A 19 Year Analysis of Small Mammals Associated with Human Hantavirus Cases in Chile

Small mammals present in areas where hantavirus cardiopulmonary syndrome (HCPS) cases had occurred in central and southern Chile were captured and analyzed to evaluate the abundance of rodents and seroprevalence rates of antibodies to Andes orthohantavirus (ANDV). Sampling areas ranged from the Coquimbo to Aysén regions (30–45° S approx.) regions. Ninety-two sites in peridomestic and countryside areas were evaluated in 19 years of sampling. An antibody against ANDV was detected by strip immunoassay in 58 of 1847 specimens captured using Sherman traps. Of the eleven species of rodents sampled, Abrothrix olivacea, Oligoryzomys longicaudatus and Abrothrix hirta were the most frequently trapped. O. longicaudatus had the highest seropositivity rate, and by logistic regression analysis, O. longicaudatus of at least 60 g had 80% or higher probability to be seropositive. Sex, age and wounds were significantly related to seropositivity only for O. longicaudatus. Across administrative regions, the highest seropositivity was found in the El Maule region (34.8–36.2° S), and the highest number of HCPS cases was registered in the Aysén region. Our results highlight the importance of long term and geographically extended studies, particularly for highly fluctuating pathogens and their reservoirs, to understand the implications of the dynamics and transmission of zoonotic diseases in human populations.


Introduction
Emerging zoonotic diseases are a major concern for public health systems, producing new or previously unrecognized diseases, or rapidly increasing incidence of infection in the human population. Ecological, environmental changes or changes in human demographics and behavior, which increase chances of contact with reservoirs, are the main factors responsible for these emergences [1]. RNA viruses represent an important group of emerging human pathogens, including several harbored by animal reservoirs [2]. Recent examples are the outbreak of pandemic influenza A subtype H1N1, human immunodeficiency virus, Ebola virus, hepatitis C virus, West Nile virus, rabies virus and severe adult respiratory distress (SARS), which have challenged control and prevention measures of health systems.
Orthohantaviruses are segmented negative-strand RNA viruses of the family Hantaviridae, which contain large (L), medium (M) and small (S) segments encoding RNA-dependent RNA polymerase, glycoproteins Gn and Gc and nucleocapsid (N) protein, respectively [3]. Several orthohantavirus strains are recognized as human pathogens [4][5][6]. Orthohantavirus infections became a concern in the Americas in 1993 after detecting an outbreak of a previously unrecognized syndrome [7]. The disease (hantavirus cardiopulmonary syndrome, HCPS), was produced by infection with the newly-discovered Sin Nombre virus (SNV), with Peromyscus maniculatus (deer mouse) identified as the reservoir [8]. Since then, more than forty-three genotypes have been reported in the Americas alone [5], with nearly half of those associated with clinical cases of HCPS. Pathogenic hantaviruses are carried by rodents of the families Muridae and Cricetidae, while hantaviruses of unknown pathogenicity have also been discovered in shrews, moles and bats [9][10][11][12][13]. Hantavirus infection is transmitted to humans via inhalation of contaminated rodent excreta and secretions or less commonly through rodent bites [5]; the case fatality rate in the Americas is around 36% [6,[14][15][16], although it may reach up to 56% [17].
In southern South America, Andes orthohantavirus (ANDV) is a major etiologic agent of HCPS, and the sigmodontine rodent Oligoryzomys longicaudatus (the pigmy rice rat) is the reservoir [18,19]. O. longicaudatus is distributed in Chile and Argentina, with higher abundances in mesic areas in the temperate forests and Patagonia [20]. Spillover has been reported to other sympatric rodent species, although with no apparent epidemiological importance [21]. Strikingly, person-to-person transmission of hantavirus infection has only been demonstrated for ANDV [16,[22][23][24][25].
In Chile, 1157 cases have been confirmed between 1995 and December 2018 in 11 administrative regions from the Coquimbo region in the north to the Aysén region in the south (30-45 • S approx.; epi.minsal.cl). Although human infections have been reported in a variety of environments, higher numbers of HCPS cases tend to be associated with people living, working or visiting rural areas where the reservoir occurs [20,26]. The settlement of human populations in areas near habitats where wild infected rodents occur increases the likelihood of contact and the probability of contracting the infection. Although the relation between the number of infected hosts and human incidence depends on complex ecological processes [27], there is evidence suggesting that higher seroprevalence in the hantavirus reservoir may lead to more human infections [28]. To understand eco-epidemiological processes that might be responsible for human hantavirus infections, we undertook an analysis of the abundance and seroprevalence of small mammals in Chile in peridomestic and countryside locations where cases of HCPS had occurred. To date, O. longicaudatus is the sole recognized reservoir for ANDV in Chile, and only a few studies have explored the participation of other small mammal species in the transmission of the disease [19,29]. Our study highlights the value of combining information from reservoir ecology, epidemiology and geography to gain insights into the persistence and transmission of zoonotic diseases.
Patricio Sánchez Reyes, Pontificia Universidad Católica de Chile (Santiago, Chile) and in the Division of Mammals of the Museum of Southwestern Biology, University of New Mexico (Albuquerque, NM). For the determination of antibodies against ANDV, the SIA (strip immunoassay vacuum-blot test) technique was used following the method described previously (2004) [20,35].
To determine the relative intraspecific seropositivity for hantavirus by locality, the presence of antibodies to the virus was standardized using the ratio between the number of seropositive at each sampled locality and the trapping success (number of captures by number of total traps-nights). All values were determined for O. longicaudatus, A. hirta and A. olivacea that resulted positive for antibodies against the ANDV [36]. In addition, the total relative seropositivity by sampling site (total seropositives of the locality/trapping effort) and by region (total seropositives of the region/trapping effort) was estimated.

