Over the 41-month trapping period at Fort Lewis and the 37-month trapping period at Molina, antibody reactive with SNV was detected in 29 (9.6%) of 302 deer mice at Fort Lewis and 36 (9.4%) of 385 at Molina; 4 (2.6%) of 155 of pinyon mice at Molina also had antibody (Table 1). For comparison, in 1993, prevalence of antibody to SNV in P. maniculatus was approximately 50% near Fort Lewis (La Plata County) and 19% near Grand Junction (Mesa County) (1). Of 112 least chipmunks (Tamias minimus), two Colorado chipmunks (T. quadrivittatus), and two western harvest mice (Reithrodontomys megalotis), none had antibody to SNV.

At Fort Lewis, trapping success (number of animals per total number of trap nights) was 0.3% to 7.6%, depending on the season (lowest rates, April–June; highest, August–October). Antibody-positive deer mice were found in 13 of 21 trapping intervals. Antibody prevalence (calculated when more than four deer mice were caught in a given trapping period) was 0% to 42.9% with a mean of 29 (9.5%) of 302. Antibody to SNV was detected in adult (10.5%), subadult (9.8%), and juvenile (12.5%) deer mice; the stages represented 63.9%, 13.1%, and 23%, respectively, of the deer mice captured. Males represented 48.8% of the deer mice (and 47.8% of recaptured deer mice) but 58.3% of the antibody-positive rodents.

At Molina, trapping success was 2.6% to 17.9% and, as at Fort Lewis, depended on the season (lowest rates, May–June; highest, July–October). Antibody-positive deer mice were found in 12 of 17 trapping intervals. Antibody prevalence was 0% to 33% in deer mice (mean 9.4%) and 0% to 18.2% in pinyon mice (mean 2.6%). Antibody to SNV was detected in adult (11.3%), subadult (1.7%), and juvenile (4.4%) deer mice; the stages represented 73.1%, 15.3%, and 11.7%, respectively, of the mice captured. Males represented 45.5% of the deer mouse population, 46.3% of the recaptured deer mice, and 60% of the antibody-positive mice. Antibody was detected in four adult (three male, one female) pinyon mice (P. truei). Of 118 pinyon mice collected, 62 (50.8%) were female and 56 (49.4%) were male. We detected seropositive pinyon mice only during May and June 1995 and April 1996.

Seroconversion

Figure 1

Figure 1. Approximate month of seroconversion in Peromyscus species at Fort Lewis and Molina, Colorado, June 1994–October 1997, by sex.

Fifteen deer mice and one pinyon mouse seroconverted (i.e., seronegative to seropositive or a fourfold or greater increase in titer) between captures (Figure 1). At Fort Lewis, 302 deer mice (150 female and 152 male) were captured. Of these, 37 female and 37 male mice were recaptured at least once. Five male and three female deer mice at Fort Lewis seroconverted. One deer mouse had antibody for the first time 14 months after it was initially captured. At Molina, 385 deer mice (212 female, 173 male) and 155 pinyon mice (85 female, 70 male) were captured. Of these, 33 female and 30 male deer mice and 12 female and 10 male pinyon mice were recaptured at least once. Five male and two female deer mice and one male pinyon mouse seroconverted. An additional three deer mice (two male, one female) at Molina were recaptured and had significant (3,200 to 25,600) but stable IgG antibody titers; we did not consider these as having seroconverted. The five male mice seroconverted at Fort Lewis during the summer (one between July and September 1994, two between July and September 1995, one at [estimated] midsummer 1995, and one between June and September 1997); two female mice seroconverted between October 1994 and May 1995, and one female mouse seroconverted during late summer (September to October) 1997. At Molina, one male deer mouse seroconverted in late spring (estimated May) 1995, one in late fall 1995, two male deer mice and a male pinyon mouse during the winter or early spring of 1995 to 1996, and one male deer mouse during late spring 1996; one female deer mouse seroconverted in late summer 1995 and one during the winter 1995 to 1996. Seropositive samples were titrated by IgG ELISA with fourfold dilutions. Titers were 100 to 102,400, with most of them at 6,400 to 25,600.

Population Densities

Figure 2

Figure 2. Minimum number of deer mice alive (MNA) (the number of individual mice captured in a month plus those mice captured on at least one previous and one subsequent occasion) and estimated...

Figure 3

Figure 3. Minimum number of deer mice alive (MNA) (the number of individual mice captured in a month plus those mice captured on at least one previous and one subsequent occasion) and estimated...

Deer mouse populations and prevalence of antibody to SNV at Fort Lewis (Figure 2) were relatively low throughout this study, except in May and June 1995 when samples included only four deer mice and one deer mouse, respectively. Mean minimum number of rodents alive was 28 in 1994 but lower from 1995 to 1997 (10.8, 13.4, and 16.4, respectively). At Molina, populations were relatively stable between 1995 and 1997 (only one collection made in 1994), with mean minimum number of rodents alive values of 31.2, 20.4, and 25.4, respectively. As at Fort Lewis, estimated standing prevalence values were commensurately low (Figure 3).

