A Systematic Review of the Distribution and Prevalence of Viruses Detected in the Peromyscus maniculatus Species Complex (Rodentia: Cricetidae)

Abstract

The North American Deermouse, Peromyscus maniculatus, is one of the most widespread and abundant mammals on the continent. It is of public health interest as a known host of several viruses that are transmissible to humans and can cause illness, including the acute respiratory disease Hantavirus Pulmonary Syndrome (HPS). However, recent taxonomic studies indicate that P. maniculatus is a complex of multiple species, raising questions about how to identify and interpret three decades of hantavirus monitoring data. We conducted a systematic review investigating the prevalence and spatial distribution of viral taxa detected in wild populations allocated to P. maniculatus. From the 46 relevant studies published from 2000 to 2022, we extracted and analyzed spatial occurrence data to calculate weighted populational prevalences for hantaviruses. We found that detection efforts have been concentrated in the Western United States and Mexico with a focus on the spread of Sin Nombre virus, the primary causative agent of HPS. There are significant gaps in the existing literature both geographically and in regard to the types of hantaviruses being sampled. These results are significantly impacted by a recent taxonomic split of P. maniculatus into four species, which results in the relabeling of 92% of hantavirus observations. Considering the uncertain, and likely multiple, phylogenetic histories of these viral hosts should be a key emphasis of future modeling efforts.

1. Introduction

The North American Deermouse, Peromyscus maniculatus (J. A. Wagner, 1845) is one of the most abundant and widespread mammals native to the continent (1). The genus Peromyscus sec. Mammal Diversity Database v1.12.1 (2–4) includes 82 recognized living species (3) and has been the subject of extensive research in ecology, development, genetics, evolution, and epidemiology (1,5–10). P. maniculatus is of particular interest to the public health sector and has been traditionally well-studied because it is known to be a source of several diseases that are communicable to humans (9,11). Research has been particularly focused on the deermouse’s role as the primary host of Sin Nombre virus (SNV), an RNA virus in the genus Orthohantavirus which was first identified in 1993 and was recently renamed Orthohantavirus sinnombreense by the International Committee on the Taxonomy of Viruses (12–15). SNV and other New World hantaviruses are the causative agents of Hantavirus Pulmonary Syndrome (HPS), an acute respiratory disease with a mortality rate of 60% at the time of the first outbreaks (8). From 1993 to 2021, a total of 850 cases of HPS have been reported in the United States (16), with most cases occurring in the Southwest (Figure 1). Despite the relative rarity of HPS cases compared to other hantavirus illnesses, the sudden emergence of the disease and potential for mutation has spurred intensive efforts to characterize the causative viral transmission pathways (17–20). Modeling the risks of hantavirus disease and novel zoonoses remains an urgent research objective that depends on comprehensive, up-to-date information about pathogen prevalence and the ecological factors influencing disease spread (21,22).

Figure 1.

Map of hantavirus pulmonary syndrome (HPS; also called hantavirus cardiopulmonary syndrome disease) human cases reported from 1993 to 2021 in the U.S. and to 2020 in Canada (16,53).

This paper presents a systematic review of the existing literature on virus detection studies of wild P. maniculatus in order to summarize findings from the last two decades and identify potential areas of further research. We focus on a spatial analysis of sampling effort and prevalence in studies that tested for hantaviruses and arenaviruses. Our results thereby add to the evidence base for ongoing research into the co-evolution of North American rodents with viral pathogens.

In addition, we discuss how study results are relevant for the reservoir host status of P. maniculatus considering recently proposed taxonomic revisions. Multiple genetic studies indicate Peromyscus maniculatus is a species complex (23). It was recently proposed to split P. maniculatus into between four and six species-level lineages (2,24,25), though this remains controversial (9). According to the Mammal Diversity Database v1.12.1 (2–4), the P. maniculatus species complex is composed of four species that are distributed as follows: (i) throughout Northern and Central Mexico, referred to P. labecula D. G. Elliot, 1903; (ii) in southern California and Baja Calaifornia, referred to P. gambelii (S. F. Baird, 1858); (iii) across the continental United States west of the Mississippi River and into northern Canada, referred to P. sonoriensis (Le Conte, 1853); and (iv) east of the Mississippi River until the Atlantic Ocean and north until the Hudson Bay, referred to P. maniculatus sensu stricto (J. A. Wagner, 1845) (Figure 2). The genetic distinctiveness of P. gambelii and P. sonoriensis from each other and from P. keeni in the Pacific Northwest (9) supports the existence of multiple species within what is typically referred to as “P. maniculatus” in ecological and biomedical studies. The latter identifications are referrable to the continentally distributed P. maniculatus sensu lato, and thus require re-interpretation relative to alternative taxonomic hypotheses.