Confirmed Human Cases of Hantavirus
The confirmed human cases of hantavirus between 2000 and 2018 were obtained using two sources of information: (i) Through a national research network established by the Hantavirus Program in which HCPS cases were enrolled in clinical research studies shortly after hospital admission, and (ii) through the National Notification report of the Ministry of Health. HCPS is a disease that must be reported as soon as the diagnosis is suspected. Human cases were classified by locality and the administrative region.

Ecological Features of the Reservoirs
To determine the relationship between seropositives and age, small mammals were sexed and classified by age into juveniles or adults according to their body mass and reproductive status (i.e., males with scrotal or abdominal testes and females with or without perforated vagina, pregnant and/or lactogenic) [37,38]. We also registered presence or absence of wounds or scars, such as bites on the ears or tail, to determine the relationship between seropositives and wounds.

Data Analysis and Statistics
The χ2 test and Fisher's exact test were used to compare the following categorical variables between groups using 2 × 2 tables: Serostatus (seropositive or seronegative) according to sex (male or female), age (adult or juvenile) and wounds (injured or not injured). The significance criterion was set at p < 0.05. Calculations were performed using the VassarStats website for statistical computation (http://www.vassarstats.net). The effect of the weight (independent variable) on the serostatus (response variable) was analyzed by using a logistic regression performed within the free data analysis environment R [39].

Serology and Human Hantavirus Cases
We determined the rodents' relative seropositivity in sites associated with human hantavirus cases, and human hantavirus cases across the 11 sampled administrative regions in Chile. The numbers of human hantavirus cases were standardized to cases per 1000 rural inhabitants (information extracted from the National Institute of Statistic of Chile, www.ine.cl).

Intraspecific Seropositivity
A total of 58 specimens were found positive for antibodies against the ANDV (seropositives) from 30 localities (Figure 1). Of the 631 O. longicaudatus captured, 41 were seropositive from 22 localities (Figure 2A), with an intraspecific seroprevalence of 6.5%. For A. hirta, 234 individuals were captured, 12 were seropositive from nine localities ( Figure 2B) with an intraspecific seroprevalence of 5.13%. For A. olivacea, 636 individuals were captured, with five seropositives from four localities ( Figure 2C) and an intraspecific seroprevalence of 0.79%. All seropositive O. longicaudatus were adults, including 32 males and nine females. For A. hirta, of the 12 seropositives, seven were adult males, four adult females and one juvenile male. For A. olivacea, five were seropositive, including three adult males, one juvenile male and one juvenile female.

