VECTOR-BORNE AND ZOONOTIC DISEASESVolume 3 Number 4 2003copy Mary Ann Liebert Inc

Research Paper

Evaluation of a Prototype Ehrlichia chaffeensisSurveillance System using White-Tailed Deer (Odocoileus virginianus) as Natural Sentinels

MICHAEL J YABSLEY12 VIVIEN G DUGAN12 DAVID E STALLKNECHT12

SUSAN E LITTLE2 J MITCHELL LOCKHART3 JACQUELINE E DAWSON4

and WILLIAM R DAVIDSON15

ABSTRACT

The natural history of Ehrlichia chaffeensis the causative agent of human monocytotropic ehrlichiosis includesthe lone star tick (LST Amblyomma americanum) as a vector and white-tailed deer (WTD Odocoileus virginianus)as both a natural reservoir of E chaffeensis and a major host of LST The goal of the current study was to imple-ment and evaluate a prototype surveillance system to delineate the geographic distribution of E chaffeensis us-ing WTD as natural sentinels To accomplish this goal serologic testing using the indirect immunofluorescent an-tibody (IFA) test was performed on WTD serum samples and to confirm serologic results polymerase chainreaction (PCR) assays and culture isolation were conducted Considerations relevant to the applicability of a sur-veillance system utilizing WTD were analyzed (eg age and gender relationships to serologic status adequacy ofsample sizes needed to distinguish between uninfected and infected populations presence of LST and ability todetect stability and spread of E chaffeensis in WTD populations) Of 3275 WTD serologically tested 549 (47)from 17 of 18 states had antibodies reactive to E chaffeensis (IFA titer $ 1128) No difference between age groupsor gender was noted with serologic testing thus these variables would not be a concern for a surveillance systemusing WTD Significantly more deer in younger age groups (15 yr) were PCR and culture positive and 46 of122 seropositive WTD populations were confirmed positive by PCR or culture isolation A significant associationbetween LST infestation and E chaffeensis seroreactivity was noted Furthermore the surveillance system wasable to detect stability of E chaffeensis within WTD populations and also spread to new populations both ofwhich were associated with LST status These data clearly demonstrate that WTD are useful as natural sentinelsfor this emerging human pathogen and establish a prototypical framework for a WTD surveillance system KeyWords Ehrlichia chaffeensismdashWhite-tail deermdashSurveillance systemsmdashAmblyomma americanummdashTick-bornemdashSerology Vector-Borne Zoonotic Dis 3 195ndash207

195

INTRODUCTION

Ehrlichia chaffeensis the causative agent of hu-man monocytotropic ehrlichiosis (HME) is

maintained in a zoonotic cycle involving white-tailed deer (WTD Odocoileus virginianus) as the

primary vertebrate reservoir and lone star ticks(LST Amblyomma americanum) as biological vec-tors (see Childs and Paddock 2003) Many pop-ulations of white-tailed deer from the southeast-ern United States have been shown to haveantibodies reactive to E chaffeensis (Dawson et al

1Southeastern Cooperative Wildlife Disease Study and 2Department of Medical Microbiology and ParasitologyCollege of Veterinary Medicine The University of Georgia Athens Georgia

3Department of Biology Valdosta State University Valdosta Georgia4Chelan-Douglas Health Department 200 Valley Mall Parkway East Wenatchee Washington5DB Warnell School of Forest Resources The University of Georgia Athens Georgia

1994a) and experimental and field studies haveproven that WTD are competent and persistentlyinfected reservoirs (Dawson et al 1994b Lock-hart et al 1997a Davidson et al 2001 Yabsley etal 2002) The LST was initially suspected to be avector based on the geographic distribution ofHME cases and this was confirmed by poly-merase chain reaction (PCR) detection of E chaf-feensis DNA in LST (Anderson et al 1993 Rolandet al 1998 Steiert and Gilfoy 2002) and by ex-perimental transstadial transmission betweenWTD (Ewing et al 1995) In addition to being im-portant reservoirs for E chaffeensis WTD are animportant host for all three mobile stages of LST(Bloemer et al 1988)

The potential utility of WTD as a sentinel fordetermining the distribution of E chaffeensiswas first suggested when antibodies reactive toE chaffeensis were detected in WTD (Dawsonet al 1994a) and a prototype WTD serologicsurveillance system for E chaffeensis recentlywas applied statewide in Iowa (Mueller-An-neling et al 2000) White-tailed deer have beenutilized as sentinel animals to monitor numer-ous human and livestock pathogens includingJamestown Canyon virus (Boromisa and Grim-stad 1987) Venezuelan equine encephalitis(Smart and Trainer 1975) Lyme disease (Gill etal 1993) vesicular stomatitis virus (Fletcher etal 1991) and bluetongue and epizootic hem-orrhagic disease viruses (Stallknecht et al1991) The advantages associated with surveil-lance systems utilizing deer include (1) a dis-tribution that includes 45 states allowing de-velopment of national regional or state-widesurveillance systems (2) relatively high abun-dances allowing use as sentinels in most loca-tions (3) regulated harvests in all states en-hancing the ease and reducing the cost ofcollecting samples from hunter-killed animals(4) limited size of deer home ranges (mean ap-proximately 400 ha) allowing identification ofexposure location (5) a much higher rate of ex-posure to tick vectors compared to humans orpotential domestic animal sentinels (6) rela-tively long life spans increasing opportunity forexposure (7) opportunities to simultaneouslymonitor for presence of tick vectors (8) occur-rence in both natural habitats and areas proxi-mal to human habitation and (9) freedom fromprior treatment with antibiotics or acaracides

Field studies have demonstrated that serologicmonitoring of deer would allow for fine-scalemapping of the distribution of E chaffeensis(Lockhart et al 1996 Mueller-Anneling et al2000) Using a sample size of only five animalsper population Lockhart et al (1996) success-fully discriminated between LST infested popu-lations which had a 100 antibody prevalenceand populations believed to be LST-free whichhad a seroprevalence of 67 Although E chaf-feensis-reactive antibodies are reported for manyWTD populations confirmation of infection us-ing PCR andor culture has only been reportedfor 10 seropositive populations in ArkansasGeorgia Kentucky North Carolina and SouthCarolina (Lockhart et al 1997a 1997b Little et al1997 1998 Yabsley et al 2002)

The overall goal of this study was to imple-ment and evaluate an extensive fine-scale(county) surveillance system for E chaffeensisutilizing WTD as natural sentinels Specific ob-jectives to achieve this goal were to (1) combinecurrent and prior WTD serologic data to con-struct an extensive regional database on E chaf-feensis-reactive antibodies in WTD (2) use PCRassays and cell culture isolation testing to con-firm E chaffeensis infection in seropositive WTDpopulations (3) document any age class or gen-der relationships to antibody prevalence andPCR detection (4) evaluate sample sizes thatwould be functionally effective at distinguish-ing between infected and uninfected popula-tions (5) affirm that infection status of WTDpopulations corresponds with the presence ofLST and (6) investigate the ability of a WTDsurveillance system to consistently discern thedistribution or spread of E chaffeensis amongWTD populations over time

