Hantaan virus surveillance targeting small mammals at Dagmar North Training Area, Gyeonggi Province, Republic of Korea, 2001–2005



In response to a hemorrhagic fever with renal syndrome case in November 2000, a seasonal rodent-borne disease surveillance program was initiated at Dagmar North Training Area (DNTA), Gyeonggi Province, Republic of Korea. From April 2001-December 2005, 1,848 small mammals were captured. Apodemus agrarius accounted for 92.5%, followed by Mus musculus (3.6%), Crocidura lasiura (2.1%), and Microtus fortis (1.1%). Three species of rodents were found to be antibody-positive (Ab+) for Hantaan virus (HTNV): A. agrarius (22.3%), M. musculus (9.1%), and M. fortis (5.0%). Ab+ rates for A. agrarius increased with increasing weight (age), except for those weighing <10 g. The peak HTNV transmission period in Korea coincided with the peak reproductive potential of A. agrarius during the fall (August/September) surveys. HTNV strains from DNTA were distinct from HTNV strains from the People's Republic of China. From these studies, more accurate risk assessments can be developed to better protect personnel from rodent-borne diseases.


Rodent-borne hantaviruses, such as Hantaan virus (HTNV), are etiologic agents for hemorrhagic fever with renal syndrome (HFRS) in the Republic of Korea (ROK) that produce acute and chronic persistent infections in reservoir rodent hosts and are transmitted to humans through inhalation of virus-contaminated excreta (feces, urine, and saliva) or by bite of an infected rodent (Lee et al. 1981, Yanagihara et al. 1985, Huggins 1997, Calderon et al. 1999, Hinson et al 2004). HTNV is responsible for approximately 70% of all reported HFRS cases in the ROK and is widespread throughout much of Asia. It is of concern to military units training in field environments, and a serious public health threat to civilian populations due to its mean duration of illness from the onset of symptoms to complete recovery (approximately 5–6 weeks), overall morbidity, and a mortality rate of approximately 5% in the presence of good medical management (Huggins 1997).

Avoiding inhaling virus laden dusts is the best prevention against hantavirus infection, as there is no U.S. Food and Drug Administration-licensed vaccine. While the ROK has an approved vaccine (HantaVax®), it is only 75% effective after three doses (Cho and Howard 1999, Cho et al. 2002). Due to an often extensive incubation period of up to 50 days, epidemiological HFRS case investigations must focus on a wide range of field training and workplace sites and environmental conditions two months before the onset of symptoms. As this doesn't adequately describe the environmental and human factors leading to infection, it was recognized that site surveillance for areas of exposure was required to characterize rodent habitats, relative population densities, bionomics, HTNV antibody-positive (Ab+) rates, unique genomic geospatial variation in viral RNA, and local military training risk assessments to develop more effective HTNV risk-reduction strategies. This report focuses on HFRS epidemiological data that characterized rodent bionomics (e.g., reproductive behavior), HTNV Ab+ rates, and environmental conditions conducive for HTNV transmission for three areas trapped at Dagmar North Training Area (DNTA) that are useful in the development of HFRS risk reduction strategies for US and ROK military personnel during training.


Rodent habitat

Seasonal small mammal surveys were conducted from April, 2001 to December, 2005 at three adjacent areas at DNTA: Areas 1 and 2 (cantonment sites associated with rice farming) and Area 3 (primary training area adjacent to the Imjin River) that was separated from Areas 1 and 2 by a 5-m rock-faced earthen berm described by Kim et al. (2011) (Figure 1). Surveillance site selection was based on several factors: 1) previous reports of HFRS patients associated with exposure at DNTA, 2) habitats associated with relatively high rodent populations, and 3) sites that were adjacent to, but did not interfere with, military training activities. Survey sites were characterized by tall grasses with various amounts of herbaceous plants [e.g., giant ragweed (Ambrosia spp.), kudzu (Humulus japonica), and Artemisia spp.) adjacent to cantonment, command and control, and vehicle and troop maneuver areas and margins along forested hillsides, fighting positions, and wetland rice farming, which were similar to environmental conditions at Twin Bridges Training Area and along the perimeters of Firing Points 10 and 60 (O'Guinn et al. 2008, Sames et al. 2009a). Relative ground cover was much reduced during the winter and early spring as the proportion of grasses to other herbaceous plants (except kudzu) decreased.

