Hantavirus in rodents in the United States: Temporal and spatial trends and report of new hosts

Abstract

In North America, the rodent-borne hantavirus pulmonary syndrome is predominantly caused by the Sin Nombre virus, typically associated with the deer mouse Peromyscus maniculatus. Utilizing data from the National Ecological Observatory Network (NEON) hantavirus program, we assessed factors that may influence spatial and temporal distribution of hantavirus in rodent populations across the United States. Between 2014 and 2019, the NEON hantavirus program conducted 104,379 small mammal captures and collected 14,004 blood samples from 49 species at 45 field sites. Our study identified 296 seropositive samples across 15 rodent species, including eight Peromyscus species. We describe six new species with hantavirus seropositive samples not previously reported as hantavirus hosts. The highest number of seropositive samples were obtained from P. maniculatus (n=116; 2.9% seroprevalence), followed by P. leucopus (n=96; 2.8%) and Microtus pennsylvanicus (n=33; 4.2%). Hantavirus seroprevalence showed an uneven spatial distribution, with the highest seroprevalence found in Virginia (7.8%, 99 seropositive samples), Colorado (5.7%, n=37), and Texas (4.8%, n=19). Hantavirus seropositive samples were obtained from 32 sites, ten of which presented seropositive samples in species other than P. maniculatus or P. leucopus. Seroprevalence was inconsistent across years but showed intra-annual bimodal trends, and in P. maniculatus and P. leucopus the number of captures correlated with seroprevalence in the following months. Seroprevalence was higher in adult males, with one seropositive sample obtained from a juvenile Peromyscus truei. Higher body mass, scrotal testes and non-pregnant status were associated with higher seropositivity. The NEON dataset, derived from a multiyear and structured surveillance system, revealed the extensive distribution of hantavirus across broad taxonomic and environmental ranges. Future research should consider winter season surveillance and continued analyses of stored samples for a comprehensive spatiotemporal study of hantavirus circulation in wildlife. Global changes are expected to affect the dynamic of rodent populations by affecting their availability or resources and demography and, consequently, may modify transmission rates of rodent-borne zoonotic pathogens such as hantavirus. This study can be considered a baseline to assess hantavirus patterns across host taxa, geographies, and seasons in the United States.

Introduction

Hantavirus infections, caused by viruses from the Hantaviridae family within the order Bunyavirales, have been documented in at least 140 animal species, including carnivores (e.g., red fox Vulpes vulpes) and bats (Guo et al. 2013; Milholland et al. 2018). Despite this broad host range, most hantavirus lineages are primarily associated to a few phylogenetically related rodent species (Milholland et al. 2018). Rodents usually present asymptomatic and persistent infections (Ermonval, Baychelier, and Tordo 2016; Forbes, Sironen, and Plyusnin 2018).

In humans, hantavirus infections can lead to severe diseases, notably hantavirus pulmonary syndrome (HPS) (Laenen et al. 2019). The first report of HPS in the United States occurred in 1993 in the Four Corners region of New Mexico, and since then, numerous cases have been documented across the country, primarily related to the Sin Nombre virus (SNV) (Luis et al. 2015; Van Hook 2018; Nichol et al. 1993). Human infections typically occurs through the inhalation of aerosolized excreta, urine, or saliva from infected rodents (Botten et al. 2002; Bagamian, Towner, et al. 2012; Forbes, Sironen, and Plyusnin 2018).

New world hantaviruses are periodically identified in multiple hosts, usually with unknown zoonotic potential (Ermonval, Baychelier, and Tordo 2016; Milholland et al. 2018; Mull et al. 2022). In North America, at least 17 hantaviruses are known to co-circulate, predominantly carried by New World rodents of the Family Cricetidae, subfamily Sigmodontinae (CDC 2012; Milholland et al. 2018; 2019). The primary recognized rodent host of SNV is the deer mouse (Peromyscus maniculatus) (Childs et al. 1994; Luong et al. 2011; Ermonval, Baychelier, and Tordo 2016). Nevertheless, SNV infections have also been identified in other rodent species, particularly cricetids such as Peromyscus leucopus, Neotoma micropus, and Sigmodon hispidus, among others (Childs et al. 1997; Milholland et al. 2018). The widespread distribution of P. maniculatus across the continental United States complicates the understanding of the roles other rodent species play in the distributional ecology of SNV (Forbes, Sironen, and Plyusnin 2018; Milholland et al. 2018; Chen et al. 2023).

Estimating the transmission risk of SNV to humans requires an comprehensive understanding of the complex hantavirus dynamics within multiple wildlife host populations that sustain viral circulation (Goodfellow et al. 2021; Shubangi J. et al. 2022; Forbes, Sironen, and Plyusnin 2018). Moreover, the biological, ecological, and environmental factors that maintain local hantavirus circulation within rodent communities remain insufficiently understood (Padula et al. 2004; Heyman et al. 2012; Bagamian, Towner, et al. 2012; Carver et al. 2015). This study aimed to enhance our understanding of the host species involved in hantavirus circulation in North America and to elucidate the spatial and temporal patterns of virus maintenance in the sylvatic cycle. To achieve this, we analyzed a comprehensive dataset from the Small Mammal Box Trapping and the Rodent Pathogen Status programs of the National Ecological Observatory Network (NEON) (Thibault 2022; NEON (National Ecological Observatory Network) 2023), which provides open and interoperable data from multiple sites across the United States, employing sustained and standardized capturing and sampling protocols (Paull et al. 2023). This rodent-pathogen NEON dataset integrates trait data of rodents such as hantavirus seroprevalence, sex, reproduction status, life stage, and morphometrical measurements (Thibault 2022; National Ecological Observatory Network (NEON) 2023a).

Methods

NEON data

The National Ecological Observatory Network (NEON) is a continent-scale observation facility designated to collect long-term, open-access ecological data, encompassing biological, chemical, and physical measurements and samples across the United States (National Ecological Observatory Network (NEON) 2023b). Between 2014 and 2019, the NEON Small Mammal Box Trapping program implemented the rodent hantavirus seroprevalence testing program (product DP1.10072.001) and generated the Small Mammal Box Trapping and Rodent Pathogen Status databases (product DP1.10064.001) (National Ecological Observatory Network (NEON) 2023b).