On the basis of the high antibody titers of these seropositive samples, our findings elsewhere in Colorado (Calisher, Beaty, and Mills, unpub. data), and the findings of others studying hantaviruses in the Southwest (9), we presumed that IgG antibody to SNV in deer mice indicated infection with SNV and not with El Moro Canyon or another hantavirus. Although we did not attempt to isolate or detect hantaviral RNA in blood or other tissues from mice with antibody, the only hantavirus specifically identified in deer mice in western Colorado has been SNV (10).

The presence of IgG antibody to hantaviruses in rodents is presumed to indicate past infection and present infection, at least in the primary vertebrate hosts of hantaviruses (Mills et al., this issue, pp. 135-142). That is, rodents infected with hantaviruses with which they appear to be closely associated coevolutionarily (e.g., deer mice and SNV, Western harvest mice and El Moro Canyon virus, rice rats [Oryzomys palustris] and Bayou virus, Black Creek Canal virus and cotton rats [Sigmodon hispidus]) do not appear ill or otherwise affected by hantaviruses specific to them. In host-virus associations that have been studied, the specific hosts become infected early or later in life, are viremic for a short period, and excrete virus in their saliva, urine, and feces, perhaps for life (11-14).

Fighting (including exchange of blood and saliva) between infected and uninfected adult rodents has been suggested as the primary mechanism by which hantaviruses are amplified epizootically (8). Infected rodents become viremic and viruric and serve as subsequent sources of infection for others in the population. Earlier studies using Seoul virus and laboratory rats as a model system had indicated that while in newborn rats infection became persistent, in older rats it was transient (15). However, evidence using Black Creek Canal virus and adult hispid cotton rats, Hantaan virus and Apodemus agrarius, and Puumala virus and Clethrionomys glareolus indicates that whereas viremia may diminish over time, virus can still be detected in various organs, including the salivary gland, for several months after infection (11-14). Given the relatively brief life span of rodents, infection and concomitant infectivity for a few weeks or months would provide a mechanism for seasonal, albeit not transseasonal, persistence of hantaviruses. Passive acquisition of maternal antibody may protect the offspring of infected dams early in their lives, but when antibody wanes, they enter the adult population as susceptibles. Infected later in life, they can become persistent shedders of virus and sources of infection for others in the population.

Deer mice infected with SNV when very young likely are able to serve as reservoirs of the virus for the remainder of their lives. Although our studies do not distinguish between death and dispersal, the life span of many deer mice at these sites may not be much more than a month. However, because some deer mice live for 1 or 2 years, longevity of even a small proportion of the deer mouse population may provide a transseasonal mechanism for virus persistence.

A second mechanism of virus transmission, an epizootic one, depends on short-term infections of deer mice infected as subadults or as adults. At periods of deer mouse population peaks (e.g., at the end of the breeding season, in late summer and fall, and during period of decreased availability of food), male mice fight one another for breeding partners, food, and territory. This premise is supported by results of serologic tests of recaptured deer mice at Fort Lewis and at Molina. At Fort Lewis, 48.8% of the deer mice and 47.8% of the recaptured deer mice were male, but 58.3% of the seropositive deer mice were male. At Molina, 45% of the deer mice and 46.3% of the recaptured deer mice were male, but 60% of the seropositive deer mice were male. These data support the hypothesis that male deer mice contribute more to the epizootic cycle of SNV than female deer mice. However, the lack of association between sex, wounds, and antibody at either Fort Lewis or Molina indicates that individual mice of either sex may fight and, through this mechanism or another, become infected with a hantavirus. That most mice with antibody to SNV are male supports the suggestion that fighting among mice, biting, and scratching can lead to hantavirus transmission from an infected to an uninfected, wounded mouse (4). The limited time these mice may be able to transmit virus might be sufficient to maintain virus infection in the population.

When deer mouse populations decrease precipitously because of decreased availability of food and water, the likelihood that SNV will disappear from the population increases. However, a few long-lived, persistently infected deer mice can serve as reservoirs until conditions are suitable for the populations to recover.