MATERIALS AND METHODS

Sample collections

White-tailed deer samples specifically forthis project were collected from sites selectedto complement and further expand four previ-ous studies where WTD in the south centraland southeastern United States were tested forE chaffeensis-reactive antibodies (Dawson et al1994a Lockhart et al 1996 Little et al 1997Yabsley et al 2002) Blood samples were col-

YABSLEY ET AL196

lected from 2101 WTD from 471 populationsin 18 states (AL AR GA FL KS KY LA MDMO MS NC NJ OK SC TN TX VA andWV) during Southeastern Cooperative WildlifeDisease Study herd health evaluation activitiesother research projects or in cooperation withstate wildlife agency personnel from hunter-harvested animals Blood samples were col-lected either aseptically from the heart fromthe jugular vein or from the body cavity Oneexception was that blood samples from 40 of79 deer from Missouri were collected on Nob-uto strips and eluted in phosphate-bufferedsaline (PBS) pH 74 as described (Mueller-An-neling et al 2000) Serum plasma or elutedblood samples were frozen at 220degC until sero-logic testing From a subset of WTD (368 from112 seropositive populations and 61 from 20seronegative populations) whole blood sam-ples were collected and frozen at 220degC forPCR testing Ticks were collected from a sec-ond subset of 715 WTD (117 populations) andsubmitted to the National Veterinary ServicesLaboratories (US Department of AgricultureAmes IA) for identification Animals werehandled and samples collected by proceduresapproved by the Animal Care and Use Com-mittee at the University of Georgia (A3437-01)

Serologic assays

Serum plasma or eluted blood sampleswere tested for antibodies reactive to E chaf-feensis by the indirect immunofluorescent anti-body (IFA) test as previously described (Daw-son et al 1994 Lockhart et al 1996) Sampleswere screened at a dilution of 1128 using Echaffeensis antigen slides (Focus TechnologiesCypress CA) A 150 dilution of fluoresceinisothiocyanate-labeled rabbit anti-deer im-munoglobulin G (Kirkegaard and Perry Labo-ratories Gaithersburg MD) was used as con-jugate Geometric mean titer (GMT) wascomputed for 152 seropositive deer

Molecular assays

DNA from 300 ml of 429 whole blood sam-ples was extracted using the GFX GenomicBlood DNA Purification Kit (Amersham Phar-macia Biotech Piscataway NJ) following themanufacturerrsquos protocol Two gene targets the

16S rRNA gene and the variable length PCRtarget (VLPT) antigen gene were utilized forscreening of whole blood for E chaffeensis DNAusing a nested PCR format Primary outsideamplification for the 16S rRNA gene consistedof 5 mL of DNA (from whole blood) in a 25-mLreaction containing 10 mM Tris-Cl (pH 83) 50mM KCl 15 mM MgCl2 02 mM each dNTP(Promega Madison WI) 25 units Taq DNAPolymerase (Promega) and 08 mM of primersECC and ECB (Dawson et al 1994b 1996a) Forthe nested PCR 1 mL of primary product wasused as template in a 25-mL reaction contain-ing the same PCR components except E chaf-feensis specific primers HE1 and HE3 (Ander-son et al 1992) were used

Nested PCR for the VLPT antigen gene wasconducted on 429 blood samples collected dur-ing this study as previously described (Sumneret al 1999) In addition 101 blood samples froman additional 10 populations previously testedusing 16S rRNA primers (Yabsley et al 2002)were tested for E chaffeensis using the VLPTgene target Amplified products were sepa-rated in 2 agarose gels stained with ethidiumbromide and visualized with UV light To con-firm identity representative secondary VLPTgene amplicons were purified with a Microconspin filter (Amicon Inc Beverley MA) se-quenced at MWG-BIOTECH (High Point NC)and compared to published E chaffeensis VLPTsequences in the GenBank database

Stringent protocols and controls were uti-lized in all PCR assays to prevent and detectcontamination DNA extraction primary am-plification secondary amplification and prod-uct analysis were performed in separate dedi-cated laboratory areas Two negative watercontrols were included in each set of DNA ex-tractions and one water control was includedin each set of primary and secondary PCR re-actions

Culture isolation

During 2001ndash2003 aseptically collectedblood samples from 72 WTD from 15 LST pos-itive populations were cultured in DH82 cellsfor isolation of E chaffeensis as previously de-scribed (Lockhart et al 1997a) Cultures werefed biweekly with MEM medium (Sigma St

197EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

Louis MO) supplemented with 10 fetalbovine serum (Sigma) and monitored for evi-dence of cytopathic effect or for a maximum of45 days Cultures showing cytopathic effectand cultures completing the 45-day incubationperiod were harvested with a cell scraper andtested by direct fluorescent antibody assay aspreviously described (Lockhart et al 1997a)

Data analysis

Serologic data from this study (n 5 2101)were combined with data from previous stud-ies conducted independently by SCWDS re-searchers (n 5 425 Lockhart et al 1996 Littleet al 1997 Yabsley et al 2002) or collabora-tively with others (n 5 749 Dawson et al1994a) to construct a comprehensive map delineating the distribution of E chaffeensisseroreactive WTD in the southeastern andsouth central United States Some of the previ-ous studies used different cutoff values for IFAtesting (eg 164 or 1128) but in order to stan-dardize results from this study and previousstudies IFA assays were classified as positiveonly if a sample had a titer of $1128 To facil-itate graphic presentation data for each popu-lation were categorized by county or parish ifone or more deer with antibodies reactive to Echaffeensis was detected that county or parishwas classified as seropositive

To confirm serologic data PCR andor cul-ture isolation were conducted on a total of 541deer from 122 seropositive populations and on61 deer from 20 seronegative populations Chi-square analysis was used to test for differencesof PCR prevalence between seropositive andseronegative populations

Prevalence of antibodies reactive to E chaf-feensis among different age classes and gendercategories were only determined for WTD inseropositive populations Serologic data fromDawson et al (1994) Lockhart et al (1996) Lit-tle et al (1997) Yabsley et al (2002) and thisstudy were combined for determining preva-lence among age classes because analysis ofWTD age data had not been previously con-ducted PCR testing was not conducted byDawson et al (1994) and Lockhart et al (1996)therefore analysis of age and PCR positivitywas conducted only on deer from Little et al(1997) Yabsley et al (2002) and this study Chi

square analysis was used to test for differencesin both seroprevalence and PCR positivityamong age classes

To accurately classify the serostatus of a pop-ulation an appropriate sample size needs to betested In previous studies testing of five deerfrom a population consistently detected E chaf-feensisndashreactive antibodies in deer populationsinfested with LST (Lockhart et al 1997 Little etal 1997 1998) and based on Dawson et al(1994) testing a mean of 21 deer failed to detectseroreactive deer in seronegative populationsHowever additional evaluation of the samplesizes required to reliably classify the serologicstatus of populations was conducted in thisstudy To determine minimal number of sam-ples needed to detect the presence of a seropos-itive deer in a population the formula n 5(1 2 (1-a)1D) (N 2 (D 2 1)2) was used as de-scribed (Thrusfield 1995) Because seronegativepopulations are more difficult to accuratelyclassify two methods were used to examine ad-equacy of sample size Larger numbers of deer(n 5 8 2 16) were tested from nine seronega-tive populations to enhance detection of po-tentially low prevalences in those populationsAlso four populations with no history of LSTinfestation were repeatedly tested for severalyears