Figure 2.

Proposed taxonomic split of the Peromyscus maniculatus species complex (24,25) into four new species as recognized by the Mammal Diversity Database (3). Note that the name P. maniculatus sensu stricto applies east of the Mississippi River in the U.S. and P. sonoriensis applies to most of the Southwest where the initial outbreak of HPS was first detected.

While hantavirus prevalence is the primary focus of biomedical research on wild-caught P. maniculatus, deer mice have been known to harbor or be susceptible to other zoonotic viruses, including SARS-CoV-2, flaviviruses associated with Tick-borne encephalitis, and arenaviruses which have been connected with human fatalities (26–28). Members of this species complex can be found in every terrestrial ecosystem across the continent, including peridomestic areas resulting in frequent contact with humans. Numerous field studies have been conducted on rodents in North America in order to determine virus prevalence and distribution, especially in regions where outbreaks have previously occurred. Significant advances have been made through these field studies as novel viruses have been detected and genetic sequencing has allowed researchers to build up-to-date virus and host phylogenies, but the approach to zoonotic virus monitoring in rodents has not been consistent or standardized.

2. Materials and Methods

We conducted a systematic review of existing literature reporting observations of zoonotic viruses in Peromyscus maniculatus. The focal population was P. maniculatus and the outcomes were positive or negative test results using a range of detection methods. Examples of this include antibody tests, Polymerase Chain Reaction (PCR) tests, and genetic sequencing. Three databases, Scopus, PubMed, and PubMed Central, were used to find existing scientific literature on this topic. Search queries were conducted in August 2022 with the common and scientific names of P. maniculatus as well as nomenclatural synonyms and the keywords “virus”, “viral”, and “viruses” (Table 1). Through these parameters, 448 papers were identified as potential candidates for the systematic review. We also added 3 ad hoc papers from other sources.

Table 1.

Search terms used in conducting the systematic review of existing literature across three separate databases. Common names as well as scientific names for species within the P. maniculatus species complex were included.

Database	Search Query Used	Search Settings Used
PubMed	(virus OR viral OR viruses) AND (“Peromyscus maniculatus” OR “Peromyscus gracilis” OR “Peromyscus bairdi” OR “Peromyscus abietorium” OR “Peromyscus nubiterrae” OR “Peromyscus argentatus” OR “Peromyscus eremus” OR “Peromyscus anticostiensis” OR “Peromyscus plumbeus” OR “Peromyscus gambelii” OR “Peromyscus labecula” OR “Peromyscus arcticus” OR “Peromyscus sonoriensis” OR “Eastern Deermouse” OR “North American Deermouse”)	Title/Abstract
PubMed Central		Abstract OR Body-All Words OR Title
Scopus	“Peromyscus maniculatus” OR “Peromyscus gracilis” OR “Peromyscus bairdi” OR “Peromyscus abietorium” OR “Peromyscus nubiterrae” OR “Peromyscus argentatus” OR “Peromyscus eremus” OR “Peromyscus anticostiensis” OR “Peromyscus plumbeus” OR “Peromyscus gambelii” OR “Peromyscus labecula” OR “Peromyscus arcticus” OR “Peromyscus sonoriensis” OR “Eastern Deermouse” OR “North American Deermouse” AND virus OR viral OR viruses	TITLE-ABSTRACT-KEY

We used the CADIMA web tool to review articles for inclusion (29). We uploaded lists of identified papers from each source to CADIMA, after which we removed duplicates and 253 papers remained. Each paper was manually reviewed by AF for inclusion, and a second person (BS) was consulted on unclear cases. Inclusion criteria were that the article must report new primary data about the results of testing wild P. maniculatus for virus occurrence or infection. For reasons of scope, we excluded lab studies based on artificial infection experiments as well as papers published prior to 2000. We excluded 192 for one or more of the following reasons: not having a full text available in English through interlibrary loan services at the author’s home institution, not meeting the inclusion criteria, or being published before the cutoff period. This left 46 papers for data extraction from the systematic review process (Figure 3).