Our data appear to support such a unified hypothesis. Fluctuations in IgG antibody prevalence in deer mice at Fort Lewis and at Molina have lagged somewhat behind but have been similar to fluctuations in deer mouse population. In male deer mice at Fort Lewis and at Molina, most seroconversions (recent infections) occurred during the summer or fall, whereas in female deer mice, most occurred between fall and spring. During winter, Colorado deer mice reduce their home range, aggregate in nests, and enter short-term torpor—strategies that together temper reduced food availability and energy loss due to cold (2). Although we did not find deer mice that had been infected for more than 3 months, we recovered a few more than 1 year (some nearly 2 years) after they first were trapped; thus, under natural conditions and despite the usual declines caused by predation, cold, heat, and decreases in food, deer mice that reach adulthood can live as long as 2 years (2), a period sufficient to allow SNV to survive adverse conditions of low populations and the resulting decreased number of susceptibles. Furthermore, whereas the overall seroprevalence of IgG antibody to SNV in deer mice at Fort Lewis was 6.8% (12 of 165) and in deer mice and pinyon mice, respectively, at Molina 7.2% (15 of 193) and 5.5% (3 of 52), the rate of seroconversion among deer mice at Fort Lewis was 16.3% (8 of 41) recaptures, and among deer mice and pinyon mice, respectively, at Molina 9.9% (7 of 64) and 3.3% (1 of 29). These results suggest that the longer deer mice live, the greater the cumulative probability they will become infected with SNV.

The deer mouse, the most numerous mammal in North America, often described as a "quintessential generalist," can survive on any dry land habitat in its range and invade and exploit areas disturbed by flood, fire, avalanche, landslides, mining, construction, extreme grazing, or land development. In ecologically stable areas, deer mice may be limited by the presence of more specialized rodent species (2), but they are found from forests to grasslands, canyons to deserts, farmlands to farm houses and suburban homes, moving into the latter more often in fall but able to take up residence whenever an opportunity presents itself. Thus, the movement of SNV-infected deer mice into human residences itself creates a risk factor for HPS. The Fort Lewis and Molina sites have not been affected by obvious ecosystem perturbations in recent years, and deer mouse populations at these sites are not high, yet are considerably decreased from the apparently inordinately high levels of 1993 (J. Mills, pers. comm. 1997). Each site seems ecologically stable, but subtle changes may have gone unnoticed.

Deer mice are omnivores, storing food for winter consumption but known to feed on acorns, nuts, insects, other small invertebrates, carrion, fungi, bone, and various plant parts, including seeds, leaves, and bark, roots, and tubers (2,16). In one study, seeds accounted for 69% to 76% of stomach contents of deer mice in Colorado, insects for 14% to 25% (2). This proportion depends on the season (i.e., availability of food supply); deer mice are more likely to feed on insects and insect larvae in spring, seeds and berries in fall.

Notwithstanding the nature of deer mice to consume a variety of foods, they rely heavily on acorns, when oaks (Quercus sp.) comprise a significant proportion of their habitat (16). Further, in the northeastern United States, the quantity of mast seems directly related to population size of white-footed mice (P. leucopus) and eastern chipmunks (T. striatus) (17). Although Gambel's oak is abundant at the Fort Lewis trapping sites and copious numbers of acorns were attached to the trees and on the ground in 1994, we did not observe acorns on trees there between spring 1995 and fall 1997. The chipmunk population at this site was, with the exception of a transient, moderate increase in June 1996, never high during our study period (June 1994 to October 1997). At the Molina sites, which do not have oaks, the chipmunk population declined considerably after June 1996 and did not return to its previous level. Chipmunks may serve as an indicator for the ready availability of acorns and other nuts or food in general.

The continued low population densities of deer mice at Fort Lewis and Molina are puzzling. If, for example, deer mouse population densities in surrounding areas are higher than at the study sites, one might expect deer mice from those areas to move into the area with the low population. However, at Fort Lewis 59% of adult and 5% of subadult deer mice were recaptured; at Molina 86% of adults and no subadult or juvenile deer mice were recaptured as adults. These data indicate that few, if any, young deer mice are immigrating to these sites or if they are, they did not survive long enough to be captured, and the survival rate of this species' young is not high.

Temperature fluctuations that affect habitat characteristics can influence rodent breeding seasons (e.g., rate of plant growth, total available nutritional biomass). However, analyses of available data (not presented) did not provide obvious evidence for such direct relationships. In contrast, a paucity of precipitation at Fort Lewis, between March 1995 and October 1996, and at Molina between May 1995 and April 1997, coincided with the usual breeding season of deer mice, least chipmunks, and other rodents at these sites, and with lower rodent population densities between the end of 1995 and the end of 1997; antibody prevalence fluctuated in parallel. Whether the two consecutive relatively wet years 1996 and 1997 will bring about conditions suitable to increase rodent populations near Fort Lewis and Molina and lead to an increase in HPS in the near term has not been determined. Availability of water may be the sine qua non of plant food availability, reproductive preparedness, gravidity rates, and attendant intraspecific fighting among individual mice within an increased population. From data (not shown) collected at these sites since October 1997, indications are that both rodent population densities and antibody prevalence are increasing at both sites. If precipitation excess correlates with rodent population, density increases, and the prevalence of hantaviruses, we will be able to predict increases in risk for hantavirus infection in the human population.