To assess the predicted epidemiologic asso-ciation between LST infestation and serologicstatus of WTD populations chi-square analy-sis was used to test for an association betweenLST presence and seroreactors at the popula-tion level Furthermore 11 populations forwhich both tick and serologic data were avail-able were evaluated during multiple yearsthree populations had LST infestation eachsample year four never had any evidence ofLST infestation and four were negative for LSTduring earlier sampling period (s) but becameLST positive during subsequent test years Chi-square analysis was used to test for differencesbetween groups

RESULTS

Regional serologic database for WTD

Among the 2101 WTD collected specificallyfor this study 984 (468 CI95 5 447

YABSLEY ET AL198

489) had antibodies reactive ($1128 titer) toE chaffeensis by IFA testing (Table 1) The meanprevalence of antibodies reactive to E chaffeen-sis in seropositive populations was 738(SD 5 258 range 20ndash100) White-tailed deerwith antibodies reactive to E chaffeensis weredetected in all states tested except West Vir-ginia The modal antibody titer for seropositivedeer was 1128 with a maximum titer of 14096The GMT of a subset of 152 seropositive WTDwas 355

The combined dataset of WTD serologicallytested during this and four prior studies (Daw-son et al 1994a Lockhart et al 1996 Little etal 1997 Yabsley et al 2002) contained a totalof 3275 WTD from 18 southeastern and southcentral states In the combined data set 1549(473) WTD examined had antibodies reactiveto E chaffeensis with seropositive populationsdetected in 17 of 18 states (Fig 1)

The highest overall seroprevalences were inArkansas and Missouri Although high preva-lences were detected in local WTD populationsin most states considerable variation in preva-lence occurred within different regions of somestates (eg Florida Kansas North Carolina

Oklahoma and Virginia) A western boundaryfor the distribution of E chaffeensis was evidentspanning the states of Kansas Oklahoma andTexas Similarly a southern boundary wasnoted across peninsular Florida Clusters ofseronegative populations were detected insouthern Mississippi and Louisiana along thelower Mississippi River floodplain and in theAppalachian Mountains from western Mary-land to northern Alabama

PCR and culture validation of E chaffeensisserologic data

Eighty-six of 368 (234) WTD from 50seropositive populations were PCR positive forE chaffeensis by either 16S or VLPT PCR Of 101WTD previously tested by 16S PCR (Yabsley etal 2002) nine deer from seven populationswere positive by VLPT PCR Collectively forthis study and Yabsley et al (2002) 95 of 469(203) WTD from 56 of 122 (46) seropositivepopulations have been confirmed using PCR(Table 1) 70 of 95 deer with both gene targets2 of 95 for only the 16S rRNA gene and 23 of95 only for the VLPT gene In contrast 16S

199

TABLE 1 RESULTS OF SEROLOGIC AND POLYMERASE CHAIN REACTION TESTING

FOR EHRLICHIA CHAFFEENSIS IN WHITE-TAILED DEER FROM 18 STATESa

Counties Number IFA Number PCRtested by positive positive

State Years IFA test no tested () no tested ()

Alabama 1981ndash2001 33 59141 (42) NTb

Arkansas 1982ndash2002 29 111154 (72) 746 (15)Florida 1983ndash2002 15 43104 (40) 936 (25)Georgia 1973ndash2002 50 119243 (49) 2698 (27)Kansas 1998ndash2002 38 3976 (51) 326 (12)Kentucky 1983ndash2001 27 2699 (26) 223 (9)Louisiana 1981ndash2001 34 54166 (33) 118 (6)Maryland 1999ndash2002 5 1225 (48) 17 (14)Missouri 1994ndash2002 19 5779 (72) 323 (13)Mississippi 1995ndash2001 38 80152 (53) 119 (5)North Carolina 1982ndash2002 33 75131 (57) 1572 (21)New Jersey 2001 2 610 (60) 05Oklahoma 1985ndash2002 13 3050 (60) 03South Carolina 1991ndash2001 26 52124 (42) 460 (7)Tennessee 1981ndash2001 20 54100 (54) 010Texas 1992ndash2002 33 73123 (59) 013Virginia 1983ndash2002 37 94168 (56) 2357 (40)West Virginia 1977ndash2001 18 0156 014Total 471 9842101 (47) 95530 (18)

aIncludes PCR results for the VLPT antigen gene for 101 WTD previously tested by PCR for the 16S rRNA gene(Yabsley et al 2002)NT not tested

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

rRNA and VLPT PCR assays on 61 deer from20 seronegative populations were all negativePresence of PCR positive deer was strongly as-sociated with seropositive populations (x2 5116 p 0001)

Of 72 culture attempts 34 (442) were lostto contamination with bacteria andor Try-panosoma cervi Seven of the remaining 38 cul-tures (184) were positive for E chaffeensisPositive cultures were obtained from two WTDfrom Greene Co AR two from Woodruff CoAR and one each from Clarke Co GA AnsonCo NC and Georgetown Co SC All isolateswere confirmed as E chaffeensis by PCR of the16S rRNA and VLPT gene targets

Relationship of age class and gender to antibody prevalence and PCR detection

Age was available on 1882 deer tested sero-logically 1086 deer from this study 517 fromDawson et al (1994) 165 from Lockhart et al

(1996) five from Little et al (1997) and 109from Yabsley et al (2002) For this combineddata set differences were not noted (x2 5 565p 5 046) in seroprevalance among age classes(Fig 2A) Similarly seroprevalence did not dif-fer between males and females (693 vs696 x2 5 0016 p 005)

Age was available for 393 deer from seropos-itive populations that were tested by PCRHigher proportions of deer were PCR positivein the younger age classes (x2 5 717 p 0001) Among deer 075 years old 531were PCR positive within the next age group(076ndash15 years) the prevalence decreased sig-nificantly to 27 (x2 5 147 p 0001) and lessthan 8 of deer 15 years old were PCR pos-itive (Fig 2B) The mean age of PCR positiveWTD was 11 years (range 025ndash4 years) In-sufficient culture isolates were made to exam-ine age-related associations however themean age for culture positive deer was 32years (range 03ndash95 years) Similar to antibody

YABSLEY ET AL200

FIG 1 Indirect immunofluorescent antibody results for Ehrlichia chaffeensis in white-tailed deer Results are from thecurrent study and from previously published work obtained in collaboration with SCWDS Darkly shaded counties pos-itive and lightly shaded counties negative for antibodies ($1128) reactive for E chaffeensis Previous serologic resultsthat differed from the results of the current study are shown as small circles within the county d previously positivefor antibodies ($1128) reactive for E chaffeensis s previously negative for antibodies reactive for E chaffeensis

prevalence PCR positivity did not differ be-tween males and females (19 vs 228 x2 5089 p 005)