Figure 1.

Satellite photo of DNTA and primary collection sites where rodent-borne disease surveillance was conducted from April, 2001 through December, 2005. The white line represents a 5 m high rock-faced (river side only) earthen berm with a 15 m base and fighting positions at the crest of the berm separating Areas 1 and 2 from Area 3.

Detachment or platoon-sized cantonment sites at Areas 1 and 2 were separated by low-lying young deciduous forested hills at their north and south boundaries and wetland rice farming on the west and from the company-sized training area (Area 3) by a 5-m rock-faced earthen berm (east boundary) (Kim et al. 2011). Areas 1 and 2 included internal earthen banks (0.3–0.5 m wide) with short cut grasses (<10 cm) separating rice paddies, the apex and base of the 5-m earthen berm separating Areas 1 and 2 from Area 3, and narrow (5–15-m wide) tall grass habitats that bordered the rice paddies, dirt roads, and forested hillsides. Area 3 is bounded by the Imjin River on its east, north, and south boundaries and bounded on the west by a 5-m rock-faced earthen berm with fighting positions at its apex that separates it from Areas 1 and 2. The central primary cantonment site was relatively barren with irregular patches of moderately tall (0.5–1.0 m) grasses and herbaceous plants (1–2 m) that provided limited cover for small mammals, while the perimeter of the central training site, troop maneuver areas, and infrequently used areas consisted of dense tall grasses and herbaceous plants provided a food source and ground cover for the small mammals during all seasons. Trapping was conducted adjacent to training area perimeters, intermittent streams, ponds, and other areas infrequently traveled by vehicles but used for troop maneuver activities.

Small mammal trapping

Seasonal trapping (March, June, August/September, and November/December) was conducted using collapsible live-capture Sherman® traps (7.7 × 9 × 23 cm; H.B. Sherman, Tallahassee, FL), that were set at 4–5 m intervals (20–50 traps/trap line) during daylight hours over one to two days (Kim et al. 2011). Traps positive for small mammals were collected the following morning, sequentially numbered, and transported to Korea University. The small mammals were anesthetized and then euthanized by cardiac puncture according to a Korea University Animal Use Protocol, identified, sexed, parity determined (2003–2005), weighed, and tissues (spleen, lung, and kidney) removed and stored at -70° C until used in this and other studies (O'Guinn et al. 2008, Payne et al. 2009). Standard procedures were followed for collection and transportation of specimens to minimize hazards from potentially infected rodents as described by Mills et al. (1995) and all personnel processing rodents in the laboratory were vaccinated using a ROK-approved hantavirus vaccine (Hantavax®).

Serology and RT-PCR test for hantavirus

Small mammal sera were diluted 1:16 in phosphate-buffered saline and examined for IgG antibodies against HTNV, Seoul (SEOV), and Prospect Hill (PHV) viruses by indirect immunofluorescent antibody (IFA) techniques (Lee et al. 1978, Song et al. 2007). RNA isolated from lung tissues of hantavirus Ab+ rodents were used in an RT-PCR assay that amplified a portion of the Gc glycoprotein-encoding M segment. Specifically, total RNA was extracted from lung tissues using the RNA-Bee Kit (TEL-TEST Inc., Friendswood, TX). RT-PCR amplification was accomplished by first reverse transcribing the RNA by random hexamer using the Superscript® II RNase H-reverse transcriptase kit (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Nested-PCR using the primers: outer primer set, 5′-TGGGCTGCAAGTGC-3′, 5′-ACATGCTGTACAGCCTGTGCC-3′; inner primer set, 5′-TGGGCTGCAAGTGCATCAGAG-3′, 5′-ATGGATTACAACCCCAGCTCG-3′; were then used to amplify a 373-nucleotide (nt) region of the hantavirus Gc glycoprotein-encoding M segment as previously published (Song et al. 2000, Xiao et al. 1992, 1994). Amplified products were size-fractionated by electrophoresis on 1.5% agarose gels containing ethidium bromide (0.5 mg/mL). PCR products of the correct size were cloned using the PST Blue-I vector (Novagen, Dormstadt, Germany), and plasmid DNA was purified using the QIAprepSpin Miniprep kit (QIAGEN Inc., Chatsworth, CA). DNA sequencing was performed in both directions from at least three clones of each PCR product, using the Big-Dye® Terminator v3.1 cycle sequencing kit (Applied Biosystems, Foster City, CA) on an automated sequencer (Model 3730, Applied Biosystems, Foster City, CA).