Figure 3.

Diagram summarizing the steps of the systematic review done in CADIMA that led to the final set of 46 which were included.

Data extraction entailed reading each scientific paper in full and extracting relevant information on sampling time, sampling location, number of hosts, various host identifiers, detection method used, material sampled, and the results of observations into a spreadsheet (Table S3). Information on all species captured and sampled was recorded for future use, though this study focuses solely on detection presented in P. maniculatus (Table S2). Summary results for P. maniculatus are available in Table S4.

To visually represent the patterns discovered in the extracted data, we created a series of maps using the Tableau software program (30). Spatially coded results were taken from the data extraction spreadsheet and translated into the maps on either the county, province, or state level, depending on whether the sample was taken in the United States, Canada, or Mexico, respectively.

Sampling effort was represented by the number of observations, which are defined as the number of times a unique host was tested for a unique virus using a specific method. Hence, if an individual rodent was tested by ELISA for both Whitewater Arroyo virus (WWAV) and Amapari virus (AMAV), this was recorded as two separate observations. Similarly, there would be multiple observations reported if an individual rodent was tested for the same virus multiple times in a recapture study. This way of individuating observations therefore provides a more fine-grained basis for collecting and analyzing test results. In the case of counties that were the site of multiple studies, the observations across each study were summed to create a single value which was then reflected in the map.

In the rare instance where one county was sampled for multiple virus species, the higherorder classification of virus species was used to create the maps. For example, if one study reported sampling for SNV and a separate study conducted in the same county tested for hantavirus antibodies but not for a specific virus species within that family, the county was represented as a hantavirus county on the map. However, all results for each type of virus tested are available in Table S2.

Virus prevalence was determined as the percentage of antibody-positive P. maniculatus reported by each study. For sampling locales with multiple studies and thus multiple prevalence results to consider, the reported prevalence results were taken and weighted by sampling effort in order to avoid over-representing results from studies that tested a smaller number of rodents. The county estimates were calculated according to the following equation:

$$P_c=\Sigma w_s\times p_s$$

The weights w_s are the number of rodents tested for that pathogen in each study divided by the sum of all rodents tested. Thus, the weights relative to a county sum to 1. The weight for each study is multiplied by the seroprevalence proportion p_s from each study, calculated as the number of positive tests divided by the total (31). Weighting is important to account for the relative precision of different studies in estimating the mean prevalence of the pathogen in the host population.

To assess the richness and distribution of the viruses detected across the selected literature, we processed the data into cumulative measures of sampling effort by virus type and geographic location. Geographic regions were analyzed at the state level for the United States and the country level for any other sampling locales. To track how sampling effort has been distributed over time, the relationship between the number of observations reported and (a) the number of viruses identified and (b) the number of unique locations sampled were also examined.

3. Results

Studies sampling populations in the P. maniculatus species complex have been conducted across eleven different U.S. states, six states in Mexico, and one Canadian province since 2000. This paper analyzed 60,791 unique observations across 46 studies, and we were able to assign geographic locations to 60,549 (99.6%) at the county (U.S.) or province level (Canada and Mexico). This includes observations that we split evenly across counties when studies only reported aggregate results for two or more counties. While we focus only on observations related to the P. maniculatus complex, we note that the full extracted dataset in Table S3 includes observations from over 100 rodent species and that about a third of these had positive test results for an arenavirus, flavivirus or hantavirus. This highlights the taxonomic scope of rodent virus “dark data” locked in scientific publications (32)

Scopus returned the largest number of potential papers compared to other databases, regardless of the type of search query used in the literature review (Table S1). PubMed Central produced more potential papers than PubMed when the search queries were broad, though the latter returned nearly 5 times as many papers when the search query included virus-related terms as well as host species names. We found that for this particular species, inclusion of common names and junior synonyms in search queries had only marginal effect (<1%) on the total number of citations returned across all three databases. This result may be due to the relatively recent application of alternative names to populations in the P. maniculatus species complex (24,25). Studies of the influence of including synonyms on search results for other taxa have found larger effects (33,34).