Evaluation of sample size adequacy

Regarding sample size for seropositive pop-ulations post hoc statistical analysis utilizing amean 73 seroprevalence indicated that test-ing of only three or four deer gives a high prob-ability (95 and 99 respectively) of detectinga positive population (Thrusfield 1995) Al-though validation of seronegative populationsis more problematic testing of larger numbersof deer per population and repeated testing ofselected populations both produced consistentnegative results In a combined dataset fromDawson et al (1994a) and this study 250 deerfrom 17 seronegative populations (9ndash30 deer

per population) were negative for antibodies toE chaffeensis with means of 208 and 104 deertested per seronegative population respec-tively Similarly none of the 131 deer from fourpopulations consistently negative for LST in-festation had antibodies reactive to E chaffeen-sis The mean number of WTD tested per pop-ulation in this study was 62 (SD 5 61 range1ndash58) and data for populations with samplesizes less than four deer were used only if aseropositive animal was detected

Association between LST and serologic status at population level

Of 117 WTD populations examined for ticks82 were infested with LST 29 with A macula-tum (Gulf Coast tick) four with Ixodes affinisand 13 with I scapularis (blacklegged tick)

201

FIG 2 (A) Prevalence of antibodies reactive to E chaffeensis in WTD among age classes Numbers in bars representnumber of deer tested (B) Prevalence of PCR-positive WTD among age classes Numbers in bars represent numberof deer tested and different letters represent significant differences at p 005

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

Eighty of the 82 (976) populations infestedwith LST contained at least one seropositiveWTD while only four of 35 (114) populationsnot infested with LST contained seropositivedeer One LST-negative population (Harris CoGA) was both seronegative and tick negativein 1986 however in 2002 the deer populationwas seropositive but the LST status was notevaluated The prevalence of antibodies reac-tive to E chaffeensis was significantly higheramong LST infested populations of WTD thanpopulations not infested by LST (x2 5 899 p 0001) None of the six populations infestedonly with A maculatum contained seropositivedeer there were insufficient populations withI affinis or I scapularis alone to permit evalua-tion

Stability and spread of E chaffeensis and LST among WTD populations

Fifty-seven populations which were testedduring previous studies (Dawson et al 1994aLockhart et al 1996) were retested during thisstudy The serologic status for 50 (88) of thesepopulations remained unchanged 35 seropos-itive populations remained seropositive and 15seronegative populations remained seronega-tive Six populations which previously testedseronegative were seropositive during this

study Three of these representing Anson CoNC Haywood Co TN and Stewart Co GAhad documented invasion of LST between thefirst reported testing and this study One Texaspopulation (Travis Co) was seropositive (n 57 714) when tested in 1992 (Dawson et al1994a) but was seronegative when sampled(n 5 10) in 2002 for this study The LST statusfor this county was not known for either sam-pling period

Among the 11 populations where LST infes-tation and serologic status were monitored repeatedly over time there was complete con-cordance between LST presence and seroposi-tive deer (Table 2) All three LST positive pop-ulations were consistently seropositive eachyear tested The mean seroprevalence for thesethree populations was 83 and two of thesepopulations (Jones Co GA and Ashley CoAR) were confirmed PCR positive for E chaf-feensis None of the four populations where LSTwas consistently absent contained seropositivedeer at any sampling time Among the fourpopulations where LST status changed fromabsent to present seropositive deer were onlydetected following the appearance of LST Themean prevalence of antibodies reactive to Echaffeensis in the most recently tested year wassignificantly higher for populations infestedwith LST for multiple years (mean 5 100)

YABSLEY ET AL202

TABLE 2 SEROLOGIC AND TICK SURVEY RESULTS FROM SEVEN POPULATIONS OF WTD COLLECTED DURING

MULTIPLE TIME PERIODS AND TESTED FOR ANTIBODIES REACTIVE FOR E CHAFFEENSIS

No of Tickyears status No of deer Mean IFA prevalence

Site Years sampled (years) tested (range)

Sites positive for ticksJones Co GA 1979ndash2001 6 1 (6) 41 92 (80ndash100)Ashley Co AR 1982ndash2001 7 1 (7) 37 84 (71ndash100)Franklin Co FL 1984ndash1996 5 1 (5) 25 68 (40ndash100)

Sites negative for ticksFloyd Co GA 1973ndash2001 6 2 (6) 58 0Tyler Co WV 1985ndash2000 4 2 (4) 21 0Hardy Co WV 1977ndash1999 7 2 (7) 37 0Monroe Co FL 1992ndash2001 4 2 (4) 15 0

Sites where tick status changedAnson Co NC 1987 2001 2 2 1a 10 0 40bConcordia Pa LA 1986 1991 1999 3 2 1 1 22 0 38 60Haywood Co TN 1989 1994 1998 3 2 2 1 16 0 0 20Stewart Co GA 1986 1998 2 2 1 10 0 20

aTick status at sequential sampling periodsbPrevalence of IFA seropositive deer at sequential sampling periods

compared to populations only recently infested(mean 5 538 x2 5 155 p 0001) One re-cently infested population (Anson Co NC)was confirmed by isolation of E chaffeensisfrom a single WTD in 2001

DISCUSSION

The overarching goal of this study was toimplement and evaluate a prototype surveil-lance system using WTD as natural sentinels todetermine the geographic distribution of Echaffeensis across most of the suspected HMEendemic region of the United States To thisend the data obtained demonstrate the fol-lowing critical attributes (1) serologic findingsreflect E chaffeensis infection (2) effective sur-veillance can be achieved with small samplesizes (3) any deer 6 month old is suitable forsurveillance purposes (4) enzootic and consis-tently negative locales were identified repeat-edly across time (5) surveillance of deer hasthe ability to detect spread to new locales and(6) serologic molecular and culture diagnosticfindings among deer could be related to pres-ence of the principal tick vector A americanumThese findings compliment and expand on arecent WTD serologic surveillance effort withinIowa (Mueller-Anneling et al 2000) howeverthat study exclusively used serologic assaysand did not include either confirmatory mi-crobiological diagnostics or contemporaneoustick vector monitoring Collectively these twostudies confirm the effectiveness of this proto-typical surveillance system for monitoring thedistribution of this emerging human pathogen

Although several studies have examined theprevalence and distribution of E chaffeensis an-tibodies among many southeastern and southcentral WTD populations (Dawson et al 1994aLockhart et al 1996 Little et al 1997 Yabsleyet al 2002) the extensive additional testingfilled many geographic gaps and provide amore complete fine scale distribution of E chaf-feensis among WTD throughout the regions ofthe United States with the highest HME risk(Paddock and Childs 2003) The combined dataevaluated herein disclosed a wide distributionof seropositive WTD populations across all ofthe 18 states represented except for West Vir-

ginia however some distributional limits alsowere discerned including a western boundaryextending across Kansas Oklahoma andTexas a southern boundary across peninsularFlorida and a large cluster of many seronega-tive populations centered along the Ap-palachian Mountains (Fig 1) Other prior stud-ies of WTD have documented an apparentnorthern range limit for E chaffeensis in south-ern portions of Iowa Illinois Indiana and Ohio(Dawson et al 1994a Irving et al 2000 Mueller-Anneling et al 2000) Along the East coast sur-veys have indicated that A americanum and Echaffeensis are present as far north as Connecti-cut Maine and Rhode Island (Keirans and La-combe 1998 Ijdo et al 2000)