Genetic and phylogenetic analyses

Alignment and comparison of partial M segment sequences of HTNVs amplified from A. agrarius captured at DNTA, with previously published hantavirus sequences, was facilitated using the Clustal W method (Lasergene program version 5, DNASTAR Inc. Madison, WI). For phylogenetic analysis, the neighbor-joining (N-J) and maximum parsimony [Phylogenetic Analysis Using Parsimony (PAUP) version 4.0b] methods and the unweighted pair-group method of assortment were employed. Genetic distances were computed by PAUP, and topologies were evaluated by bootstrap analysis of 1,000 iterations (Swofford 2003).


Small mammal collections

A total of 7,295 trap nights (2,150 Area 1; 2,445 Area 2; and 2,700 Area 3) was conducted seasonally from April 2001-December, 2005. A total of 1,848 small mammals, belonging to eight rodent (seven genera) and one soricomorph species, was collected at Areas 1 (293), 2 (379), and 3 (1,176), with an overall trap rate for all trapping periods of 25.3%. Apodemus agrarius, the striped field mouse and primary reservoir for HTNV, accounted for 92.5% (1,710) of all the small mammals captured, followed by Mus musculus (house mouse, 3.6%, 66), Crocidura lasiura (Ussuri white-toothed shrew, 2.1%, 38), Microtus fortis (reed vole, 1.1%, 20), and Micromys minutus (Korean harvest mouse, 0.5%, 10). Only one each of Rattus norvegicus (Norway rat), Rattus rattus (roof rat), Tscherskia (=Cricetulus) triton (greater long-tailed hamster), and Myodes (=Eothenomys) regulus (Royal vole) were collected. While trap rates for A. agrarius were seasonally variable, overall mean trap rates were similar for Areas 1 (12.8%) and 2 (14.4%) but were >two to three-fold higher at Area 3 (40.9%) (F=15.4, df=2, P<0.001) where tall grass habitats were more expansive. Gravid female rates were significantly higher (χ2= 163.3, df = 3, P<0.001) during the fall (45.3%, range 25.5–57.3%) period than during the spring (24.6%, range 0.0–41.8%), summer (0.5, range 0.0–2.5), and winter (2.0, range 0.0–4.4) seasons.


IgG antibodies against hantaviruses were detected by IFA in A. agrarius (22.3%, 381/1,709), M. musculus (9.1%, 6/66), and M. fortis (5.0%, 1/20), while HTNV RNA was detected by RT-PCR only from A. agrarius (62.7%, 239/381) and M. musculus (16.7%,1/6). None of the other species of small mammals were Ab+ for hantaviruses. The mean Ab+ rates were 24.0% for Areas 1+2 and 20.8% for Area 3, but were variable over time, ranging from 0–38.5% for Areas 1+2 and 0–34.1% for Area 3 (Figure 2). Also for Area 3, a total of three M. musculus (15.8%) and one M. fortis (50.0%) were Ab+ for hantaviruses during spring 2005 when the HTNV rate in A. agrarius was 34.1%. Three other M. musculus were Ab+ for hantaviruses during spring 2002 (Area 3), summer 2003 (Area 2) and winter 2004 (Area 3) when HTNV rates in A. agrarius were 22.2%, 11.1%, and 8.7%, respectively. While seasonal Ab+ rates for HTNV were variable and asynchronous at all three areas surveyed, overall Ab+ rates, while not significantly different (χ2= 7.6, df = 3, P = 0.054), were generally higher during the fall periods and were associated with dramatic increases in the number of HFRS cases as reported by the Korea Centers for Disease Control and Prevention (KCDC).