Intraspecific Seropositivity
A total of 58 specimens were found positive for antibodies against the ANDV (seropositives) from 30 localities (Figure 1). Of the 631 O. longicaudatus captured, 41 were seropositive from 22 localities (Figure 2A), with an intraspecific seroprevalence of 6.5%. For A. hirta, 234 individuals were captured, 12 were seropositive from nine localities ( Figure 2B) with an intraspecific seroprevalence of 5.13%. For A. olivacea, 636 individuals were captured, with five seropositives from four localities ( Figure 2C) and an intraspecific seroprevalence of 0.79%. All seropositive O. longicaudatus were adults, including 32 males and nine females. For A. hirta, of the 12 seropositives, seven were adult males, four adult females and one juvenile male. For A. olivacea, five were seropositive, including three adult males, one juvenile male and one juvenile female.
Among sites, the highest relative seropositivity for O. longicaudatus was reported in Las Peñas 2002, San Fernando, O'Higgins region and Los Queñes, Romeral, Maule region (3.6), and the lowest in Paso El León, Cochamó, Los Lagos region and Colico, Cunco, Araucanía region (0.08). A. hirta showed the highest relative seropositivity in Caleta Rollizo, Cochamó, Los Lagos region (4.02) and the lowest in Playa San Julián, Corral, Los Ríos region (0.14). A. olivacea showed the highest relative seropositivity in Los Queñes, Romeral, Maule region (3.6) and Ucúquer, Litueche, O'Higgins region (0.32). Of the 631 O. longicaudatus captured, 127 had wounds, of which 26 (20.5%) were seropositive to ANDV, including 23 adult males and three adult females. Of the 101 seronegatives with wounds, 90 were adults and 11 juveniles; 55 were males and 46 females. Of the 234 A. hirta captured, 67 specimens showed wounds, including five (7.5%) seropositive adults (three males and two females). For the 62 showed wounds, including five (7.5%) seropositive adults (three males and two females). For the 62 seronegative A. hirta with wounds, 51 were adults and 11 were juveniles, and 45 were males and 17 females. Of the 636 A. olivacea captured, 91 specimens had wounds, including one seropositive juvenile female and one seropositive adult male. Among the 89 seronegative A. olivacea with wounds, there were 82 adults and seven juveniles, and 59 males and 30 females.

Association Between Seropositives and Biological Traits
By using contingency tables, we analyzed the relationships between species (O. longicaudatus, A. hirta and A. olivacea) versus sex, wounds and age ( Table 2) to determine the differences among seropositives. For O. longicaudatus, seropositivity was significantly higher in males than in females (p = 0.0028), in wounded (p < 0.0001) specimens and in adults (p = 0.0066). For A. hirta and A. olivacea, neither sex, wounds nor age were associated with seropositivity. Logistic regression ( Figure 3A) using serostatus and weight showed that for O. longicaudatus, there is an increase in the probability to be seropositive for individuals of higher weight. Individuals of at least 60 g in weight have 80% or higher probability to be seropositive. This probability value drastically decreases with individuals of lower weight (i.e., an individual of 40 g has a 21% probability to be seropositive). Contrasting data were found for A. hirta ( Figure 3B). In spite of the trend in which the higher the weight higher chances to be seropositive, our study showed that an individual of 70 g has a probability lower than 20% to be seropositive.

Association Between Seropositives and Biological Traits
By using contingency tables, we analyzed the relationships between species (O. longicaudatus, A. hirta and A. olivacea) versus sex, wounds and age ( Table 2) to determine the differences among seropositives. For O. longicaudatus, seropositivity was significantly higher in males than in females (p = 0.0028), in wounded (p < 0.0001) specimens and in adults (p = 0.0066). For A. hirta and A. olivacea, neither sex, wounds nor age were associated with seropositivity.
Logistic regression ( Figure 3A) using serostatus and weight showed that for O. longicaudatus, there is an increase in the probability to be seropositive for individuals of higher weight. Individuals of at least 60 g in weight have 80% or higher probability to be seropositive. This probability value drastically decreases with individuals of lower weight (i.e., an individual of 40 g has a 21% probability to be seropositive). Contrasting data were found for A. hirta ( Figure 3B). In spite of the trend in which the higher the weight higher chances to be seropositive, our study showed that an individual of 70 g has a probability lower than 20% to be seropositive.