The accuracy of diagnostic assays utilized iscritical to any pathogen surveillance system In-fection with E chaffeensis was confirmed innearly half (46) of 122 seropositive WTD pop-ulations tested by PCR and was supplementedby isolation from others In contrast PCR evi-dence of infection was not detected in any of20 seronegative WTD populations Thus PCRand culture testing applied at the populationlevel were effective at confirming infection inmany seropositive populations and demon-strating a strong association between the sero-logic status of populations and infection withE chaffeensis

Because WTD in these regions are known to be infected with three other related rickettsiae(E ewingii A phagocytophilum and an Ana-plasma sp [WTD-agent]) in addition to E chaf-feensis (Dawson et al 1996b Little et al 1998Brandsma et al 1999 Magnarelli et al 1999Yabsley et al 2002) the potential for serologiccross-reaction is an important considerationAlthough antibodies reactive with E chaffeensisin WTD from Maryland have been confirmedby immunoblotting using the 28- to 29-kDa anti-gens of E chaffeensis (Walls et al 1998) the ex-tent of serologic cross-reactions among thesefour species is not fully understood (Dumlerand Walker 2001) However data from otherstudies suggests that infection with these otherrickettsiae may not commonly result in pro-duction of antibodies that cross-react with Echaffeensis Cross-reactions between E chaffeen-sis and E ewingii have been noted in infectedhumans and dogs however not all E ewingii

203EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

infected dogs or humans develop antibodies toE chaffeensis antigens (Murphy et al 1998 Pad-dock et al 2001) Similarly a WTD fawn ex-perimentally infected with E ewingii did not de-velop antibodies reactive with E chaffeensis(Yabsley et al 2002) Cross-reactions between Echaffeensis and A phagocytophilum have been re-ported in humans and may complicate diagno-sis (Comer et al 1999) Far fewer WTD in thesoutheastern states have antibodies reactive toA phagocytophilum (25) than to E chaffeensisantigens however although some deer reactedto both E chaffeensis and A phagocytophilum ator above the 1128 cut-off many also reacted toonly one of these antigens at titers $1128 (Lit-tle et al 1998 Walls et al 1998 VG Dugan andMJ Yabsley unpublished data) The high per-centage of populations where seropositive orseronegative status was confirmed by PCR orculture together with data from these otherstudies provide considerable evidence that theseroreactivity reported herein largely repre-sents specific seroconversion to E chaffeensis

The distribution of antibodies within WTDage and gender categories was investigated be-cause these variables are an important consid-eration for a WTD natural sentinel systemOnly a limited investigation of geometric meantiters among age categories has been reportedfor WTD (Lockhart et al 1995) and any signif-icant influence of age or gender on the occur-rence of antibodies would require samplingstratified according to these host attributes Im-portantly differences in antibody prevalencewere not noted among age or gender categoriesindicating that all WTD particularly animals $6 mo of age are suitable for use in a WTD sur-veillance system In contrast to the high stableprevalence of antibodies among age classes(Fig 2A) PCR evidence of rickettsemia de-clined with age (Fig 2B) This pattern conformsto proposed infection dynamics in naturally in-fected WTD whereby animals from approxi-mately 6 months to 15 years of age are morelikely to be rickettsemic than older adults (Pad-dock and Childs 2003 Dawson et al 1994Lockhart et al 1997a Davidson et al 2001) Inthe present study the prevalence of PCR pos-itivity dropped dramatically from over half(53) of WTD younger than 075 years to lessthan 8 of WTD older than 15 years The ma-

jority of deer tested from some seropositivepopulations were 15 years and only deer15 years from other populations were avail-able for PCR testing which may explain whynot all seropositive populations were con-firmed by PCR Because the probability of aWTD being rickettsemic declines with ageserology represents a better surveillance toolthan PCR and under natural conditions of re-exposure to ticks titers likely do not declineover time as often occurred in single exposureexperimental infections of WTD (Dawson et al1994b Ewing et al 1995 Davidson et al 2001)

Previous serologic surveys of E chaffeensis inWTD (Dawson et al 1994a Lockhart et al 1996Little et al 1997 Yabsley et al 2002) have uti-lized relatively small sample sizes per popula-tion but except for one (Mueller-Anneling etal 2000) have not addressed the adequacy ofthese sample sizes to correctly classify infectedand uninfected populations Data from the cur-rent and prior studies (Dawson et al 1994aLockhart et al 1996 Little et al 1997) haveshown that infected WTD populations in thesoutheastern and southcentral United Statestypically have high (70) seroprevalencesCalculations (Thrusfield 1995) based on a mean73 seroprevalence indicated that use of a sam-ple size of five WTD per population should re-liably detect nearly all positive populations Al-though it was beyond the scope of this studyto test enough deer to detect very low popula-tion prevalences larger sample sizes and re-peated sampling of seronegative populationsdid not reveal seropositive populations thatwould have been misclassified as seronegativebased on smaller sample sizes A chance of mis-classifying newly infected populations of WTDexists because seroprevalence may be low ini-tially however field evidence shows that anti-body prevalence can increase rapidly once apopulation becomes infected with E chaffeensis(Lockhart et al 1995) Utilizing per county sam-ple sizes equivalent to this study statewideserotesting of WTD from Iowa demonstrated aclear north-south gradient of seroprevalencethat corresponded to key epidemiologic factors(eg distributions of ticks deer and habitats)(Mueller-Anneling et al 2000)

Because the distributions of pathogens oftenare reflected by some combination of both sta-

YABSLEY ET AL204

tic and changing geographic distributions animportant attribute of an effective surveillancesystem is the ability to identify both persis-tently enzootic locales and spread to new loca-tions Testing of WTD from LST infested pop-ulations over the span of 20 years showed thatonce this biologically important vector is pres-ent and WTD populations are infected with Echaffeensis there is a high probability that thepopulations will remain infected and that WTDsentinels reflect these enzootic conditions Con-versely populations with no history of LST in-festation remained consistently seronegativewhen tested multiple times These results arecompatible with a strong site-specific associa-tion between E chaffeensis antibodies and thepresence of LST (Lockhart et al 1996) Of in-terest each of two populations (Choctaw CoAL and Bolivar Co MS) that were earlier re-ported as seropositive but LST negative (Lock-hart et al 1996) were shown in this study to beboth LST positive and seropositive for E chaf-feensis The rare failure to find LST in seropos-itive WTD populations may have occurred be-cause many collections were made during thefall hunting season when LST infestations areundergoing rapid seasonal declines (Allan2001)

Four instances of E chaffeensis spread to newlocations were detected by monitoring WTDfor multiple years (Table 2) At each of these lo-cations the change in population serologic sta-tus from negative to positive coincided withdocumented appearance of LST These findingsare very similar to those from a 12-year-longmonitoring of a WTD population that demon-strated the introduction of E chaffeensis was at-tributable to the establishment of LST (Lock-hart et al 1995) Because all of these changes toseropositive status were linked to the principalvector they provide further evidence that thissurveillance system reflects important epi-demiologic factors and that it should discernchanging local HME risks