Figure 2.

Seasonal1 and mean percent of A. agrarius Ab+ for HTNV for all habitats of Areas 1+2 and Area 3, DNTA, by collection period, March, 2001-December, 2005. A total of six and one, M. musculus and M. fortis, respectively, were Ab+ for HTNV (not included).2

The percent of Ab+A. agrarius males were significantly higher than for females captured at Areas 1 and 2 (range 29.5–40.9%♂, 18.3–24.1%♀) (χ2= 16.1, df = 2, P<0.001), while male and female Ab+ rates for Area 3 were comparable (χ2= 0.95, df = 1, P = 0.33). Similarly, while the overall annual percentage of Ab+ males and females were similar for spring (22.9♂, 20.7♀) (χ2= 0.80, df = 1, P = 0.37) and summer (24.0♂, 20.7♀) (χ2= 0.66, df = 1, P = 0.42) seasons, the percentage of Ab+ males were significantly higher during the fall (31.6♂, 23.3♀) (χ2= 4.96, df = 1, P = 0.026) and winter (24.2♂, 13.5♀) (χ2= 4.56, df = 1, P = 0.033) trapping periods. Most A. agrarius (88.6%) weighed between 10–30 g, while only a few (0.9%) weighed <10 g (Table 1). With the exception of those that weighed <10 g, Ab+ rates among the different weight classes increased as weights (age) increased (Figure 3).

Table 1.  Seasonal and total number of A. agrarius Ab+ for HTNV/number captured by weight category (%) at DNTA, March, 2001 to December, 2005.
Trapping SeasonWeight Class (g)TOTAL
  1. 1One mouse, which was caught in the door of the trap, is not included.

Figure 3.

Percent of A. agrarius Ab+ for HTNV at DNTA by weight category, March, 2001-December, 2005.

While the overall HTNV Ab+ rates (number Ab+/total number rodents tested) for A. agrarius were not significantly different between rodents captured at Areas 1 and 2 or Area 3 (χ2= 2.27, df = 1, P = 0.13), the numbers of HTNV Ab+ rodents/100 traps were 2.5-fold higher at Area 3 due to the greater number of rodents captured per 100 traps (Figure 4). The number of HTNV Ab+A. agrarius/100 traps was the lowest for narrow earthen banks separating rice paddies (1.7), and ranged from 1.7–4.5 for all habitats surveyed at Areas 1 and 2, while the number of Ab+ rodents was highest for crawling vegetation (12.2) and ranged 5.58–12.2 for all habitats surveyed at Area 3 (Figure 5).

Figure 4.

Seasonal1 number and mean of A. agrarius Ab+ for HTNV per 100 traps for Areas 1 and 2 and Area 3, DNTA, March, 2001 to December, 2005.

Figure 5.

Trap rates (bars), HTNV Ab+ rate (%), and number of HTNV Ab+ per 100 traps (line with squares) for A. agrarius captured at selected habitats1 for Areas 1 and 2 and Area 3, DNTA, March, 2001 to December, 2005.

Sequence analysis

The genetic diversity and phylogenetic relationship among Korean strains of HTNV was determined for strains obtained from the lung tissue of Ab+ rodents at DNTA. Partial sequencing of the Gc glycoprotein-encoding M segment from the HTNV strains yielded a 373-nt fragment, and a 324-nt length, which excluded primer sequences and were trimmed to match the lengths of the end sequences, was used for analysis. All sequenced HTNV isolates were submitted to GenBank (Accession numbers; FJ844368-FJ844395). The partial M segment sequence (coordinates 1991 to 2314) of 61 HTNV isolates from DNTA correspond to these reference numbers: 01–326, 01–339, 01–353, 01–356, 01–501, 01–513, 01–526, 01–599, 01–621, 02–128, 03–2, 03–4, 03–33, 03–53, 03–225, 03–239, 03–240, 03–260, 03–276, 03–287, 03–288, 03–297, 03–312, 03–524, 03–531, 03–534, 03–551, 03–556, 03–561, 03–578, 03–612, 03–794, 03–838, 03–839, 04–6, 04–115, 04–392, 04–405, 04–409, 04–441, 04–442, 04–445, 04–457, 04–494, 04–501, 04–502, 04–504, 04–505, 04–509, 04–511, 04–519, 04–1082, 04–1097, 04–1117, 05–166, 05–169, 05–197, 05–587, 05–603, 05–616, 05–645). The nucleotide and amino acid identity of the 61 HTNV isolates from DNTA varied between 0–3.1% and 0–2.8%, respectively.