The primary serological method used across all studies was an ELISA, though RT-PCR and IFAT, among other tools, were also utilized (Table S2). The primary antigen in the studies was SNV, which was the target of 37 studies. Whitewater Arroyo mammarenavirus (WWAV) and Amapari virus (AMAV) were detected in two studies; Powassan virus (POWV) and Monongahela virus (MGLV) were detected in one study. Five studies did not test for a specific virus species and instead sought to detect generic hantavirus or arenavirus antibodies. The seroprevalence ranged greatly across studies from 0% to 100% of individuals sampled in a given subpopulation. Sampling effort varied as well, with the smallest study reporting 9 unique observations and the largest reporting 11,391.

Hantaviruses in general were the most common virus presented in the various studies, with 91% of papers reporting on some virus within the Hantaviridae family (Table S2). The predominant virus in the recent literature is SNV, which was examined across the widest longitudinal range from California to Indiana (Figure 4). At least one study has been published in almost every year on this virus since 2000. Other viruses were sampled less frequently and at a lower volume, with a maximum number of observations of 353 for both WWAV and AMAV collected in a single study (Table S2). POWV and arenavirus were also tested for in only a single study each. In New Mexico, one study reported finding POWV in deer mice caught in the northern central region, but precise details on location are unavailable. In Texas, multiple studies focused on detecting arenaviruses in P. maniculatus, including WWAV and AMAV.

Figure 4.

Geographic distribution of sampling effort and zoonotic virus type tested for in P. maniculatus field studies. Legend refers to the number of observations in each county, province, or state, where observations are defined as the number of unique hosts tested for a unique virus. Hantavirus and arenavirus refer to studies in which a specific virus species was not identified.

Since 2000, sampling has been concentrated on the western half of the U.S. despite the fact that rodents in the P. maniculatus species complex can be found across the entirety of North America (Figure 2, Figure 4). Counties in Montana, Utah, and Colorado had the highest average sampling effort, with fifteen different counties hosting more than 1,500 unique observations each. This increased level of observations is partially due to multiple studies being carried out in the same county, as some locations were sites of more than one study while others were only sampled once within the last 13 years (Table S2). In studies that took place over multiple counties, all collection sites are represented. One study was conducted in Indiana and another in West Virginia, but the rest of the U.S. Midwest and East Coast were not sampled for P. maniculatus between 2000 and 2022.

Sampling effort has been more evenly distributed across geographic locations than across virus types. In terms of the number of studies conducted in each state, Montana had eleven while Nevada, Arizona, West Virginia, Oregon, and Indiana each boasted one (Table S2). However, the number of studies alone does not accurately represent the total sampling effort in a geographic area; both West Virginia and Oregon only hosted one study, but the former had a total of 15 observations while the latter had 3,175. Montana had the highest number of observations at 29,857, while Colorado and Utah had the second and third highest counts with 10,303 and 7,889, respectively.

As the number of observations reported increased between 2000 and 2022, so did the number of locations sampled. However, the total number of locations sampled across this time period was only 18 within 11 U.S. states as well as Mexico and Canada, which does not accurately cover the entire range of rodents within the P. maniculatus species complex.

The dominance of studies aimed at detecting SNV appears to correlate with historical data on the risk of HPS (Figure 5). There is a moderate positive relationship between the number of reported human HPS cases in a location and the sampling effort in that location (coefficient=0.056, standard error=0.007, no intercept; R²=0.52; p-value < 0.0001). This trend, however, is driven strongly by a handful of data points. Notably, Montana and Utah are overrepresented in terms of sampling effort compared to the number of historic hantavirus cases in those locations (studentized residuals of 4.7 and 3.7, respectively). On the other hand, Alberta, Arizona, and Washington are undersampled (studentized residuals of −2.2, −2.1, −1.8, respectively).

Figure 5:

Comparison of field studies published since 2000 in which testing for viruses in P. maniculatus was conducted against the total number of reported human hantavirus cases across states and Canadian provinces.