Human HCPS Cases
During the 19-year study period, 1060 human cases were confirmed as positive for HCPS in the 11 administrative regions in the southcentral regions of Chile sampled in this study (information available from the web page of Ministerio de Salud de Chile, http//:epi.minsal.cl). Rodent relative seropositivity was highest in the Maule region, whereas rates approaching 1.5% were found in O'Higgins, Araucanía, Biobío, Los Ríos and Los Lagos regions. We did not detect seropositive rodents in Coquimbo and Aysén regions (Figure 4), probably due to a bias in the low sampling effort (one sampling in each region). No seropositive rodents were also found in peridomestic sites in the Metropolitan region despite sampling in 14 localities (Table 1) and registering 54 HCPS cases. During the sampling period (2000-2018), the higher number of HCPS cases was registered in Los Lagos region (184 cases), and the lowest in Coquimbo (one case). However, when standardized by people inhabiting in rural areas, human hantavirus cases were notably higher in Aysén region, followed by Los Ríos, Ñuble and Los Lagos (Figure 4). During the 19-year study period, 1060 human cases were confirmed as positive for HCPS in the 11 administrative regions in the southcentral regions of Chile sampled in this study (information available from the web page of Ministerio de Salud de Chile, http//:epi.minsal.cl). Rodent relative seropositivity was highest in the Maule region, whereas rates approaching 1.5% were found in O'Higgins, Araucanía, Biobío, Los Ríos and Los Lagos regions. We did not detect seropositive rodents in Coquimbo and Aysén regions (Figure 4), probably due to a bias in the low sampling effort (one sampling in each region). No seropositive rodents were also found in peridomestic sites in the Metropolitan region despite sampling in 14 localities (Table 1) and registering 54 HCPS cases. During the sampling period (2000-2018), the higher number of HCPS cases was registered in Los Lagos region (184 cases), and the lowest in Coquimbo (one case). However, when standardized by people inhabiting in rural areas, human hantavirus cases were notably higher in Aysén region, followed by Los Ríos, Ñuble and Los Lagos (Figure 4).

Discussion
The extension of the geographical distribution of a zoonotic disease is intrinsically linked to the niche of its host. How hosts interact with human activity and settlements, and the degree of pathogen transmission among host and human populations, will determine the scale of infections that will produce the disease in humans. Hantaviruses are known to be transmitted horizontally [21,40], therefore ecological interactions are key to understanding the spread of the virus among reservoir populations. It is also known that a virus may be harbored by two or more species, with higher opportunities for host switch particularly for sympatric species [41][42][43]. Our study shows the importance of conducting long-term studies that cover a large part of the distribution of a pathogen and its hosts, allowing us to infer patterns of infection and the scale of a host's epidemiologic importance.

