Ecoregional Dominance in Spatial Distribution of Avian Inﬂ uenza (H5N1) Outbreaks

Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 13, No. 8, August 2007 1269 When all identifi ed strains of M. cosmeticum are considered, this species is clearly present in diverse geographic regions and in healthcare institutions. These fi ndings suggest that it may be widely distributed in the environment and should be regarded, along with other rapidly growing mycobacteria species, as a potential pathogen.


Ecoregional Dominance in Spatial Distribution of Avian Infl uenza (H5N1) Outbreaks
To the Editor: Recent articles in Emerging Infectious Diseases (1,2) and elsewhere (3,4) have highlighted the role of Anatidae migration in dispersal of the H5N1 subtype of highly pathogenic avian infl uenza (HPAI) virus. Although these articles point out that identifying the geographic origin of migrating waterfowl is needed to understand and predict pathogen dispersal, study analyses have been limited to pathways with nominal reference to climatic and vegetation patterns that control spatiotemporal patterns of this migration.
We propose that a better understanding of the threat of future spread can be obtained by identifying specifi c climatic and vegetation zones that are important in the life cycle of Anatidae, and which account for a disproportionately large number of HPAI outbreaks. The concept of ecoregions (5,6), i.e., distinct assemblages of natural communities determined by climate, geology, and evolution, is a useful zonal classifi cation for evaluating HPAI outbreaks. A World Wildlife Fund classifi cation delineating 825 terrestrial ecoregions (7), combined with a Google Earth map of 3,133 avian infl uenza outbreaks from November 24, 2003, to November 21, 2006 (8), provided the basis for this analysis.
All fi les were converted to shapefi les (Environmental Systems Research Institute, Redlands, CA, USA), and overlay analysis was performed by using ArcGIS software (Environmental Systems Research Institute).
The online Appendix Figure  (available from www.cdc.gov/EID/ content/13/8/1269-appG.htm) shows a chloropleth map (display of quantitative or qualitative information about subentities in terms of symbols or colors) of ecoregions with numbers of avian infl uenza cases (each spatially and temporally isolated set of individual events, regardless of number of deaths, is recorded as a case). Panels A, B, and C of this fi gure show enlargements of specifi c ecoregions with large numbers of known cases in regions of Eurasia, Southeast Asia, and Africa, respectively. Twenty-fi ve ecoregions, representing 8.8% of the terrestrial surface area, accounted for 2,407 (76.8%) cases. A total of 132 of 825 ecoregional classifi cations had >1 recorded case of an avian infl uenza outbreak, but most (83) had <10 cases each.
Regionally, Southeast Asia has 12 ecoregions that collectively account for 1,651 cases (online Appendix Figure, panel B) that have occurred consistently, albeit cyclically, since 2003. Among these ecoregions, the freshwater wetlands of the Chao Phraya, Tonle Sap, and Red Rivers are known migratory waterfowl wintering habitats in which 719 cases were located. Recent phylogenetic evidence suggests that this area is a local hotspot for an endemic strain of avian infl uenza (H5N1) that demonstrates bidirectional dispersal among localities within the region (9).
In the Eurasian region (online Appendix Figure, panel A), 12 ecoregions accounted for 712 cases. The easternmost ecoregions, the Kazakh forest steppe (location of Lake Chany, an Anatidae habitat and breeding area) and the Kazakh Steppe, accounted for 132 cases, with the fi rst case recorded  Our results may be skewed by several confounding factors, e.g., low national surveillance capabilities resulting in unreported cases and effects of the poultry trade. Nonetheless, the fi ndings have implications for global monitoring of avian infl uenza (H5N1) outbreaks. Although migratory pathways and the poultry trade should continue to be scrutinized, monitoring efforts should focus on wintering and breeding habitats of migrating waterfowl, especially wetlands located within ecoregions with a disproportionately large number of avian infl uenza outbreaks. These hotspots are also likely to give rise to endemic local strains with regional dispersal characteristics (9). In Response: Ecoregions are large units of land that refl ect the distribution of species and communities and are used to defi ne priority areas for biodiversity conservation. In their letter (1), Sengupta et al. suggest that ecoregions could help explain the spatial distribution of observed outbreaks of HPAI (H5N1). Since most outbreaks they observed occurred in ecoregions with waterfowl habitats, they concluded that monitoring efforts should focus on these areas.

Raja
Habitat mapping is useful for predicting the spread of wildlife-associated diseases. However, because ecoregions are large areas where agriculture and other human activities abound, many confounding variables exist. To assume that ecoregions with HPAI (H5N1) outbreaks are chiefl y characterized by the natural biotopes is too simplistic.
In addition, the authors assume that wild waterfowl are the main factors in the dispersal of HPAI (H5N1). Although wild birds are known to be involved in the dispersal of low pathogenic strains of avian infl uenza viruses (2,3), the precise role of these birds in the dispersal of HPAI (H5N1) viruses remains controversial. It is now clear that the expansion of HPAI (H5N1) in Asia and to Africa was primarily associated with the poultry trade (4), which is not represented on a map of ecoregions. Also noteworthy is that HPAI (H5N1) did not persist in wild birds after the spring of 2006, and no case has been detected in the wetlands of America despite the migration of wild ducks and waders from Siberia (their main breeding ground) to Asia, Europe, and North America.
We believe that human activity is the primary vehicle for the spread of HPAI (H5N1) virus. Rather than the monitoring of waterfowl habitats, increased surveillance of the legal and illegal trade of birds and bird products is more likely to help in predicting the spread of this virus. (1) discuss the role of ecoregions in the distribution of HPAI (H5N1) outbreaks. Although the concept of ecoregions is undoubtedly useful in global biogeography, we do not understand the point they are trying to make. In our article (2), which is cited in their letter, we undertook a descriptive study to determine whether spread of HPAI (H5N1) virus was consistent in time with ecologic drivers of bird migration and in space with distribution of major migratory fl yways of Anatidae. It is obvious that the distribu-tion pattern of Anatidae is dependent on ecologic variables, and some of these variables are summarized by the ecoregion concept.
However, apart from a strictly descriptive point of view, we do not see how the ecoregion concept applies to describe patterns in HPAI (H5N1) spread and distribution. Sengupta et al. list ecoregions where reports of HPAI (H5N1) were concentrated. However, what do they infer from this? They observe regions with many reports of HPAI (H5N1) and conclude that these ecoregions are at risk. We fi nd this reasoning completely circular, and any geographic zonation would provide the same observation. They may mean that ecoregions defi ne boundaries within which secondary spread of HPAI is more likely than across ecoregions. However, this hypothesis would need to be more clearly demonstrated and quantifi ed before the ecoregion concept can be used for global monitoring of HPAI (H5N1) outbreaks.