This database contains individual reports of small mammals captured and sampled at terrestrial field sites across 18 NEON domains (Figure 1). Due to regulatory changes and conservation concerns, blood samples were not collected in Hawaii, Puerto Rico, and Yellowstone National Park field sites (Thibault 2022; National Ecological Observatory Network (NEON) 2023a). At each field site, captures are organized within multiple 90×90 meter grids, each containing 100 traps spaced 10 meters apart. Blood samples for hantavirus serology were collected in three designated “pathogen grids” at each NEON site. Captures and sampling typically occurred over one to three consecutive nights per sampling bout, with flexibility to accommodate logistical challenges posed by extreme or unexpected events such as flooding, fires, or severe weather conditions (Chesney, Thibault, and Paull 2023; Paull et al. 2023). Sampling was scheduled to occur once every lunar cycle (approximately every four weeks) at 19 core sites, and every second lunar cycle at the 27 gradient sites (Springer et al. 2016). A maximum of 20 individuals were expected to be sampled per grid and collection date (Thibault 2022; Paull et al. 2023). From 2017 onwards, only a subset of collected blood samples, up to 140 samples per site per year, were analyzed (Thibault 2022).

Figure 1.

Distribution of NEON field sites, blood samples, and seroprevalence. The figure illustrates the geographic distribution of NEON field sites across the United States (black dots), and sites with hantavirus seropositive samples (yellow squares). Data are aggregated by state to depict the number of blood samples (indicated by circle size) and the seroprevalence (represented by color scale). The number of blood samples are defined for three groups: (A) all rodent species, (B) Peromyscus maniculatus, and (C) Peromyscus leucopus. Sites with no seropositive samples (seroprevalence=0) are shown as transparent circles.

Species sampled for hantavirus serology belonged to the families Cricetidae, Muridae, and Dipodidae. Each field site prioritized specific species, with Peromyscus species, particularly P. maniculatus and P. leucopus, identified as priority species for pathogen testing (Springer et al. 2016; Paull et al. 2023). Samples were collected only from individuals weighing more than 10 grams and appearing to be in good physical condition (Thibault 2022). NEON guidelines stipulate that individuals should not be resampled within an ongoing bout, though they may be resampled in subsequent bouts (Thibault 2022). Serum samples were screened using IgG ELISA, following methodologies described by Klingström et al. (2002).

We standardized NEON data using the neonUtilities package, with all R codes to process NEON data products available (Lunch et al. 2018). The NEON Small Mammal Trapping database includes detailed individual data such as taxonomic identification, age (life stage), sex, reproductive condition (pregnancy status and testes, nipples, and vagina conditions), and standardized measurements of hind foot length and body mass. The protocol mainly considers the distances between clitoris or genital papilla and anus to determine sex, testes presence and size for age, and vaginal opening to determine reproductive conditions (Paull et al. 2023). Additional measurements, such as ear length, tail length, and total length, are taken optionally to facilitate more accurate species identification (Paull et al. 2023).

Data Analyses

Species of the genus Peromyscus, particularly P. maniculatus and P. leucopus, were designated as priority species for hantavirus testing in NEON protocols (Springer et al. 2016; Paull et al. 2023), as they are considered pivotal in the transmission dynamics of the SNV (Childs et al. 1994; Milholland et al. 2019; Luong et al. 2011). All our analyses were conducted on three datasets: (1) all small mammal species captured, (2) dataset filtered to only include P. maniculatus, and (3) dataset of P. leucopus. Seroprevalence was estimated as a proportion of the seropositive samples in relation to all blood samples. We examined associations between hantavirus seroprevalence and sex, reproductive status, and life stage using the prop.test() function in R (RStudio v2023.09, RStudio Team 2023), which performs parametric test for equality of proportions based on chi-square (X²) statistics. Additionally, we assessed potential biases in the number of captures, blood samples and seroprevalence metrics.

To explore temporal patterns, we constructed time series models to assess the relationship between two time series: the number of captures (𝑥𝑡) and seroprevalence (𝑦𝑡). We employed the sample cross-correlation function (CCF) in the “tseries” package in R (Trapletti, Hornik, and LeBaron 2024) to identify lags of the x-series that might predict 𝑦𝑡 (Tenzin, Dhand, and Ward 2011; Shrestha et al. 2024). The CCF calculates correlations between 𝑥𝑡+ℎ and 𝑦𝑡 for h=0, ±1, ±2, ±3, examining whether the x-variable serves as a leading indicator for the y-variable, thereby determining the effect of capture numbers on future hantavirus seroprevalence. One or more 𝑥𝑡+ℎ values with negative ℎ is considered a correlation between x-variable at a time before 𝑡 and the y-variable at time 𝑡 and interpreted as the x-variable leading the y-variable (i.e., “x leads to y”). Conversely, positive ℎ values suggest that the x-variable lagged behind the y-variable (i.e., “x lags to y”). The level of statistical significance for CCF analysis was set as α=0.05 (Shrestha et al. 2024).

In addition, we applied a multiplicative decomposition to the time series to describe trends and patterns of change across years and seasons (forecast Package) (Hyndman and Athanasopoulos 2018). The time-series analysis included three components: Trends, indicating the direction and variability rate over years, allowing for the identification of stable or changing patterns; Seasonality, reflecting regular fluctuations at consistent intervals, such as seasons or years, and Reminders, representing residuals or noise, which captures irregular or random variations not explained by the other components and reflects data uncertainty. Finally, to assess the relationships between hantavirus seroprevalence and body measurements (i.e., body mass, hind foot length, total body length, ear length, tail length), we used the wilcox_test() function from the rstatix package in R (Kassambara 2023). This test is a non-parametric alternative to the t-test, suitable for comparing two groups when data distributions are skewed.

Results

From 2014 to 2019, the NEON small mammal trapping program recorded 104,379 captures, with P. maniculatus and P. leucopus comprising 21.9% and 17.1% of these captures, respectively (Table 1). A total of 127 species were captured and 49 species were sampled, resulting in the collection of 14,004 blood samples over the six-year period (Appendix S1: Table S1). Blood samples collected and analyzed represented 13.4% of all captures. For example, blood was collected for 17.1% of P. maniculatus captures (3919 blood samples) and for 19.2% of P. leucopus captures (3460 blood samples). From the pool of blood samples, Peromyscus species accounted for 64.7% of the total blood samples collected (n=9059), principally P. maniculatus representing 28.1% and P. leucopus 25% of the sample pool (Figure 2, Table 1). We identified 135 blood sample records labelled with ambiguous scientific names, including 88 as P. gossypinus/leucopus, 47 as P. leucopus/maniculatus, and 71 blood sample records were labelled at genus level including Peromyscus and Myodes spp.

Table 1.