In summary the biological reasons a sentinelWTD surveillance system functions effectivelyare that the force of transmission of E chaffeen-sis is focused on WTD and that aspects of WTDbiology allow attribution of infection to a spe-cific geographic location From an operationalperspective the system has the desirable at-

tributes of applicability on a wide geographicscale and logistic feasibility by using ldquofreerdquo(hunter-harvested) samples The present workwas done in a research context however it ispossible to implement sentinel WTD serologicsurveillance for E chaffeensis in an operationalpublic health context as clearly demonstratedpreviously in Iowa (Mueller-Anneling et al2000)

ACKNOWLEDGMENTS

We are indebted to over 100 individuals rep-resenting state wildlife agency personnelSCWDS staff and white-tailed deer researchersfor help in field collection of samples but es-pecially the following individuals for coordi-nating specific collections Chris Cook (AL)John Morgan (FL) Lloyd Fox (KS) Jon Gassettand David Yancy (KY) Jeff Beringer (MO)Douglas Roscoe (NJ) Evin Stanford (NC) AAlan Kocan (OK) Charles Ruth Jr (SC) BenLayton (TN) Matt Knox (VA) and Jim Crum(WV) Sarah Cross Page Luttrell and DarrellKavanaugh from SCWDS provided excellentfield and laboratory help Work was supportedprimarily by the National Institutes of Allergyand Infectious Diseases (5 R01 AI044235-02)Further support was provided by the FederalAid to Wildlife Restoration Act (50 Stat 917)and through sponsorship from fish and wildlifeagencies in Alabama Arkansas Florida Geor-gia Kansas Kentucky Louisiana MarylandMississippi Missouri North Carolina Okla-homa Puerto Rico South Carolina TennesseeVirginia and West Virginia

REFERENCES

Allan SA Ticks (Class Arachnida Order Acarina) InSamuel WM Pybus MJ and Kocan AA ed ParasiticDiseases of Wild Mammals Ames IA Iowa State Uni-versity Press 2001 72ndash106

Anderson BE Sumner JW Dawson JE et al Detection ofthe etiologic agent of human ehrlichiosis by polymerasechain reaction J Clin Microbiol 1992 30775ndash780

Anderson BE Sims KG Olson JG et al Amblyommaamericanum a potential vector of human ehrlichiosisAm J Trop Med Hyg 1993 49239ndash244

Bloemer SR Zimmerman RH Fairbanks K Abundanceattachment sites and density estimators of lone star

205EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

ticks (Acari Ixodidae) infesting white-tailed deer JMed Entomol 1988 25295ndash300

Boromisa RD Grimstand PR Seroconversion rates toJamestown Canyon virus among six populations ofwhite-tailed deer (Odocoileus virginianus) in Indiana JWildl Dis 1987 2723ndash33

Brandsma AR Little SE Lockhart JM et al NovelEhrlichia organism (Rickettsiales Ehrlichieae) in white-tailed deer associated with lone star tick (Acari Ixodi-dae) parasitism J Med Entomol 1999 36190ndash194

Childs JE Paddock CD The ascendancy of Amblyommaamericanum as a vector of pathogens affecting humansin the United States Annu Rev Entomol 2003 48307ndash337

Comer JA Nicholson WL Olson JG et al Serologic test-ing for human granulocytic ehrlichiosis at a national re-ferral center J Clin Microbiol 1999 37558ndash564

Dawson JE Childs JE Biggie KL et al White-tailed deeras a potential reservoir of Ehrlichia spp J Wildl Dis1994a 30162ndash168

Dawson JE Stallknecht D Howerth EW et al Suscep-tibility of white-tailed deer (Odocoileus virginianus) toinfection with Ehrlichia chaffeensis the etiologic agent ofhuman ehrlichiosis J Clin Microbiol 1994b 322725ndash2728

Dawson JE Anderson BE Fishbein DB et al Poly-merase chain reaction evidence of Ehrlichia chaffeensisetiologic agent of human ehrlichiosis in dogs fromsoutheast Virginia Am J Vet Res 1996a 571175ndash1179

Dawson JE Warner CK Baker V et al Ehrlichia-like 16SrDNA sequence from wild white-tailed deer (Odocoileusvirginianus) J Parasitol 1996b 8252ndash58

Davidson WR Lockhart JM Stallknecht DE et al Per-sistent Ehrlichia chaffeensis infection in white-tailed deerJ Wildl Dis 2001 37538ndash546

Dumler JS Walker DH Tick-borne ehrlichiosis LancetInfect Dis 2001 021ndash28

Ewing SA Dawson JE Kocan AA et al Experimentaltransmission of Ehrlichia chaffeensis (RickettsialesEhrlichieae) among white-tailed deer by Amblyommaamericanum (Acari Ixodidae) J Med Entomol 199532368ndash374

Fletcher WO Stallknecht DE Kearney MT et al Anti-bodies to vesicular stomatitis New Jersey type virus inwhite-tailed deer on Ossabaw Island Georgia 1985ndash1989 J Wildl Dis 1991 27675ndash680

Gill JS McLean RG Neitzil DF et al Serologic analysisof white-tailed deer sera for antibodies to Borreliaburgdorferi by enzyme-linked immunosorbent assayand western blotting J Clin Microbiol 1993 31318ndash322

Ijdo JW Wu C Magnarelli LA et al Detection ofEhrlichia chaffeensis DNA in Amblyomma americanumticks in Connecticut and Rhode Island J Clin Microbiol2000 384655ndash4656

Irving RP Pinger RR Vann CN et al Distribution ofEhrlichia chaffeensis (Rickettsiales Rickettsiaeceae) inAmblyomma americanum in southern Indiana and preva-lence of E chaffeensis-reactive antibodies in white-taileddeer in Indiana and Ohio in 1998 J Med Entomol 200037595ndash600

Keirans JE Lacombe EH First records of Amblyommaamericanum Ixodes (Ixodes) dentatus and Ixodes (Cer-atixodes) uriae (Acari Ixodidae) from Maine J Parasitol1998 84629ndash631

Little SE Dawson JE Lockhart JM et al Developmentand use of specific polymerase chain reaction for thedetection of an Ehrlichia-like organism in white-taileddeer J Wildl Dis 1997 33246ndash253

Little SE Stallknecht DE Lockhart JM et al Naturalcoinfection of a white-tailed deer (Odocoileus virgini-anus) population with three Ehrlichia spp J Parasitol1998 84897ndash901

Lockhart JM Davidson WR Dawson JE et al Tempo-ral association of Amblyomma americanumwith the pres-ence of Ehrlichia chaffeensis reactive antibodies in white-tailed deer J Wildl Dis 1995 31119ndash124

Lockhart JM Davidson WR Stallknecht DE et al Site-specific geographic association between Amblyommaamericanum (Acari Ixodidae) infestations and Ehrlichiachaffeensis-reactive (Rickettsiales Ehrlichieae) antibod-ies in white-tailed deer J Med Entomol 1996 33153ndash158

Lockhart JM Davidson WR Stallknecht DE et al Iso-lation of Ehrlichia chaffeensis from wild white-tailed deer(Odocoileus virginianus) confirms their role as naturalreservoir hosts J Clin Microbiol 1997a 351681ndash1686

Lockhart JM Davidson WR Stallknecht DE et al Nat-ural history of Ehrlichia chaffeensis (RickettsialesEhrlichieae) in the Piedmont physiographic province ofGeorgia J Parasitol 1997b 83887ndash894