Discussion
The extension of the geographical distribution of a zoonotic disease is intrinsically linked to the niche of its host. How hosts interact with human activity and settlements, and the degree of pathogen transmission among host and human populations, will determine the scale of infections that will produce the disease in humans. Hantaviruses are known to be transmitted horizontally [21,40], therefore ecological interactions are key to understanding the spread of the virus among reservoir populations. It is also known that a virus may be harbored by two or more species, with higher opportunities for host switch particularly for sympatric species [41][42][43]. Our study shows the importance of conducting long-term studies that cover a large part of the distribution of a pathogen and its hosts, allowing us to infer patterns of infection and the scale of a host's epidemiologic importance. The eleven rodent species analyzed during the nineteen years of sampling in areas associated with human hantavirus cases showed different patterns of abundance in Chile. The highest relative abundance within each sampling site for O. longicaudatus was found in the Los Ríos region, for A. olivacea and A. hirta in the Biobío region, for A. longipilis and Mus musculus in the Valparaíso region, for A. sanborni in the Los Lagos region, for L. micropus and R. rattus in the Araucanía region and for P. darwini and R. norvegicus in the O'Higgins region (Table 1, Appendix A). The geographic heterogeneity among rodent abundances is expected and linked to their ecology. For example, despite the wide distribution of O. longicaudatus in Chile (from the southern of Atacama desert to Patagonia), the primary habitat of this species occurs in mesic areas that are more prominent between 35-50 • S in the Valdivian and Patagonian rain forest ecoregions [36,44]. Our study supports this inference, as more O. longicaudatus were captured in localities of Biobío, Araucanía, Los Ríos and Los Lagos regions that fall within the Valdivian rain forests (Appendix A). To understand the scale of the relation between abundance of host and prevalence of the pathogen, we compared the host relative abundance and pathogen infection measured as relative seropositivity. Comparing sampling localities, we found heterogeneous results (Table 1). For example, high abundance of O. longicaudatus and absence of seropositives were registered in Vilumanque (Concepción), and Ñancul (Panguipulli), contrasting with localities with low abundance of this species and high seropositivity (e.g., Las Peñas, San Fernando, O'Higgins and La Palma, Valparaíso). This expected pattern was also reported by several studies [20,30], emphasizing the uncertainty of sampling the host carrying the virus despite the epidemiological and ecological analyses to determine the most probable site of human infection [45]. We acknowledge that our estimations of seropositivity might be biased due to an error associated to false positives and/or negatives determined by the SIA method [46]. However, we expect a low error in this parameter [46,47].
In the present study, the seroprevalence of 3.14% for all small mammals and 3.63% for sigmodontine rodents was higher to that reported previously in Chile [20,30,48], an expected result due to our sampling scheme linked to recent confirmed HCPS cases. The variation of total seroprevalence reported in the Americas includes similar percentages to those of our study, such as in Argentina (1.5% in the north, 2.6% in the center and 5.4% in the south) [49], Brazil (2.1%) [50] and North America (average 5%) [51], but differs from others with higher rates such as Panama, Mexico and United States of America [52][53][54][55].
For O. longicaudatus, we found an intraspecific seroprevalence of 6.5%, a value within the range of other studies in Chile reporting seroprevalences of 10.4% [20] and 2.5% [30]. The heterogeneity observed in the seroprevalence of rodent populations depends on several factors impacting the ecological dynamics of the host and the virus, such as climate [56], landscape structure [52,57], density-based and/or frequency-based dependence [58][59][60], dilution effect [61], viral recrudescence [47] and habitat disturbance [26,62]. Our sampling scheme also allowed us to show new evidence for the prevalence of antibody against ANDV in Chile in species other than the major host. We found an intraspecific seroprevalence of 5.13% for A. hirta and 0.79% for A. olivacea. For many years, A. longipilis was identified as the second species with higher seroprevalence in Chile. However, new taxonomic rearrangements show that those results were skewed, because the results should be reassigned to A. hirta [33,42]. Taking this into account, the seroprevalence for A. hirta ranges from 9.3% in Los Ríos region [63] to 1.84% in Los Lagos region [21], representing the highest seroprevalence for hantavirus in Chile after the main reservoir O. longicaudatus. Compared with O. longicaudatus (6.5%), our seroprevalence data for A. hirta (5.13%) acquires new relevance, particularly in understanding its role in maintaining the virus in nature, or potentially as host of a new lineage of hantavirus of unknown pathogenicity [5,36,64]. Based on ours and all previous published data, no A. longipilis have been found seropositive to ANDV in Chile. For A. olivacea, the most abundant species captured, an intraspecific seroprevalence of 7.5% and 0.73% has been reported in Aysén region [29,65,66], 2.2% in Los Lagos region [66] and 1.5% in Chubut, Argentina [67]. Hantaviruses are capable of infecting multiple host species [5,45], and host-switching events have been reported between species of the same and different families and also separate orders [68]. Cross-species transmission is more likely to occur among sympatric and evolutionary related species, so the study of their ecological interactions is key to understand the processes leading to species jumps and emergence of outbreaks. ANDV experimental transmission was demonstrated from O. longicaudatus to A. olivacea [21], but the epidemiological importance of A. hirta and A. olivacea remains unclear [42]. We highlight that our study is one of the few long-term involving only areas associated with human hantavirus cases, therefore the seroprevalences may also differ from others covering alternative sampling designs.