Captures, sites, blood sampling and hantavirus serology test results for All rodents, Peromyscus maniculatus and P. leucopus.

Variable	All rodents	P. maniculatus	P. leucopus
No. captures	104,379	22,904 (21.9% of all captures)	17,999 (17.1% of all captures)
No. sites with captures	46	31 (67.4% of all sites)	23 (50% of all sites)
No. sites sampled	45 (97.8%)	30 (96.8%)	21 (91.3%)
No. blood samples collected	14,004 (13.4%)	3919 (17.2%)	3460 (19.2%)
No. seropositive samples	296 (2.1%)	116 (3.0%)	96 (2.8%)
No. sites with seropositive samples	32 (71.1%)	17 (56.7%)	12 (57.1%)

Note: Information on other species is detailed in Appendix S1: Table S1.

Figure 2.

Captures, blood samples and hantavirus seroprevalence by rodent species. (A): Hantavirus seroprevalence (seropositive/total blood samples) for each of the 15 seropositive rodent species. (B) Number of blood samples and the number of captures for each species. Notably, Myodes pennsylvanicus and Peromyscus keeni exhibit high seroprevalence, although both species have relatively low numbers of blood samples and captures. In contrast, P. maniculatus and P. leucopus present high seroprevalence as well as substantial numbers of captures and blood samples.

Seropositivity was detected in 15 rodent species (30.6% of the species sampled) (Figure 2), with a total of 296 seropositive samples (2.1% seroprevalence). Most of the seropositive samples were collected from Peromyscus species (79.1%, 234 seropositive samples). The highest seroprevalence was obtained from Peromyscus truei (4.9%, 9 seropositive samples), followed by Microtus pennsylvanicus (4.3%, 33 seropositive samples) and P. maniculatus (3.0%, 116 seropositive samples). Filtering for species that contributed with ≥1% (n>140) of the blood samples from the total pool, we found no seropositives for Myodes rutilus, Neotoma albigula, Zapus hudsonius, and Onychomys torridus (Appendix S1: Table S1). We found seropositive samples in six species for which we did not find previous reports of hantavirus exposure or infection: Peromyscus keeni, Peromyscus gossypinus, Peromyscus polionotus, Myodes gapperi, Podomys floridanus, and Napaeozapus insignis.

Spatial trends

Small mammals were captured across 46 terrestrial sites, and blood samples were collected from all sites except YELL, located in Yellowstone National Park. Captures and sampling activities in Puerto Rico (LAJA and GUAN sites) stopped in 2017. The site with the highest number of captures was SRER in Arizona, with 7285 captures, followed by ONAQ in Utah with 7157 captures, and HARV in Massachusetts with 7010 captures (refer to Figure 1, Table 1, Appendix S1: Table S2). At the state level, Utah recorded the highest number of captures (9379 captures), followed by Kansas (8097), and North Dakota (7780). At the NEON site level, the KONZ site in Kansas yielded the highest number of blood samples, with 1109 samples collected, representing 23.6% of captures at that site. This was followed by the WOOD site in North Dakota, with 764 blood samples, accounting for 14.9% of site captures. The median number of captures and blood samples per site was 415 and 77, respectively. At the state level, the largest numbers of blood samples were collected in Kansas (2014 blood samples), Virginia (1261), and North Dakota (1144). Seropositive samples were identified at 32 NEON sites, which constitutes 71.1% of the sampled sites, distributed across 17 of the 23 surveyed states (73.9%) (Figure 1, Table 1; Appendix S1: Table S2).

At the site level, the highest seroprevalence was observed at CPER in Colorado, with 14.3% based on one seropositive sample from the seven blood samples collected, which reflects that blood samples were collected in 0.5% of the 1602 captures in the site. Other NEON sites with notable seroprevalence included SCBI (11% seroprevalence, 65 seropositive samples) and MLBS (10.2% seroprevalence, 30 seropositive samples), both located in Virginia.

At the state level, Virginia exhibited the highest seroprevalence at 7.9% (n=99 seropositive samples), followed by Colorado at 5.7% (n=37), and Texas at 4.8% (n=19). Notably, five of the 45 sites (11%) where hantavirus sampling was conducted did not yield any seropositive samples. These sites included locations in Michigan (826 blood samples), Alaska (570 blood samples), and Arizona (473 blood samples). Similarly, Kansas (2014 blood samples), Wisconsin (899 blood samples), and Massachusetts (745 blood samples) exhibited low seroprevalence (<1.2%) despite the high number of blood samples collected (Figure 1; Appendix S1: Table S2).

Peromyscus maniculatus were captured at 31 out of 46 sites, with the highest capture numbers recorded at ONAQ in Utah (2218 captures), followed by UNDE in Michigan (2057 captures) and HARV in Massachusetts (1997) (Table 1). Blood samples from P. maniculatus were collected at all sites where this species was captured, except in YELL site, with seropositive samples detected at 56.6% of these sites (17 sites) (Table 1, Figure 1). The highest seroprevalence for P. maniculatus was observed at the STER site in Colorado, reaching up to 15.7% (27 seropositive samples). At the state level, North Dakota had the highest number of P. maniculatus captures (2937 captures), followed by Colorado (2610) and Utah (2606). The highest number of blood samples from P. maniculatus were collected in Kansas (505), Michigan (427), and Wisconsin (423). Notably, the highest seroprevalences for P. maniculatus were observed in Virginia (14.5%), Colorado (7.8%), and Texas (7.1%).

Peromyscus leucopus were captured at 23 sites across 14 states, with the highest capture numbers at SCBI in Virginia (2768 captures), followed by ORNL in Tennessee (2327 captures), and HARV in Massachusetts (2239 captures) (Table 1). Blood samples from P. leucopus were collected at 21 of the 23 sites where this species was captured (91.3%), with seropositive samples identified at 12 (57.1%) of them (Figure 1, Table 1). The nine sites sampled for P. leucopus that did not yield seropositive samples presented a low number of blood samples (e.g., DELA with three samples, LENO with two, OAES with 21, SRER with one) (Figure 1; Appendix S1: Table S2). The SCBI site in Virginia had the highest number of blood samples from P. leucopus (457 blood samples), as well as the highest seroprevalence (12.9%, 59 seropositive samples). At the state level, Virginia also recorded the highest number of blood samples (n=865) and the highest prevalence (8.2%) for P. leucopus. Following SCBI, the KONZ site in Kansas had the second highest number of blood samples from P. leucopus (n=411). Hantavirus seropositive samples were identified in 10 sites where no seropositive samples were identified in P. maniculatus or P. leucopus. These sites were distributed across southeastern states (Alabama, Florida, Georgia), northern regions (North Dakota, Washington, Wisconsin), and Colorado (Figure 1).

