The potential exposure sites were chosen by selecting all places where patients had been living or working or had visited during the 6 weeks before onset of symptoms. Rodents were trapped in all these sites, which were classified into six categories: domestic and peridomestic urban, domestic and peridomestic rural, other urban, and other rural. Peridomestic urban and rural categories were all sites in the immediate vicinity of homes or buildings, including yards, parks, driveways, adjoining lands, outbuildings, vegetable gardens, and fence lines. The peridomestic rural category includes ponds, natural or planted woodlots, weeds, sugar cane or plantain plantations, and corn stubble in the immediate vicinity of the house. All other trapping sites distant from the previously mentioned settings were considered other urban or other rural. Other urban includes sites from the outskirts of towns and natural and artificial corridors that could allow the access of sigmodontine rodents to urban areas, such as railroad rights-of-way and roadsides inside the perimeter of the town. In other rural sites rodents were captured in open fields, where the representative habitats of each area were sampled, including natural and modified land, such as cultivated areas and weeds.

In the southern and central zones, rodents were trapped as soon as HPS case reports were received. In the remote northern zone, three expeditions were organized to trap rodents at sites frequented by six persons with HPS reported in previous months, and only rarely was trapping conducted inside houses. The three expeditions took place in July 1995, October 1996, and May 1998; rodents were trapped at 18 sampling sites.

From August 1994 to April 1998, 46 sampling sites were selected in the central zone. In the southern zone, we included 51 sampling sites from November 1996 to April 1998 (Table 1).

Each site was sampled with Sherman (8 x 9 x 23 cm) and Tomahawk (14 x 14 x 40 cm) live-capture traps. The number of traps depended on the area available for trap placement at each site. Animals were trapped and sampled according to established safety guidelines (9) and were anesthetized with Isoflurane (Abbott Laboratories) before blood was drawn from the retroorbital sinus. Carcasses were tentatively identified in the field and kept in a solution of 10% formalin for confirmation of identification at the Museum of Natural Sciences "Bernardino Rivadavia," Buenos Aires.

During 26,458 Sherman and 1,322 Tomahawk trap-nights, 2,299 small mammals belonging to two orders (Rodentia and Didelphimorphia) and three families (Muridae, Caviidae, and Didelphidae) were captured. These animals belonged to 32 species, with the murid subfamily Sigmodontinae representing 86.3% of the total sample.

The introduced murine rodents R. rattus and M. musculus, as well as Cavia aperea (Caviidae), were captured in all three areas. Sigmodontine rodents were represented by different species in the three regions.

In all three regions, M. musculus was found in domestic urban sites (Table 1). In two of the three areas, we also observed rodents inside urban homes; this is the first documented occurrence of sigmodontine species entering homes in Argentina.

We also found sigmodontine rodents inside rural homes: one Calomys laucha and one O. flavescens in the central zone and one Akodon spegazzinii in the northern zone. Sigmodontine rodents, including the reservoirs for Lechiguanas and Andes viruses, were also captured in the peridomestic urban sites, especially in the central and southern zones. In peridomestic rural habitats next to open fields, captures of sigmodontines were expected. The trap success values for hantavirus reservoir species in peridomestic rural sites were similar or higher than those in open fields represented by other rural sites. The relative proportion of rodent species among site categories includes all species antibody positive and the species that were numerically dominant but antibody negative in each zone. The category "others" includes species that were less representative in each zone; the high values observed in PU and OU sites in the northern zone were due to the low number of captures and in OR to the high diversity of species captured (Table 2). The relative proportion was compared by chi-square test with Epi Info version 6.04. Only site categories with >30 captures could be tested. An increase in the relative proportion of O. flavescens (host of the genotype Lechiguanas, associated with human disease) in the central zone and O. longicaudatus (putative reservoir of the genotype Orán, also associated with human disease) in the northern zone was seen in peridomestic rural settings in comparison with other rural. O. longicaudatus (proposed reservoir for Andes virus) was captured in similar relative proportions in both peridomestic and other rural sites. In all cases, these findings emphasize the risk linked to peridomestic settings.

We tested 2,159 (93.9%) rodents in IgG ELISA by using Sin Nombre virus antigen (CDC, SPR293). We used a recombinant nucleocapsid protein as antigen applied to the solid phase of a microtiter plate. Hantavirus-specific IgG in test samples of rodent whole blood was allowed to bind to the antigen. A mixture of two conjugates (anti-Peromyscus leucopus and anti-Rattus norvegicus, Kirkegaard and Perry) was used to detect immune globulins from various murid rodent phyla. This was followed by 2,2'-azino-di(3-ethybenthiazonline sulfonate) substrate (Kirkegaard and Perry Laboratories, Inc.) and read with a Bio-Tek Microplate autoreader at 405 and 450 nm. A titer >1:400 was considered positive (10).

Of 330 rodents tested in the north, 5 (1.5%) were positive (Table 3). In the central zone, we found 35 (2.6%) positives among 1,326 rodents, associated with eight HPS cases. In the south, 27 (5.4%) of 503 rodents tested had positive results. In the northern zone, the presence of infected O. longicaudatus was associated with HPS cases in peridomestic rural habitats. The importance of detecting infected O. chacoensis and Akodon varius associated with an HPS case in peridomestic urban and rural sites cannot be assessed until data on the viral genotypes of the rodents and the case patients are available.

In the central zone, apart from O. flavescens, already shown to be associated with HPS, another species found infected was A. azarae, the putative reservoir of the Pergamino genotype, which has not yet been associated with human disease. Spatial and temporal association between an HPS case and an infected A. azarae does not confirm this species as the source of infection. Further genetic studies are under way to determine if Pergamino virus was responsible for the HPS cases.

In the southern zone, human cases were associated with O. longicaudatus captured in peridomestic and other rural settings (Table 3). In the three zones, in all other site categories, no seropositive animals were found. Nevertheless, because of small sample sizes, any conclusions concerning lack of infection in these site categories are tentative.

Infected hantavirus reservoir hosts (as evidenced by antibody positivity) were found within peridomestic environments in all three HPS-endemic zones in Argentina. Reservoir species were captured inside urban houses in two of the three endemic zones. Although host species were not captured in homes in the northern zone, sampling was not sufficient to exclude the possibility that they enter homes occasionally.

The presence of hantavirus reservoir species in peridomestic environments indicates risk for human inhabitants. The primary measure for reducing the risk is preventing access of rodents to homes (11). The efficacy of proposed and currently used exclusion methods in Argentina needs to be evaluated (12).

Sigmodontine rodents, including known hantavirus reservoir species, were frequently captured in the rural and small-town peridomestic environments we studied. At many of the case sites, the level of hygiene was suboptimal. The widespread presence of such conditions underscores the importance of local habitat management to prevent wild (sigmodontine) rodents from entering domestic areas in towns, villages, and urban centers and of health education for the local population to reduce the risk for hantavirus infection.