Magnarelli LA Anderson JF Stafford KC et al Anti-bodies to multiple tick-borne pathogens of babesiosisehrlichiosis and Lyme borreliosis in white-footed miceJ Wildl Dis 1997 33466ndash473

Mueller-Anneling L Gilchrist MJ Thorne PS Ehrlichiachaffeensis antibodies in white-tailed deer Iowa 1994and 1996 Emerg Infect Dis 2000 6397ndash400

Murphy GL Ewing SA Whitworth LC et al A molec-ular and serologic survey of Ehrlichia canis E chaffeen-sis and E ewingii in dogs and ticks from OklahomaVet Parasitol 1998 79325ndash339

Paddock CD Childs JE Ehrlichia chaffeensis a prototyp-ical emerging pathogen Clin Microbiol Rev 2003 1637ndash64

Paddock CD Folk SM Shore GM et al Infections withEhrlichia chaffeensisand Ehrlichia ewingii in persons coin-fected with human immunodeficiency virus Clin InfectDis 2001 31586ndash1594

Roland WE Everett ED Cyr TL et al Ehrlichia chaffeensisin Missouri ticks Am J Trop Med Hyg 1998 59641ndash643

Smart DL and Trainer DO Serologic evidence ofVenezuelan equine encephalitis in some wild and do-mestic populations of southern Texas J Wildl Dis 197511195ndash200

Stallknecht DE Blue JL Rollor EA et al Precipitatingantibodies to bluetongue and epizootic hemorrhagicdisease viruses in white-tailed deer in the southeasternUnited States J Wildl Dis 1991 27238ndash247

Steiert JG Gilfoy F Infection rates of Amblyomma ameri-canum and Dermacentor variabilis by Ehrlichia chaffeensis

YABSLEY ET AL206

1994a) and experimental and field studies haveproven that WTD are competent and persistentlyinfected reservoirs (Dawson et al 1994b Lock-hart et al 1997a Davidson et al 2001 Yabsley etal 2002) The LST was initially suspected to be avector based on the geographic distribution ofHME cases and this was confirmed by poly-merase chain reaction (PCR) detection of E chaf-feensis DNA in LST (Anderson et al 1993 Rolandet al 1998 Steiert and Gilfoy 2002) and by ex-perimental transstadial transmission betweenWTD (Ewing et al 1995) In addition to being im-portant reservoirs for E chaffeensis WTD are animportant host for all three mobile stages of LST(Bloemer et al 1988)

The potential utility of WTD as a sentinel fordetermining the distribution of E chaffeensiswas first suggested when antibodies reactive toE chaffeensis were detected in WTD (Dawsonet al 1994a) and a prototype WTD serologicsurveillance system for E chaffeensis recentlywas applied statewide in Iowa (Mueller-An-neling et al 2000) White-tailed deer have beenutilized as sentinel animals to monitor numer-ous human and livestock pathogens includingJamestown Canyon virus (Boromisa and Grim-stad 1987) Venezuelan equine encephalitis(Smart and Trainer 1975) Lyme disease (Gill etal 1993) vesicular stomatitis virus (Fletcher etal 1991) and bluetongue and epizootic hem-orrhagic disease viruses (Stallknecht et al1991) The advantages associated with surveil-lance systems utilizing deer include (1) a dis-tribution that includes 45 states allowing de-velopment of national regional or state-widesurveillance systems (2) relatively high abun-dances allowing use as sentinels in most loca-tions (3) regulated harvests in all states en-hancing the ease and reducing the cost ofcollecting samples from hunter-killed animals(4) limited size of deer home ranges (mean ap-proximately 400 ha) allowing identification ofexposure location (5) a much higher rate of ex-posure to tick vectors compared to humans orpotential domestic animal sentinels (6) rela-tively long life spans increasing opportunity forexposure (7) opportunities to simultaneouslymonitor for presence of tick vectors (8) occur-rence in both natural habitats and areas proxi-mal to human habitation and (9) freedom fromprior treatment with antibiotics or acaracides

Field studies have demonstrated that serologicmonitoring of deer would allow for fine-scalemapping of the distribution of E chaffeensis(Lockhart et al 1996 Mueller-Anneling et al2000) Using a sample size of only five animalsper population Lockhart et al (1996) success-fully discriminated between LST infested popu-lations which had a 100 antibody prevalenceand populations believed to be LST-free whichhad a seroprevalence of 67 Although E chaf-feensis-reactive antibodies are reported for manyWTD populations confirmation of infection us-ing PCR andor culture has only been reportedfor 10 seropositive populations in ArkansasGeorgia Kentucky North Carolina and SouthCarolina (Lockhart et al 1997a 1997b Little et al1997 1998 Yabsley et al 2002)

The overall goal of this study was to imple-ment and evaluate an extensive fine-scale(county) surveillance system for E chaffeensisutilizing WTD as natural sentinels Specific ob-jectives to achieve this goal were to (1) combinecurrent and prior WTD serologic data to con-struct an extensive regional database on E chaf-feensis-reactive antibodies in WTD (2) use PCRassays and cell culture isolation testing to con-firm E chaffeensis infection in seropositive WTDpopulations (3) document any age class or gen-der relationships to antibody prevalence andPCR detection (4) evaluate sample sizes thatwould be functionally effective at distinguish-ing between infected and uninfected popula-tions (5) affirm that infection status of WTDpopulations corresponds with the presence ofLST and (6) investigate the ability of a WTDsurveillance system to consistently discern thedistribution or spread of E chaffeensis amongWTD populations over time

MATERIALS AND METHODS

Sample collections

White-tailed deer samples specifically forthis project were collected from sites selectedto complement and further expand four previ-ous studies where WTD in the south centraland southeastern United States were tested forE chaffeensis-reactive antibodies (Dawson et al1994a Lockhart et al 1996 Little et al 1997Yabsley et al 2002) Blood samples were col-

YABSLEY ET AL196

lected from 2101 WTD from 471 populationsin 18 states (AL AR GA FL KS KY LA MDMO MS NC NJ OK SC TN TX VA andWV) during Southeastern Cooperative WildlifeDisease Study herd health evaluation activitiesother research projects or in cooperation withstate wildlife agency personnel from hunter-harvested animals Blood samples were col-lected either aseptically from the heart fromthe jugular vein or from the body cavity Oneexception was that blood samples from 40 of79 deer from Missouri were collected on Nob-uto strips and eluted in phosphate-bufferedsaline (PBS) pH 74 as described (Mueller-An-neling et al 2000) Serum plasma or elutedblood samples were frozen at 220degC until sero-logic testing From a subset of WTD (368 from112 seropositive populations and 61 from 20seronegative populations) whole blood sam-ples were collected and frozen at 220degC forPCR testing Ticks were collected from a sec-ond subset of 715 WTD (117 populations) andsubmitted to the National Veterinary ServicesLaboratories (US Department of AgricultureAmes IA) for identification Animals werehandled and samples collected by proceduresapproved by the Animal Care and Use Com-mittee at the University of Georgia (A3437-01)