The relationships between species (O. longicaudatus, A. hirta and A. olivacea) versus sex, wounds and age have been frequently studied to infer major determinants of spread and maintenance of viruses in nature. It has been proposed that most of seropositive O. longicaudatus are adult males, which was evaluated in this study. A positive relationship between weight and seropositivity in O. longicaudatus suggests that adults are more likely to be infected than juveniles. Although we did not search for viral RNA in seronegative rodents, there it might be a proportion of rodents that are viremic and without antibodies that might be excreting the virus into the environment. We found that adult males (32/41 of positives) are more likely to be seropositive than females (9/41 of positives; Table 2), a result congruent with studies in several wild reservoirs [20,67,[69][70][71][72][73]. Our logistic regression (Figure 2) also shows that O. longicaudatus weighing 60 g or more were more likely (80%) to be infected with ANDV. A slightly different analysis in the same species in Argentina showed that individuals not injured of 50 g (approx.) and those of 43 g (approx.) with wounds, have a probability of 0.80 to be infected with the virus [73]. Our analysis was performed including both wounded and non-wounded O. longicaudatus from all ages. We chose this approach because the role of wounds in hantavirus transmission is still in debate (see below). Adults seem to be more likely to contract the virus because transmission of the virus requires a prolonged period of exposure [74]. O. longicaudatus adult females were less likely to be seropositive, suggesting that adult males may have a major role in maintaining the virus transmission in nature. However, the frequency of seropositive females found across several studies is evidence that they may play a more important role than the one currently recognized. Space use and social mating system may contribute to reveal the factors maintaining the virus in wild populations and explaining sex differences of infected hosts [75]. However, viral transmission through aggressive encounters including frequent bites among adult males have been proposed as major hypothesis to explain sex differences [56,[76][77][78].
We found 127 O. longicaudatus with wounds of which 26 were seropositive (23 males and three females). Our results provide evidence that O. longicaudatus adult males with wounds would be more likely to be seropositive ( Table 2). An association between wounds and seropositive status suggests that antagonistic encounters between reproductive individuals would be a mechanism of transmission between host populations [79][80][81]. Interestingly, the probability that an O. longicaudatus adult male in Argentina is seropositive is five times higher in injured individuals [73]. The previous study showed that adult males with wounds with a body-mass greater than 44g have a probability of 0.84 to be seropositive, while individuals with wounds and a body-mass less than 30 g have a probability of being seropositive of 0.2. Although wounds seem to be the mechanism for intraspecific transmission of hantaviruses, a few studies have found different results. For example, in Peromyscus maniculatus, the prevalence seemed to reach a peak much earlier in individuals with wounds, possibly indicating that aggression is not the only mechanism of transmission of Sin Nombre orthohantavirus among deer mice [69]. The relationship between hosts with wounds and seropositivity might also be a consequence of the infection [78,80,[82][83][84], or it might be related to a greater population density of the reservoir, resulting in viral transmission caused by the greater environmental contamination rather than by aggression among rodents [80]. No relationship between infection and individuals with wounds was found among the O. longicaudatus in southern Argentina [67]. There are also social behaviors such as identification, mounting, smelling and persecution that favor physical unaggressive contact among individuals [85]. Field experiments in O. longicaudatus do not support a relationship between the presence of wounds and the transmission of ANDV, indicating that other factors such as grooming or aerosol transmission may be playing a major role for efficient viral transmission [21].
Unlike O. longicaudatus, seropositive juveniles were identified in the two species of Abrothrix. For A. hirta, one juvenile male out of 12 was infected. Three adult males and two juveniles (one male and one female) were seropositive for A. olivacea. The number of Abrothrix infected with hantavirus is frequently low or absent compared to that of sympatric O. longicaudatus, both in areas with or without human cases [20,30,86]. In addition to global seroprevalence and sex, we tested if wounds were related to infected Abrothrix. We found no association in both species of Abrothrix (5/12 seropositives of A. hirta and 2/5 for A. olivacea with wounds; Table 2), similar to a previous study [63]. Despite some efforts to study viral ecology of coexisting species with the main reservoir, the epidemiological relevance of Abrothrix is still unknown [21,29,30,42,87].