The highest average seroprevalences were found in New Mexico and California, with Rio Arriba, Cibola, and Inyo counties all reporting greater than 40% of individuals as positive (Table S2). The majority of counties hovered between seropositivity levels of 10% and 30% of sampled rodents (Figure 6). Several counties in southern California, which fall within the range of P. gamelii, found no rodents with virus antibodies among those sampled (Table S2, Figure 6).

Figure 6.

Seroprevalence for viruses studied in P. maniculatus by United States county, Canadian province, and Mexican state. In areas in which there was more than one detection study conducted, the average seroprevalence across studies was weighted by sampling effort.

The total number of unique virus species confirmed across all studies was five (Table S2), though additional viruses within the Hantaviridae and Arenaviridae families were detected but not identified and could conceivably be unique species. The relationship between the number of observations and the number of virus species identified in the literature is not linear. This is because the overwhelming majority of studies are focused on SNV antibodies and thus add to the volume of observations without introducing a novel virus.

Most of studies were conducted within the range of the proposed species P. sonoriensis, which covers the majority of the western U.S. including the aforementioned states of Montana, Utah, and Colorado (Figure 7). Sampling was also conducted within the ranges of three other putative species within the P. maniculatus species complex: P. gambelii, P. labecula, and P. maniculatus (sensu stricto). The arenavirus studies correspond with the range of P. labecula, while the SNV and hantavirus studies were concentrated within the range of P. sonoriensis.

Figure 7.

Seroprevalence compared to host ranges in the P. maniculatus species complex.

Based on the expert species range maps, we were able to assign 94% (57,232 out of 60,791) of all observations unambiguously to one of the four species (Table S4). We assigned an observation unambiguously if the boundaries of the corresponding county or province fell entirely within the range of one species. In some cases, the county or province boundaries fell partly or wholly outside the range of all four species. We counted these as unambiguously assigned only if the county or province boundaries overlapped with a single species. Of the 94% we could assign, P. sonoriensis received by far the most, 90%, compared to 6% for P. gambelii, 2% for P. maniculatus sensu stricto, and <1% for P. labecula.

4. Discussion

Zoonotic viruses in the species P. maniculatus have been non-systematically monitored over the last twenty years. In scientific literature published since 2000, we found reports of 60,791 unique observations of wild deermice interacting with five different virus species from three viral families. This body of research covers 11 unique U.S. states as well as regions in Mexico and Canada, but it does not reflect the entirety of the range of the P. maniculatus species complex, indicating that there are significant gaps in the literature within the 2000 – 2022 period.

Sampling effort varied significantly by both geographical location and virus genotype, with Sin Nombre virus (SNV) overrepresented in the literature relative to other hantaviruses, especially in the states of Montana and Colorado. With 80% of the literature reporting on SNV, the SNV prevalence in P. maniculatus over the last fifteen years is the best representation of hantaviral sharing dynamics in this group of rodents. Results on Powassan virus, Whitewater Arroyo Virus, Amapari virus, and Monongahela virus are informative in that they indicate the presence of these other hantaviruses within the the P. maniculatus complex. However, further conclusions cannot be drawn about these viral distributions without additional targeted sampling.

This systematic review suggests that, over time, even as more populations referrable to the P. maniculatus species complex are being sampled for zoonotic viruses, new viruses are not being discovered. While at first glance this seems to suggest that virus sampling for P. maniculatus is adequate, it more likely indicates that sampling effort has not been sufficiently dedicated to discovering other known or novel viruses (7). Instead, researchers have been using virus-specific antibody or PCR tests given their focus on understanding the public health risks posed by the distribution and prevalence of certain known viruses. Hantaviruses in general were sampled far more than arenaviruses or flaviviruses, which reflects public health concerns relative to HPS cases. However, both arenaviruses and flaviviruses are known to be capable of spreading to humans (35,36), suggesting that increased sampling effort dedicated to these viral families will provide valuable public health information.