Sex and life stage

A significantly higher number (X²=366.76, df=1, p=9.48×10⁻⁸²) of captured rodents were identified as males (n=52,433) compared to females (n=46,441). This pattern was consistent with the number of sampled individuals (X²=85.57, df=1, p=2.24×10⁻²⁰) and seroprevalence (X²=26.94, df=1, p=2.10×10⁻⁷), with 7501 sampled males (2.6% seroprevalence) and 6409 sampled females (1.6% seroprevalence; Figure 3; Appendix S1: Table S3). A total of 5.3% of the captures (n=5535) and 0.7% of the blood samples (n=94) were recorded with undetermined sex or lacked sex identification. Seroprevalence was significantly higher (X²=13.55, df=1, p=0. 24×10⁻³) in males with scrotal testes (n=122, 3.6% seroprevalence) compared to non-scrotal males (n=70, 1.6% seroprevalence), and pregnancy was negatively correlated with seroprevalence (X²=63.21, df=1, p=1.85×10⁻¹⁵), with 1.7% seroprevalence in pregnant females (15 of 866 blood samples) and 1.9% in non-pregnant ones (128 of 6567 blood samples; Figure 3; Appendix S1: Table S3).

Figure 3.

(A, B) Seroprevalence, (C, D) numbers of captures, and (E, F) numbers of blood samples according to sex, life stage, and reproductive status. These variables are shown separated for all rodents and for Peromyscus maniculatus and P. leucopus grouped. For life stage, juveniles and subadults were grouped. Significant differences of seroprevalence were identified for scrotal testes and pregnancy status. For better visualization, all variables are presented as binomial, with two possible outcome categories. More details about the results of these variables can be found in Appendix S1: Table S4.

Regarding life stage, the majority of the captures (X²=124997, df=1, p=2.2×10⁻¹⁶) and blood samples (X²=16857, df=1, p=2.2×10⁻¹⁶) were from adults (captured males=87,022; male blood samples=11,889; Figure 3). Higher seroprevalence was found in adults (2.3%, 277 of 11,889 blood samples) compared to subadults (1.0%, 16 of 1580 blood samples) and juveniles (0.2%, one of 475 blood samples; X²=487.69, df=1, p=4.54×10⁻¹⁰⁸). The only seropositive sample from a juvenile was a P. truei rodent (Appendix S1: Table S3).

Time and seasonal trends

Captures were conducted annually from 2014 to 2019 at 13 sites, although blood samples were consistently collected each year only at the SCBI site in Virginia. The median number of captures per year per site was 415, while the median number of blood samples was 77. Captures tended to peak in May and September, corresponding to spring and summer months, exhibiting a moderate bimodal intra-annual pattern (Figure 4; Appendix S1: Figure S1). Conversely, captures decreased during winter months from December to February, a period when blood samples were infrequently collected. In January, only 43 blood samples were collected, all from 2018, with a single seropositive sample identified in a P. boylii rodent. In February, 73 blood samples were collected, with 51 samples from 2018 and 23 in 2019, yet no seropositive sample were found (Appendix S1: Figure S2).

Figure 4.

Numbers of captures and seroprevalence for (A) all rodents, (B) Peromyscus maniculatus, and (C) P. leucopus across the 2014–2019 period. Peaks in seroprevalence (e.g., February 2016 in all rodents and in P. leucopus, December 2018 in P. maniculatus) tend to coincide with a lower number of captures. The number of blood samples and seropositive rodents are shown in Appendix S1: Figure S3.

Captures trends (Figure 4) displayed greater consistency across years compared to seroprevalence, as indicated by the more uniform residual structure (Appendix S1: Figure S2). Higher inconsistencies were noted for P. maniculatus compared to P. leucopus (Appendix S1: Figure S3, Figure S4). Hantavirus seroprevalence tended to peak during spring and fall, specifically in September (year 2014), between May and June (year 2015), between March and April, as well as in October-November (years 2016, 2017, 2018) (Figure 4; Appendix S1: Figure S3). Intra-annually, a bimodal seasonal trend was observed across all rodent species, yet no significant lagged correlations were found between the number of captures and seroprevalence within a 95% confident interval. Thus, the number of rodents captured did not explain or predict seroprevalence variation when considering all rodents collectively (Appendix S1: Figure S2).

For P. maniculatus, seroprevalence peaks typically occurred between September and November (years 2014, 2016, 2017, 2018) and between March and May (years 2018, 2019). In November 2018, a peak in P. maniculatus seroprevalence was observed when only two blood samples were collected, both testing positive, resulting in a seroprevalence of 100%. The cross-correlation function showed a positive correlation (CCF(r)=0.2432 p=0.039) at a −2-month lag in P. maniculatus, suggesting that an increase in the number of captures is associated with an increase in the average seroprevalence two months later (Figure 4; Appendix S1: Figure S2, Figure S4).

The seroprevalence of P. leucopus exhibited bimodal peaks in spring (March-April) and fall (September-October) (Figure 4; Appendix S1: Figure S1). The 2016 peak in P. leucopus seroprevalence observed in April (20%) was based on four seropositive samples out of 20 collected, while the 2018 peak observed in March resulted from two seropositive samples out of 13 collected. We identified that the average number of P. leucopus captures could predict future seroprevalence, showing both positive and negative cross-correlations. Significant positive correlations were observed at lags −9 (CCF(r)=0.3297; p=0.005), −10 (CCF(r)=0.3082 p=0.008), and −11-months (CCF(r)=0.2455; p=0.037), suggesting that an increase in P. leucopus captures is associated with increased seroprevalence 9 to 11 months later. Conversely, significant negative correlations were found at lags −4 (CCF(r)=−0.344; p=0.003) and −5-months (CCF(r)=−0.3354; p=0.004), indicating that an increase in P. leucopus captures is associated with a decrease in seroprevalence 4 to 5 months later (Figure 4; Appendix S1: Figure S1).

