions between Great Salt Lake and Salton Sea Eared Grebe populations may facilitate the transmission of disease agents between these two sites resulting in genetic homogenization, as measured by AFLP analysis, of the P. multocida isolates cultured from waterfowl of this geographic region. Application of AFLP analysis to P. multocida isolates from additional sites along waterfowl migratory routes may provide information on the transmission of avian cholera useful to wildlife resource managers for predicting where future disease outbreaks might occur.
The previously perceived homogeneity of P. multocida serotype 1 has been a limiting factor in understanding the epidemiology of avian cholera infections in wild bird populations (Samuel et al., 2007). Using AFLP analysis, we have demonstrated that the P. multocida genome is subject to genetic drift and that there is sufficient diversity among P. multocida serotype 1 isolates to distinguish regional and temporal epidemiologic patterns. The utility of this technique for tracing avian cholera outbreaks among wild birds was most clearly exemplified by the group of genetically distinct P. multocida isolates originating from central California bird and environmental samples. Further application of AFLP technology to the study of avian cholera epidemiology has the potential ability to link bacterial isolates from the environment to those from infected hosts, to determine spatiotemporal patterns of isolate types, to identify specific affiliations between P. multocida genotypes and bird species, to correlate virulence with bacterial genotypes, and to understand the role of specific bird species in transmitting or spreading this disease. Accomplishment of these goals will provide wildlife resource managers with the means to predict local bird population impacts during ongoing avian cholera die-offs, to identify bird species that may be at greatest risk for infection during a mortality event, and to correlate bird movement patterns with the potential for disease spread. Amplified fragment length polymorphism analysis promises to be a valuable tool to both enhance understanding of the epidemiology of avian cholera infection and more effectively manage this disease among wild birds.
|
ACKNOWLEDGMENTS |
|---|
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
BLANCHONG, J. A., M. D. SAMUEL, AND G. MACK. 2006. Multi-species patterns of avian cholera mortality in Nebraska’s Rainwater Basin. Journal of Wildlife Diseases 42: 81–91.
BOTZLER, R. G. 1991. Epizootiology of avian cholera in wildfowl. Journal of Wildlife Diseases 27: 367–395.[Abstract]
EARL, A. M., R. LOSICK, AND R. KOLTER. 2007. Bacillus subtilis genome diversity. Journal of Bacteriology 189: 1163–1170.
FRIEND, M. 1999. Avian cholera. In Field manual of wildlife diseases: General field procedures and diseases of birds, M. Friend and J. C. Franson (eds.). Information Technology Report 1999–2001, Biological Resource Division, US Geological Survey, Madison, Wisconsin, pp. 75–92.
GEERT, H., R. COOPMAN, P. JANSSEEN, AND K. KERSTERS. 1996. High-resolution genotypic analysis of the genus aeromonas by AFLP fingerprinting. International Journal of Systematic Bacteriology 46: 572–580.
JEHL, J. R, JR. 1993. Observations on the fall migration of eared grebes, based on evidence from a mass downing in Utah. The Condor 95: 470–473,
Jehl JR, A. R. HENRY, AND S. I. BOND. 1999. Flying the gantlet: Population characteristics, sampling bias, and migration routes of eared grebes downed in the Utah desert. The Auk 116: 178–183.
JOHNSON, R. A., AND D. W. WICHERN. 1982. Applied multivariate statistical analysis. Prentice-Hall, Englewood Cliffs, New Jersey, 543 pp.
LEHR, M. A., R. G. BOTZLER, M. D. SAMUEL, AND D. J. SHADDUCK. 2005. Associations between water quality, Pasteurella multocida, and avian cholera at Sacramento National Wildlife Refuge. Journal of Wildlife Diseases 41: 291–297.
LEVIN, B. R., AND C. T. BERGSTROM. 2000. Bacteria are different: Observations, interpretations, speculations, and opinions about the mechanisms of adaptive evolution in prokaryotes. Proceedings of the National Academy of Sciences of the United States 97: 6981–6985.
OWEN, R. J. 1989. Chromosomal DNA fingerprinting: A new method of species and strain identification applicable to microbial pathogens. Journal of Medical Microbiology 30: 89–99.
RHOADES, R. R., AND R. B. RIMLER. 1991. Pasteurellosis. In Diseases of poultry, 9th ed., B. W. Calnek, H. J. Barnes, C. W. Beard, W. R. Reid and H. W. Yoder (eds.). Iowa State University Press, Ames, Iowa, pp. 145–171.
ROSEN, M. 1969. Species susceptibility to avian cholera. Journal of Wildlife Diseases 5: 195–200.
SAMUEL, M. D., R. G. BOTZLER, AND G. A. WOBESER. 2007. Avian cholera. In Infectious diseases of wild birds, N. J. Thomas, B. J. Hunter and C. T. Atkinson (eds.). Blackwell Publishing, Ames, IA, pp. 239–269.
———, D. J. SHADDUCK, D. R. GOLDBERG, AND W. P. JOHNSON. 2003a. Comparison of methods to detect Pasteurella multocida in carrier waterfowl. Journal of Wildlife Diseases 39: 125–135.[Abstract]
———, ———, ———, M. A. WILSON, D. O. JOLY, AND M. A. LEHR. 2003b. Characterization of Pasteurella multocida isolates from wetland ecosystems during 1996 to 1999. Journal of Wildlife Diseases 39: 798–807.[Abstract]
———, ———, AND ———. 2004. Are wetlands the reservoir for avian cholera? Journal of Wildlife Diseases 40: 377–382.
———, ———, ———, AND W. P. JOHNSON. 2005. Avian cholera in waterfowl: The role of lesser snow geese and Ross’s geese as disease carriers in the Playa Lakes region. Journal of Wildlife Diseases 41: 48–57.
SOKAL, R. R., AND C. D. MICHENER. 1958. A statistical method for evaluating systematic relationships. University of Kansas Science Bulletin 38: 1409–1438.
VOS, P., R. HOGERS, M. BLEEKER, M. REIJANS, T. VAN DER LEP, M. HORNES, A. FRIJTERS, J. POT, J. PELEMAN, M. KULPER, AND M. ZABEAU. 1995. AFLP: A new technique for DNA fingerprinting. Nucleic Acids Research 23: 4407–4424.
WARD, J. H. 1963. Hierarchical groupings to optimize an objective function. Journal of the American Statistical Association 58: 236–244.
WILSON, M. A., R. B. DUNCAN, G. E. NORDHOLM, AND B. M. BERLOWSKI. 1995a. Serotypes and DNA fingerprint profiles of Pasteurella multocida isolated from raptors. Avian Diseases 39: 94–99.[Medline]
———, ———, ———, AND ———. 1995b. Pasteurella multocida isolated from wild birds in North America: A serotype and DNA fingerprint study of isolates from1978 to 1993. Avian Diseases 39: 587–593.[Medline]
WOBESER, G. 1992. Avian cholera and waterfowl biology. Journal of Wildlife Diseases 28: 674–682.[Medline]
Received for publication 9 April 2007.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