The ecological properties of host and vectors are influenced by environmental features [88], which may be strong determinants in the transmission of the virus to humans across different spatial and temporal scales [36,89]. The wide latitudinal range of our samplings ranges from a Mediterranean heterogeneous vegetation mosaic to a mixed evergreen-deciduous temperate and Patagonian forest (30 • 30'S-45 • 30'S) [90,91]. These areas are characterized by contrasting landscapes resulting in strong differences in population structure of O. longicaudatus [92,93], A. hirta [94] and A. olivacea [95]. Variability in the population dynamic of O. longicaudatus populations across their latitudinal range in Chile may impact transmission and infection rates of ANDV in rodent hosts [36]. Therefore, humans may be differentially exposed to viral infection. A variety of studies reported high differences in seroprevalence of O. longicaudatus in Chile across administrative regions, ranging from 23% for the O'Higgins region [66], 12.74% and 12,7% for the Aysén region [29,65], 13.51% in the Los Ríos and Los Lagos regions [63] to 1.44% in the Magallanes region [96]. Interestingly, we found that rodents' relative seropositivity was higher in the Maule region (around 34-36 • S). This region is within the Mediterranean ecoregion, which showed the highest relative seropositivity in Chile [36]. Complex interacting determinants may be involved in explaining this result [97,98]. Landscape structure and hosts' responses to disturbances and habitat modifications may strongly influence reservoir populations [99] and infectious disease dynamics [98]. For example, hosts may become the dominant species in degraded ecosystems [100], and agroecosystems may play a relevant but still unknown role in harboring infected hosts, and therefore in the transmission of virus to humans [101]. Agriculture and habitat modification, features that characterize the Maule region, are also found in several other regions where we detected high seroprevalence (Figure 4). These inferences acquire more relevance when small mammal studies are linked to areas where humans are at higher risk of contracting the pathogen such as domestic/peridomestic settings [26,49,102]. Similar values of relative seropositivity were found across O'Higgins, Araucanía, Biobío, Los Ríos and Los Lagos regions. We did not detect seropositive rodents in Coquimbo and Aysén regions (Figure 4), probably due to the low comparative sampling effort (one sampling locality in each region). No seropositive rodents were found in peridomestic sites in the Metropolitan region despite sampling in 14 localities (Table 1) and registering 54 confirmed HCPS cases. Whether these cases contracted the infection outside of this region or we were unable to detect seropositives due to our sampling scheme remains to be elucidated. We observed a trend of finding an increased number of HCPS cases from north to south (Figure 4). The high number of reported HCPS cases found the Aysén region was likely influenced by the approach we used (HCPS cases/1000 rural inhabitants). In fact, rural inhabitants in Aysén were significantly lower compared to the other regions reported in this study; also 61 HCPS cases were confirmed in the Aysén region during the study period while a higher number of cases were reported for Los Lagos (N = 184), Araucanía (N = 158) and Biobío (N = 156) regions. Despite the differences in the sampling efforts across administrative regions, the geographic and temporal span of our study highlights the heterogeneity of HCPS cases and the spatial fluctuations of seropositive hosts. Consequently, these data emphasize that integrating ecological understanding of host and pathogens, spatial and temporal surveillance, epidemiology and public health agencies is fundamental to i) understanding the driving processes that ultimately lead to the transmission of virus to humans and ii) successfully designing policies and campaigns to prevent the infection from zoonotic diseases [26,52,103,104].

Conclusions
Our study assessed how ecology and geography influence host and viral dynamics in areas associated with HCPS cases in Chile. We confirmed the major role of O. longicaudatus as a reservoir of ANDV in Chile and the significance of adult males in maintenance and transmission of the virus in nature. Our results indirectly show that wounds might play a significant role in the intraspecific transmission of the virus, which coupled with alternative routes previously reported, suggest that to fully understand the host-viral transmission, further investigation is required on social and behavior interactions of rodents. We highlighted that the epidemiological importance of A. hirta and A. olivacea in maintaining and/or transmitting the virus among wild populations and to humans remains unclear.
The high variation in rodents' abundance and seroprevalence across a wide geography, emphasizes the need for i) understanding the drivers that determine the dynamic of fluctuating populations of reservoirs, ii) determining the role of other interacting species for the viral spread and iii) integrating efforts from different disciplines to more accurately determine transmission rates of pathogens within host populations, host switch and ultimately to humans. Elucidating these patterns is critical to prevent future zoonotic outbreaks by implementing effective intervention strategies.