Morphometry and hantavirus seroprevalence

Body mass data were available for 87,570 (83.9%) captures and 12,863 (91.9%) blood samples, while hind foot length was recorded for 83,239 (79.7%) captures and 12,356 (88.2%) blood samples (Appendix S1: Figure S5). Total length measurements were missing in 91.5% of captures (n=95,719) and 89.6% of blood samples (n=12,549). Ear length was not recorded in 80.2% of captures (n=83,737) and in 72.3% of blood sample records (n=10,124), while tail length was absent in 73.0% (n=76,239) of captures and 63.9% of blood samples (n=8950). Specifically, for Peromyscus species, body mass was recorded in 44,499 (88.3%) captures and in 8454 (93.32%) blood samples, while hind foot length was available for 42,445 (84.2%) of the captures and in 8253 (91.1%) of the blood sample records.

In P. maniculatus, seropositive individuals were found having larger body masses (20.8 g) when compared with seronegative ones (18.6 g; W=−6.67, df=104.8, p=1.22×10⁻⁹), and ear lengths tended to be smaller in seropositive rodents (16.7 g) when compared to seronegatives (17.0 g; W=3.72, df=33.4, p=7.26×10⁻⁴). Similarly, for P. leucopus, seropositive rodents tended to have a larger body mass (23.5 g) when compared with seronegative individuals (22.7 g; W=−2.14, df=91.2, p=0.035) (Figure 5; Appendix S1: Table S4 shows detailed values for all body measurements).

Figure 5.

Differences of biological measurements in (A) P. maniculatus and in (B) P. leucopus between seropositive and seronegative individuals. Body mass was found to be significantly different between seropositive and seronegative individuals in both Peromyscus maniculatus and P. leucopus, and ear length was significantly different only for P. maniculatus. Body mass, which has the larger number of records, tends to have a consistent distribution in both species. In contrast, ear length and hindfoot length presented higher variability.

Discussion

This study assessed hantavirus seroprevalence across 49 species in the United States using data from the National Ecological Observatory Network (NEON). We described the distribution of hantavirus exposure across species, seasons, and geographical regions, contributing to the understanding of hantavirus circulation among stablished and newly identified host species. Our findings indicate that Peromyscus species generally exhibit higher seroprevalence levels, in agreement with scientific literature identifying P. maniculatus as the primary host of Sin Nombre virus (SNV) (Childs et al. 1994). Nevertheless, other rodent species also play significant roles in supporting hantavirus circulation, particularly in regions where P. maniculatus and P. leucopus show lower seroprevalence and multiple hantaviruses coexist (Rowe et al. 1995; Ermonval, Baychelier, and Tordo 2016).

Since hantaviruses were described three decades ago, new hosts and hantaviruses continue to be identified, aided by enhanced screening efforts and advancements in molecular tools (Perry et al. 1997; Milholland et al. 2018; Bellomo et al. 2021). Globally, researchers have identified at least 140 small mammal species hosting nearly 30 different recognized hantaviruses (ICTV 2024), with at least 42 species hosting hantaviruses with zoonotic potential (Guo et al. 2013; Guzmán, Mattar, and Calderón 2015; Forbes, Sironen, and Plyusnin 2018; Milholland et al. 2018; Chen et al. 2023; ICTV 2024). Although hantaviruses usually circulate within a single or a few rodent species, these viruses can circulate among host assemblages, and multiple hantaviruses can infect a single host species (Guzmán, Mattar, and Calderón 2015; Milholland et al. 2018).

The serological test employed by NEON does not distinguish hantavirus lineages (Klingström et al. 2002; Thibault 2022). Even though our study identified seropositive samples in species previously recognized as SVN hosts (Figure 2) (Ermonval, Baychelier, and Tordo 2016; Milholland et al. 2018; Chen et al. 2023), from these findings, we cannot rule out that the immune response detected in blood samples reflects an exposure to other hantaviruses different to SNV or coinfections with different hantaviruses. For instance, P. truei has been described as a host for both SNV and Prospect Hill virus (Orthohantavirus prospectense) (Rollin et al. 1995; Neill et al. 1996; Milholland et al. 2018; ICTV 2024), and S. hispidus has been found infected by SNV and Black Creek Canal viruses (O. nigrorivense) (Lewis 2005; Milholland et al. 2018). Since not all hantaviruses are recognized to be pathogenic to humans (Mills 1998; MacNeil, Nichol, and Spiropoulou 2011; Springer et al. 2016), understanding the cocirculation of pathogenic and nonpathogenic hantaviruses is critical to assess human health risks (Luong et al. 2011; Chen et al. 2023).

In our study, we identified six seropositive species that, to our knowledge, represent newly described hosts for hantavirus, including P. keeni P. gossypinus, P. polionotus, Myodes gapperi, Podomys floridanus, and Napaeozapus insignis. Nevertheless, the taxonomic identification of these potential novel hosts might have been misclassified, or they may have been recorded under invalid or synonymous scientific names in earlier reports. For instance, while we found no prior reports of hantavirus infection in P. keeni, the Integrated Taxonomic Information System (ITIS; www.itis.gov) recognizes 16 synonyms for this species and also accepts it as a subspecies of P. maniculatus. Consequently, past records of hantavirus infection in P. keeni might have been documented under different species name, including P. maniculatus. Similarly, we found no report of hantavirus in Myodes gapperi, a species recently prioritized for hantavirus surveillance (Mull et al. 2022) . Nonetheless, SNV was reported in Clethrionomys gapperi (Kuenzi et al. 2001), an accepted synonym of Myodes gapperi (Kryštufek et al. 2020). We identified one seropositive sample of Myodes gapperi in Massachusetts (HARV site), which is distant from the previous report of Clethrionomys gapperi in Montana (Kuenzi et al. 2001). Thus, if the previous report truly corresponds to Myodes gapperi, the specimen identified in this study would represent the first record of the species in northeastern United States.

The seropositive samples found in Myodes gapperi were collected in sites where seropositive samples were also collected from P. maniculatus or P. leucopus rodents. Similar to previous findings of Kuenzi et al. (2001), the seroprevalence in Myodes gapperi was low, with results also based on an antibody test that does not distinguish among hantavirus lineages. Therefore, the epidemiological role of Myodes gapperi may not be crucial, as there are other well-known reservoirs in the area, and there is no evidence that it may carry another hantavirus for which it may have more relevance.