Phylogenetic analysis

The N-J phylogenetic analysis, based on a 324-nt region of the Gc glycoprotein-encoding M segment of HTNV, demonstrated that HTNV isolates from A. agrarius from various HFRS-endemic areas in the ROK align to form two subgroups (prototype 76–118 subgroup and Lee subgroup) of group M1, as supported by a high bootstrap probabilities (89% and 74%) (Figure 6). Previously, analysis of the M segment sequences from available HTNV strains indicated that they formed seven groups (M1-M7), and HTNV showed extremely high genetic diversity in Guizhou, China (Wang et al. 2000; Zou et al. 2008). Our data indicate that the ROK HTNVs all belong to group M1 with Bao14 strain from Heilongjiang, China, and are evolutionarily distinct from HTNVs from China.

Figure 6.

Neighbor-joining phylogenetic cladogram based on a 324-nt region of the G2 glycoprotein-encoding M segment of the HTNVs amplified from A. agrarius captured at DNTA areas 1(DNI), 2(DNII), and 3(DNIII) (GenBank accession numbers; FJ844368-FJ844395). HTNV (Bao 14) and Soochong viruses (SOO-1 and H5) from the People's Republic of China originally isolated from the Korean field mouse (A. peninsulae), and Dobrava virus (DOB) originally isolated from the yellow-necked field mouse (A. flavicollis) are included from comparison. Genbank accession numbers for Muju (MUJ) 04–4, Puumala (PUU) Sotkamo, HTN Lee and HTN HoJo are EF198313, ×61034, D00377 and D00376, respectively. Branch lengths are proportional to the number of nucleotide substitutions, while vertical distances are for clarity only. The numbers at each node are bootstrap probabilities (expressed as percentages), as determined for 1,000 iterations by PAUP version 4.0b.


A seasonal rodent-borne disease surveillance program was initiated at DNTA in December, 2000 in response to a U.S. soldier acquiring HFRS attributed to exposure while training at DNTA in November, 2000. Our overall objectives were to identify HFRS risk factors including seasonal rodent bionomics and population densities, HTNV Ab+ positive rates (including density and distribution of Ab+ rodents), and environmental factors associated with training activities (O'Guinn et al. 2008, Song et al. 2009, Kim et al. 2011).

Competition for available habitat (vegetation and ground cover) also plays an important role in population densities and transmission of HTNV. For example, at Twin Bridges training area where three HFRS cases were attributed to exposure in 2005, unmanaged lands consisting of tall grasses and other herbaceous vegetation bordered the primary training sites and dirt roads and provided habitat for large populations of small mammals. Conversely, at Firing Point-60, where two HFRS cases were attributed to exposure in 2000 and 2005 (Song et al. 2009, Kim et al. 2011), small mammal populations were significantly lower for space-limited tall grass habitats at Areas 1 and 2. These differences were reflected in overall capture rates of 40.3% for Area 3, while for narrow habitats (<20 m wide) at Areas 1 and 2 that bordered dirt roads and forested hillsides, capture rates were 15.3%, and for even more limited habitats, e.g., (0.5–1.0 m) earthen banks separating rice paddies, the overall capture rate was only 5.5%. Conversely, overall HTNV Ab+ rates tended to be highest for habitats of limited space and lowest in areas of extensive grasses and crawling vegetation; potentially a consequence of greater competition for space via wounding/biting and rodent-to-rodent transmission of HTNV (Mackelprang et al. 2001, Hinson et al. 2004, O'Guinn et al. 2008, Sames et al. 2009a, Song et al. 2009, Kim et al. 2011. However, even through Ab+ rates were lower for expansive tall grass habitats, the total number of HTNV Ab+ rodents was, on average, higher than for restricted habitats. Therefore, while increasing HTNV Ab+ rates are indicators of higher disease risks, it may be misleading to only review HTNV Ab+ rates as the primary indicator of HFRS transmission risks.