Only 11 U.S. states were represented in the literature even though the P. maniculatus species complex is widespread across the continental U.S. aside from the Southeastern region where other Peromyscus species are found (e.g., P. polionotus, P. leucopus). While Indiana and West Virginia each received only one study, the rest of the U.S. Midwest and East Coast has remained unsampled for P. maniculatus-to-hantavirus interactions despite regional virus detection studies being conducted in other rodent species over the same time period (37–39). Hantavirus disease cases have been reported only sparingly in states east of the Mississippi since 1993 (Figure 1), so it is unsurprising that recent sampling efforts have not been dedicated to these regions where the risk of disease transmission to humans is sufficiently low (Figure 4). Interestingly, a handful of states where a significant number of hantavirus disease cases have historically been reported—Washington, Wyoming, North Dakota, and South Dakota—were not sampled for P. maniculatus occurrences in any of the studies, even though new hantavirus disease cases have occurred in each of these places within the last decade (16).

Among the geographic locations that were sampled (Figure 4), most populations fall largely within the subdivided range of P. sonoriensis and the dominant virus over this range was SNV (Figure 7). Surprisingly, only two sampling locations are now associated with P. maniculatus sensu stricto: Marion County, Indiana, and Randolph County, West Virginia. The latter was found positive for Monongahela virus, and the former was a positive for a generic hantivirus ELISA test.

Seroprevalence varied greatly across geographic regions, and the associated variation in sampling effort makes it difficult to draw strong conclusions about regional trends of virus prevalence. Several counties in Montana, Utah, and Colorado reported seropositivity of greater than 30%, while numerous studies conducted in counties in California, New Mexico, and Texas found zero positive individuals. Since sampling effort was highly inconsistent across all locations, the seropositivity results should be considered as reflections of not only the prevalence of hantavirus in the given region but also of the number of rodents sampled and the time period over which sampling took place. More consistent sampling efforts across the wider range of P. maniculatus would be necessary in order to fully understand the geographic trends in virus prevalence.

We also note that virus taxonomy is changing in parallel with the proposed revisions to P. maniculatus (40). The recent taxonomic union of Sin Nombre virus and New York virus, for example, would implicate both P. maniculatus and P. leucopus as reservoir hosts for different strains of a single virus species whose range spans from the Pacific to Atlantic coasts of North America. This viral taxonomic change is based on hierarchical genetic clustering, in part due to the high similarity of nucleoprotein and glycoprotein amino acid sequences of Sin Nombre and New York viruses. However, a human HPS patient infected with the New York variant showed no serologic reactivity to the Sin Nombre glycoprotein, potentially indicating different seroneutralization responses in humans (41).

Further topics for research include a deeper investigation into the evidence for particular Peromyscus species as reservoir hosts for hantavirus and other species. Scientific definitions of reservoir host status vary significantly and prioritize different types of biological relationships and evidence, ranging from simple detection in a host to persistent pathogen maintenance with or without serious symptoms or a history of co-evolution (42–48). Historically, hantaviruses were thought to closely co-evolve with single host species responsible for indefinitely maintaining the pathogen in the environment. Recent analyses have added nuance to this picture, showing that a number of hantavirus species infect multiple rodent hosts. More generally, biologists increasingly define reservoir hosts as composed of meta-populations or ecological assemblages of multiple species. Our results provide evidence that P. maniculatus is not the reservoir host for Sin Nombre virus in the narrow sense of being the sole biological species responsible for the pathogen’s maintenance. Instead, evidence suggests Sin Nombre has multiple reservoir species, especially in light of the taxonomic union of Sin Nombre and New York viruses. More broadly, the Peromyscus genus is likely not monophyletic (49–51), indicating the need for a broader survey of what is known about pathogens in other North American rodent species. In this respect, we lack an up-to-date, comprehensive analysis of the evidence for rodent reservoirs of hantaviruses that is consistent with leading frameworks for assessing future zoonotic disease risk (52).

5. Conclusion

Hantavirus pulmonary syndrome and other diseases contracted via the spread of zoonotic viruses pose a risk to human health in the United States and surrounding areas. Rodents within the P. maniculatus species complex have been identified as hosts for the viruses that cause HPS and other human diseases, but the recent scientific literature on virus prevalence in this group of species has many gaps, both in terms of the types of viruses being sampled and the geographic regions in which sampling is occurring. viral sampling has been uneven relative to the known HPS incidences. Taxonomic changes in the P. maniculatus species complex have large (and quantifiable) impacts on our knowledge of especially SNV prevalence and thus also HPS risk. It is important to establish a systematic framework with which to approach the sampling of wild hosts in order to optimize resources and identify regions and populations that pose a particular threat to human health.