On the other hand, P. gossypinus, another plausible novel host for hantavirus, can be misclassified in the field as P. maniculatus (Perry et al. 1997). Seropositive samples of P. gossypinus, however, were collected in all the sites from Alabama and southern Georgia in which the species was captured. These sites fall outside the accepted distributional range of P. maniculatus by the International Union for Conservation of Nature and Natural Resources (IUCN (International Union for Conservation of Nature) 2024b). Besides P. gossypinus, two other new hosts were found in these southern areas, including P. polionotus and Podomys floridanus. In the southern US, the circulation of hantavirus could be related to Black Creek Canal or Bayou hantaviruses, both known to cause human pulmonary syndrome in humans (Morzunov et al. 1995) and associated with Sigmodon hispidus and Oryzomys palustris, respectively (Chen et al. 2023; Holsomback et al. 2009; McIntyre et al. 2005). Considering that these areas are beyond P. maniculatus and P. leucopus distributions, and the local presence of at least two zoonotic hantaviruses, the description of these new hosts are a valuable contribution in the assessment of human health risk.

Detecting hantavirus exposure in rare species or in those with low prevalence requires large sampling size and extended field efforts. The comprehensive NEON hantavirus program explored a wide range of individuals, species, geographies, seasons, and ecosystems, which partly explains the detection of hantavirus exposure in new areas and in multiple and novel host species, revealing hidden aspects of hantavirus dynamics. The identification of rare events would probably remain undetected with lower sample size and in studies limited in time and space. For instance, we found only a few previous reports for species such as Reithrodontomys megalotis (Abbott, Ksiazek, & Mills, 1999; Bi, Formenty, & Roth, 2008), and only one report for Onychomys leucogaster (Banther-McConnell et al. 2024). Studies with modest sampling effort may lead to over- or underestimation of hantavirus prevalence in wildlife (Carver et al. 2010). For example, we found peaks up to 100% seroprevalence in periods with low sample sizes (e.g., November 2018 in P. maniculatus) and high seroprevalence in some species with low sample sizes such as P. truei (Figure 2; Appendix S1: Table S1).

Despite the large number of species and blood samples, and the great geographical extension of the NEON program, we did not find seropositive samples in 34 of the species sampled, some of them previously described as hantavirus hosts. This absence of seropositive samples underscore the relevance of large-scale, continuous surveillance programs, particularly for pathogens like hantaviruses, which typically exhibits low prevalence and can infect rare species (Milazzo et al. 2012; Milholland et al. 2019; Mull et al. 2022). For instance, Mull (2023) utilized taxonomical and ecological characteristics to identify over 100 species in the Americas that should be targeted as potential hantavirus hosts. Furthermore, Milholland et al. (2018) found that out of nearly 37 endemic Peromyscus species in Mexico, 28 remain untested for hantaviruses, indicating a significant taxonomic gap in host identification. On the other hand, NEON program, like other studies, relies on serological testing that does not differentiate between hantaviruses lineages or assess their pathogenic potential (Bellomo et al. 2021). To deepen our understanding of hantavirus–host dynamics, it is crucial to identify hosts, explore their distributions and interactions, and assess the zoonotic potential of the hantaviruses they carry (Milholland et al. 2018).

Spatial distribution

The primary recognized reservoir of the SNV, P. maniculatus, was not captured at 15 sites, 14 of which were located beyond its known distribution range, reflecting its consistent abundance within its stablished range (Dragoo et al. 2006). There were no seropositive samples at nearly half of the sites where P. maniculatus was sampled (13 of 30) (Figure 1). Most of these sites were near other NEON sites that presented seropositive P. maniculatus, indicating exposure to hantavirus in surrounding areas, as represented at the state-level (Figure 1). For instance, although no seropositive samples were collected at NIWO in Colorado, which it is approximately 35 km from RMNP site where P. maniculatus seropositive samples were collected. This suggests that fine-scale ecological factors may be influencing hantavirus circulation, which should be explored as a means to better identify drivers of hantavirus transmission risk. Interestingly, some of the sampled sites without seropositive P. maniculatus were located at the edges of the species’ distributional range (IUCN (International Union for Conservation of Nature) 2024b), which may represent areas with lower population densities or with a different rodent species richness structure. Seronegative P. maniculatus sites grouped in two clusters, one in southwestern United States (New Mexico and Arizona) and another in the eastern regions (Maryland and Virginia). Notably, the eastern sites, especially those in Virginia, showed high seroprevalence in P. leucopus (Figure 1).

Among the 21 sites where P. leucopus rodents were sampled, in nine of them no seropositive samples were collected. Similar to the trend observed for P. maniculatus, these seronegative sites were typically located at the edge of the species’ distributional range, such as SRER (Arizona), DELA and LENO (Alabama), BART (New Hampshire), and UNDE (Michigan). The only exception was OAES in Oklahoma, potentially explained by the limited number of blood samples collected (21 blood samples) (Appendix S1: Table S1). Thus, hantavirus exposure in P. maniculatus and P. leucopus appears widespread but not continuous across their range. On the other hand, the highest seroprevalence for P. leucopus was found in Virginia and Maryland, specifically where no seropositive P. maniculatus were collected. In fact, at two of these sites (BLAN and SERC) P. leucopus was the only seropositive species. Therefore, in the eastern areas, at the borders of P. maniculatus distribution, P. leucopus may play a crucial role in sustaining hantavirus circulation.

Our study identified ten sites where seropositive samples did not correspond to the well documented hantavirus hosts P. leucopus or P. maniculatus, despite being sampled in four and in six of those sites, correspondingly. Interestingly, in six of these sites, the seropositive samples were obtained from species identified in this study as potential novel host species. For instance, Podomys floridanus was the sole seropositive species in Florida (OSBS), and along with P. gossypinus, the only seropositive species in Georgia (JERC). Additionally, a blood sample obtained from Napaeozapus insignis was the only seropositive sample at STEI, in Wisconsin. In Washington, the two available NEON sites (ABBY and WREF) yielded three seropositive samples all from P. keeni, another potential novel host for hantaviruses. These two sites from Washington, however, are located outside the distribution range described for P. keeni (IUCN (International Union for Conservation of Nature) 2024a), although the sites tend to overlap with occurrences documented in the Global Biodiversity Information Facility (GBIF 2024). The seropositive records collected from P. keeni would represent the first documentation of hantavirus in this species but also provide additional evidence of the species presence beyond the recognized boundaries currently stablished by the IUCN.