Convoys on dirt roads, which lead to and within the training areas, off-road maneuvers in rodent infested habitats, backblast from artillery, and helicopter resupply operations contribute to aerosolized dusts that are potentially contaminated with virus-rodent excreta increase HTNV transmission risks. Annual burning of the training sites, which reduce vegetation cover during non-growth periods, has been proposed to reduce rodent populations. However, the effect of increased exposure to predation, movement of rodents in search of more suitable habitats, and changes in corresponding transmission of HTNV (Ab+ rates) for disturbed rodent populations is unknown. Thus, pre- and post-rodent-borne disease surveillance that evaluates intervention effects to limit exposure through rodent population reduction is necessary.

Annual A. agrarius populations and reported HFRS cases remain relatively constant in the ROK (KCDC 2010). Seasonal populations (trap rates) are lowest during the early fall as a result of an abundance of natural food and mortality during the previous months when reproduction is low, whereas winter populations rebound due to high fall reproductive rates (O'Guinn et al. 2008, Kim et al. 2011). The infusion and movement of naïve young rodents during the fall that seek suitable winter habitat may increase HTNV acute infections in the rodents through competition (biting) (Hinson et al. 2004). Yet, while rodent-to-rodent transmission may be high during this period, HTNV Ab+ rates in the rodents at DNTA and other survey sites (FP-10 and FP-60) have been found to be highly variable seasonally, i.e., high rates observed during the winter, spring, or summer seasons without a corresponding increase in HFRS cases (KCDC 2010). These data suggest that the infusion of young HTNV naïve A. agrarius in late August/September results in greater numbers of acute infections with corresponding increases in virus shedding and HFRS cases from September-December, while subsequent low reproductive rates and a higher proportion of chronic infections associated with lower levels of viral shedding result in fewer observed HFRS cases during most of the rest of the year (Niklasson et al. 1995, Olsson et al. 2009).

Before 2005, data were insufficient for identifying sites of human HTNV transmission, as most training exercises were conducted at several training sites over a one- to four-week period. New published information shows that HTNV strains are variable geographically and that at any given site, a minimum number of genetic variants are found circulating in associated rodent populations (Song et al. 2009). However, while genetic strains of HTNV predominated at each of the three areas surveyed at DNTA, the areas were only separated by a narrow forested hillside and 5-m rock-faced berm, which may have allowed for intermixing of rodent populations and associated virus strains. In summary, field conditions during training that increase the potential for HTNV transmission can be determined and applied to disease mitigation strategies for military training areas throughout the ROK that have been surveyed. In addition, epidemiological surveys conducted subsequent to HFRS cases can be utilized for developing site-specific HFRS mitigation protocols.


We thank COL Brian Allgood (deceased, and for whom the Brian Allgood Army Community Hospital (BAACH) was recently named) and Hee-Choon (Sam) Lee for their support. We thank Suk Hee Yi, Force Health Protection and Preventive Medicine, 65th Medical Brigade, for analysis of data and GIS mapping; Jiun Yoon and Amy Nguyen, attending physicians, BAAHC, and Rex Bergren, Chief, Laboratory Services, BAAHC, U.S. Army MEDDAC-Korea, for their support; and K. Kenyon, USAMRIID, for her editorial suggestions. Funding for portions of this work was provided by the Division of Global Emerging Infections Surveillance and Response System (GEIS) Operations, Armed Forces Health Surveillance Center, Silver Spring, MD, and the National Center of Medical Intelligence, Fort Detrick, MD. The views expressed in this article are those of the authors and do not reflect the official policy or position of the Department of Defense, the Department of the Army, or the U.S. Government.