In the remaining five sites with seropositive samples unrelated to P. maniculatus or P. leucopus, the rodents involved have been previously described as hantavirus hosts. For example, in Colorado (STER), seropositive samples were collected from Onychomys leucogaster, a species recently identified as hantavirus host with two positive cases reported in New Mexico (Banther-McConnell et al. 2024). At NOGP site in North Dakota, the seropositive samples were collected exclusively from Microtus pennsylvanicus, previously associated with Prospect Hill virus (O. prospectense), with pathogenic potential for humans (Lee et al. 1982). Although blood samples from M. pennsylvanicus were collected across seven states, seropositive samples were only obtained in Virginia and North Dakota. The low sample size in the other states (three blood samples in five states) limits our ability to rule out hantavirus exposure in M. pennsylvanicus across its extensive distribution in the United States (IUCN (International Union for Conservation of Nature) 2024b). Nevertheless, our findings suggest that M. pennsylvanicus plays a role in areas where seropositive samples were collected, possible supporting the cocirculation of Prospect Hill virus and SNV.

Finally, seropositive samples were collected from Reithrodontomys megalotis at CPER, Colorado, where no seropositive samples were identified for P. maniculatus or P. leucopus. Reports of hantavirus infection in R. megalotis are rare in the literature, and have been associated with SNV, El Moro Canyon, and Huitzilac hantaviruses, with unknown pathogenic potential for humans (OPS 1999; Lewis 2005; Milholland et al. 2018). In summary, R. megalotis may have a role in hantavirus circulation particularly in central and southwestern areas of the United States, where hantavirus exposure has been described in humans (New Mexico, California, Colorado) (Hjelle et al. 1994; Lewis 2005; Milholland et al. 2019), and where Peromyscus species may play a minor epidemiological role.

Temporal trends and seasonality

Agreeing with previous studies, our analyses revealed variable rodent abundance and hantavirus seroprevalence across different seasons and years (Semmens et al. 2001; Carver et al. 2010; Luis et al. 2010; Bagamian, Douglass, et al. 2012) . We observed interannual variations in captures and seroprevalence (Appendix S1: Figure S2, Figure 3). Such variations of rodent captures have been linked to environmental events that cause drastic landscape changes, such as vegetation flowering, forest fires, climate anomalies (e.g., ENSO), among others, especially at higher latitudes (Olsson et al. 2002; Mölle et al. 2022). Therefore, despite the high number of captures and blood samples for species such as P. maniculatus, the six-years period of hantavirus serological data available in NEON does not allow robust time series analyses capturing signals of long-term processes (Pankratz 1983; Hyndman and Athanasopoulos 2018). The patterns observed in our analysis, however, allowed us to identify correlations and seasonal signals of hantavirus dynamics, useful to guide epidemiological surveillance.

Our study identified a consistent intra-annual trend of increased rodent captures during late spring and summer (May to August) across all data-groups (All rodents, P. maniculatus, P. leucopus), followed by a sharp decline in winter (November to March) (Figure 4). This pattern partially aligns with the dynamics of Peromyscus species described in North America (Karunarathna, Wells, and Clark 2024). Unlike previous studies, we did not observe a robust peak in abundance during the fall (October-November) (Sullivan and Sullivan 2004; Wilder and Meikle 2006). Nevertheless, the compiled monthly values present a high variability that hampers the identification of clear trends in some months (Appendix S1: Figure S4). This variability may be attributed to the broad geographical coverage of the NEON program, spanning from Puerto Rico to Alaska, which may introduce greater variability between populations at different latitudes and longitudes. In contrast, studies conducted in more localized areas are subject to more uniform environmental conditions and lower seasonal variability (Karunarathna, Wells, and Clark 2024).

In NEON database, the sharp decline of the number of rodent captures during fall and winter primarily reflects lower fieldwork efforts. Even though the NEON protocol aims to have consistent capture efforts, it is assumed that it will be reduced during periods with challenging field conditions, considering personnel safety, logistic constraints, and the impacts of field captures for animal welfare (Paull et al. 2023). Captures in December occurred only in states with temperate weather or mild winter conditions, such as California, Puerto Rico, and Florida, while those in January and February were exclusively collected from California. Consequently, blood sampling also decreased, resulting in only 43 blood samples in January, all from 2018, with just one seropositive sample (Figure 4). Therefore, the information on hantavirus dynamics during winter remains limited, hampering our understanding of seasonal patterns (Bagamian, Douglass, et al. 2012).

The temporal patterns of seroprevalence exhibited inconsistent intra-and inter-annual trends, with a general bimodal pattern and some extreme peaks (Figure 4; Appendix S1: Figure S1, Figure S2). Seroprevalence is greatly influenced by samples sizes, often resulting in extreme values with lower robustness. For instance, in November 2018 and in late 2015, fewer than 10 P. maniculatus were captured and sampled, potentially mispresenting infection expressed as 100% seroprevalence in 2018 and 0% in 2015.

While no temporal association was found between captures and seroprevalence when considering all rodents collectively, such association was observed in P. leucopus and in P. maniculatus, where abundance (captures) predicted seroprevalence. The association between population abundance on hantavirus transmission is inconsistent in the literature, with studies reporting no association (Pearce-Duvet et al. 2006; Clay et al. 2009), positive association (Boone et al. 1998; Douglass et al. 2007), or negative association (Calisher et al. 1999; Semmens et al. 2001; Douglas et al. 2021). These associations should consider the diagnostic tests used and whether the test indicates current or previous infection (Lewis, 2005; Banther-McConnell et al., 2024). NEON’s hantavirus diagnosis relies on a serological test (ELISA IgG), which does not necessarily indicates hantavirus viremia (Klingström et al. 2002; Thibault 2022) and therefore does not inform about the time of infection, complicating temporal-trends interpretations. Molecular virus detection may also fail to identify the time of transmission, as RNA can persist in blood for 4–6 months post-infection (Bagamian et al. 2013) and vary depending of the organ analyzed (Botten et al. 2002; Quizon et al. 2022). Future research should explore the establishment and duration of hantavirus antibodies in wild rodents to improve inferences about the time of infection.

The negative correlation delay observed in P. leucopus (4–5 months), and the positive correlation delay in P. leucopus (9–11 months) and in P. maniculatus (2-months) suggests that, even though with different effects and at different lags, rodent abundance influences the transmission and immune response development. In P. leucopus, the capture peak observed from July to August (breeding season) predicts a rise in seroprevalence in the following spring (9–11 months lag), and a decreased seroprevalence observed from November to January (4–5 months lag). On the other hand, in P. maniculatus, the summer peak of captures predicted an increased seroprevalence in the following fall (September-November 2-months lag). Our findings align with previous studies that demonstrate an association between abundance and transmission (Pearce-Duvet et al. 2006; Bagamian, Douglass, et al. 2012). For instance, Luis et al. (2010, 2015) described a positive temporal lag of 8–16 months between the rise in rodent abundance and the subsequent rise in hantavirus seroprevalence.

The differing temporal lags between abundance and prevalence observed in P. leucopus compared with P. maniculatus may reflect distinct dynamics of hantavirus transmission. In general, both species of Peromyscus tend to present similar population dynamics across time (abundance, breeding season), highly dependent on resources availability, habitats, latitudes, vegetation structure, climate, among others (Wilder and Meikle 2006; Wang et al. 2009; Sullivan and Sullivan 2022; Wolff 1985). In both species, northern latitudes and higher altitudes may influence the range of temporal variability, while winter conditions decrease population survival (Wang et al. 2009). Peromyscus maniculatus, with larger geographical extension compared with P. leucopus, are exposed to larger ranges of latitudes, and therefore would tend to present higher dispersion of seasonal captures and seroprevalence values (Figure 4; Appendix S1: Figure S4) (Wang et al. 2009). This variability, together with potential interspecies differences in the establishment of a detectable immune response and persistence of antibodies in circulation (Schönrich et al. 2008), may explain in part the different time-lag associations of captures and seroprevalence between P. maniculatus and P. leucopus.

Sex, Life stages and Body measurements

Our findings align with previous studies, confirming that adult rodents exhibit higher hantavirus seroprevalence (Douglass et al. 2007; Luis et al. 2015). Notably, we detected only one seropositive sample in a juvenile P. truei. The persistence of antibodies, which can remain in circulation for weeks, months or permanently through a rodents’ life (Tischler et al. 2005; Schönrich et al. 2008; Engdahl and Crowe 2020), likely contributes to the higher seroprevalence in adults. This may reflect the cumulative persistence of antibodies rather than a lack of exposure during early life stages. Additionally, the juvenile stage represents a brief period of the rodents’ life, potentially insufficient for developing a detectable immune response (Schönrich et al. 2008; Luis et al. 2010).

There is limited evidence regarding the effects of maternal hantavirus antibodies. An experimental study on Puumala hantavirus in Myodes glareoulus suggested that maternal antibody transfer could delay the timing of hantavirus infection (Kallio et al. 2006). Consequently, the low hantavirus infection rates in juveniles could be attributed to early protection conferred by undetected maternal antibodies.

In terms of sex differences, our study found higher seroprevalence in males compared to females (Bennett et al. 1999; Bagamian et al. 2013), and a correlation between the presence of scrotal testes and seroprevalence (Figure 5; Appendix S1: Table S3). Sexually active rodents are typically more aggressive and exhibit greater spatial dispersal (Bagamian et al. 2013; Maroli et al. 2015), facilitating hantavirus transmission through increase direct contact and overlap between infected and uninfected individuals (Pearce-Duvet et al. 2006; Bagamian, Douglass, et al. 2012).

Conversely, pregnancy was negatively correlated with seroprevalence, indicating that non-pregnant females tended to have higher values of seroprevalence. Although pregnancy involves close contact with multiple males (Cornejo-Latorre, Cortés-Calva, and Álvarez-Castañeda 2021), which would typically suggest higher seroprevalence, our results may be explained by the short gestation period in Peromyscus species, lasting 24–27 days (Wilson & Ruff, 1999), probably insufficient for developing a detectable immune response post-infection during mating. Additionally, NEON’s sampling protocol excludes blood collection from females in advance pregnancy (Paull et al. 2023), limiting our ability to assess the impact of pregnancy on hantavirus infection.

In both P. leucopus and P. maniculatus, seropositivity was positively associated with body mass, though not with hind foot length or total body length (Figure 5; Appendix S1: Table S5). Given that most captured rodents were adults, body mass likely reflects body condition and, to a lesser extent, age-related changes. Thus, we infer that heavier, more robust, sexually active and potentially more aggressive rodents are more susceptible to hantavirus exposure and transmission (Bagamian, Towner, et al. 2012). Consistent with previous studies, these findings also suggest that SNV infection has minimal impact on the fitness of Peromyscus species (Mills et al. 1998; Abbott, Ksiazek, and Mills 1999; Quizon et al. 2022).

In addition to body mass, ear length also differed between seropositive and seronegative individuals in P. maniculatus. The reliability of this difference, however, remains uncertain due to the smaller sample size and the high variability in measurements (Figure 5). This low precision may be attributed to the challenges in field measurements given the small size of the ears and the influence of individual researcher skills.

Future directions

The NEON program delivers one of the largest datasets of standardized hantavirus surveillance in small mammals in the United States and the world, providing a unique opportunity to study hantavirus dynamics in nature. The discontinuation of the NEON hantavirus program in 2019 limits its contribution for the understanding of long-term processes and phenomena that occur gradually, including the effects of climate change and inter-annual transmission dynamics within the hantavirus system. Fortunately, NEON maintains a sample biorepository, providing a source of rodent blood for further testing to identify hantavirus infections. In addition, NEON databases provide valuable data of the ecology and dynamics of several pathogen hosts.

The NEON hantavirus program, as well as most of the scientific studies reviewed, lack of a robust number of samples during fall and winter seasons, generating a significant gap for the understanding of temporal patterns of infection and transmission. We recommend the implementation of hantaviruses surveillance studies that aim to increase sample collection during fall and winter seasons and compensate the systematic lack of evidence for such periods. These programs should consider animal welfare and personal availability.

Finally, the identification of specific hantavirus circulating would give crucial information about the risks for human health and the epidemiological role of each rodent host. Not all hantaviruses have known zoonotic potential and new hantaviruses are periodically described. The implementation of additional analytical tests and specimen collection (e.g., organs) to characterize hantaviruses would enable deeper studies regarding hantavirus ecology and evolution.

Conclusions

Hantaviruses have a complex dynamic that involves multiple hosts and pathogenic and not pathogenic viral strains. We found that multiple hosts that were exposed to hantavirus and identified new hosts and new areas of hantavirus circulation, highlighting the value of extended, permanent and broad surveillance programs. Rodent captures presented an inconsistent pattern across years, with a typical peak in summer. Seroprevalence exhibited a bimodal trend, although biased by the lower sample size during winter months. Captures were associated with seroprevalence with distinct delays in P. leucopus and in P. maniculatus. Seroprevalence and captures presented great variability across species, time and geographical locations, revealing the potential relevance of fine-scale features and the long-term gradual processes involved. This article advances our understanding hantaviruses circulation in wildlife across the United States and expands the current list of plausible hantavirus hosts.

Data Availability Statement

Original data were obtained from NEON (National Ecological Observatory Network (NEON) 2023a; 2023b).