Serologic assays

Serum plasma or eluted blood sampleswere tested for antibodies reactive to E chaf-feensis by the indirect immunofluorescent anti-body (IFA) test as previously described (Daw-son et al 1994 Lockhart et al 1996) Sampleswere screened at a dilution of 1128 using Echaffeensis antigen slides (Focus TechnologiesCypress CA) A 150 dilution of fluoresceinisothiocyanate-labeled rabbit anti-deer im-munoglobulin G (Kirkegaard and Perry Labo-ratories Gaithersburg MD) was used as con-jugate Geometric mean titer (GMT) wascomputed for 152 seropositive deer

Molecular assays

DNA from 300 ml of 429 whole blood sam-ples was extracted using the GFX GenomicBlood DNA Purification Kit (Amersham Phar-macia Biotech Piscataway NJ) following themanufacturerrsquos protocol Two gene targets the

16S rRNA gene and the variable length PCRtarget (VLPT) antigen gene were utilized forscreening of whole blood for E chaffeensis DNAusing a nested PCR format Primary outsideamplification for the 16S rRNA gene consistedof 5 mL of DNA (from whole blood) in a 25-mLreaction containing 10 mM Tris-Cl (pH 83) 50mM KCl 15 mM MgCl2 02 mM each dNTP(Promega Madison WI) 25 units Taq DNAPolymerase (Promega) and 08 mM of primersECC and ECB (Dawson et al 1994b 1996a) Forthe nested PCR 1 mL of primary product wasused as template in a 25-mL reaction contain-ing the same PCR components except E chaf-feensis specific primers HE1 and HE3 (Ander-son et al 1992) were used

Nested PCR for the VLPT antigen gene wasconducted on 429 blood samples collected dur-ing this study as previously described (Sumneret al 1999) In addition 101 blood samples froman additional 10 populations previously testedusing 16S rRNA primers (Yabsley et al 2002)were tested for E chaffeensis using the VLPTgene target Amplified products were sepa-rated in 2 agarose gels stained with ethidiumbromide and visualized with UV light To con-firm identity representative secondary VLPTgene amplicons were purified with a Microconspin filter (Amicon Inc Beverley MA) se-quenced at MWG-BIOTECH (High Point NC)and compared to published E chaffeensis VLPTsequences in the GenBank database

Stringent protocols and controls were uti-lized in all PCR assays to prevent and detectcontamination DNA extraction primary am-plification secondary amplification and prod-uct analysis were performed in separate dedi-cated laboratory areas Two negative watercontrols were included in each set of DNA ex-tractions and one water control was includedin each set of primary and secondary PCR re-actions

Culture isolation

During 2001ndash2003 aseptically collectedblood samples from 72 WTD from 15 LST pos-itive populations were cultured in DH82 cellsfor isolation of E chaffeensis as previously de-scribed (Lockhart et al 1997a) Cultures werefed biweekly with MEM medium (Sigma St

197EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

Louis MO) supplemented with 10 fetalbovine serum (Sigma) and monitored for evi-dence of cytopathic effect or for a maximum of45 days Cultures showing cytopathic effectand cultures completing the 45-day incubationperiod were harvested with a cell scraper andtested by direct fluorescent antibody assay aspreviously described (Lockhart et al 1997a)

Data analysis

Serologic data from this study (n 5 2101)were combined with data from previous stud-ies conducted independently by SCWDS re-searchers (n 5 425 Lockhart et al 1996 Littleet al 1997 Yabsley et al 2002) or collabora-tively with others (n 5 749 Dawson et al1994a) to construct a comprehensive map delineating the distribution of E chaffeensisseroreactive WTD in the southeastern andsouth central United States Some of the previ-ous studies used different cutoff values for IFAtesting (eg 164 or 1128) but in order to stan-dardize results from this study and previousstudies IFA assays were classified as positiveonly if a sample had a titer of $1128 To facil-itate graphic presentation data for each popu-lation were categorized by county or parish ifone or more deer with antibodies reactive to Echaffeensis was detected that county or parishwas classified as seropositive

To confirm serologic data PCR andor cul-ture isolation were conducted on a total of 541deer from 122 seropositive populations and on61 deer from 20 seronegative populations Chi-square analysis was used to test for differencesof PCR prevalence between seropositive andseronegative populations

Prevalence of antibodies reactive to E chaf-feensis among different age classes and gendercategories were only determined for WTD inseropositive populations Serologic data fromDawson et al (1994) Lockhart et al (1996) Lit-tle et al (1997) Yabsley et al (2002) and thisstudy were combined for determining preva-lence among age classes because analysis ofWTD age data had not been previously con-ducted PCR testing was not conducted byDawson et al (1994) and Lockhart et al (1996)therefore analysis of age and PCR positivitywas conducted only on deer from Little et al(1997) Yabsley et al (2002) and this study Chi

square analysis was used to test for differencesin both seroprevalence and PCR positivityamong age classes

To accurately classify the serostatus of a pop-ulation an appropriate sample size needs to betested In previous studies testing of five deerfrom a population consistently detected E chaf-feensisndashreactive antibodies in deer populationsinfested with LST (Lockhart et al 1997 Little etal 1997 1998) and based on Dawson et al(1994) testing a mean of 21 deer failed to detectseroreactive deer in seronegative populationsHowever additional evaluation of the samplesizes required to reliably classify the serologicstatus of populations was conducted in thisstudy To determine minimal number of sam-ples needed to detect the presence of a seropos-itive deer in a population the formula n 5(1 2 (1-a)1D) (N 2 (D 2 1)2) was used as de-scribed (Thrusfield 1995) Because seronegativepopulations are more difficult to accuratelyclassify two methods were used to examine ad-equacy of sample size Larger numbers of deer(n 5 8 2 16) were tested from nine seronega-tive populations to enhance detection of po-tentially low prevalences in those populationsAlso four populations with no history of LSTinfestation were repeatedly tested for severalyears

To assess the predicted epidemiologic asso-ciation between LST infestation and serologicstatus of WTD populations chi-square analy-sis was used to test for an association betweenLST presence and seroreactors at the popula-tion level Furthermore 11 populations forwhich both tick and serologic data were avail-able were evaluated during multiple yearsthree populations had LST infestation eachsample year four never had any evidence ofLST infestation and four were negative for LSTduring earlier sampling period (s) but becameLST positive during subsequent test years Chi-square analysis was used to test for differencesbetween groups

RESULTS

Regional serologic database for WTD

Among the 2101 WTD collected specificallyfor this study 984 (468 CI95 5 447

YABSLEY ET AL198

489) had antibodies reactive ($1128 titer) toE chaffeensis by IFA testing (Table 1) The meanprevalence of antibodies reactive to E chaffeen-sis in seropositive populations was 738(SD 5 258 range 20ndash100) White-tailed deerwith antibodies reactive to E chaffeensis weredetected in all states tested except West Vir-ginia The modal antibody titer for seropositivedeer was 1128 with a maximum titer of 14096The GMT of a subset of 152 seropositive WTDwas 355

The combined dataset of WTD serologicallytested during this and four prior studies (Daw-son et al 1994a Lockhart et al 1996 Little etal 1997 Yabsley et al 2002) contained a totalof 3275 WTD from 18 southeastern and southcentral states In the combined data set 1549(473) WTD examined had antibodies reactiveto E chaffeensis with seropositive populationsdetected in 17 of 18 states (Fig 1)

The highest overall seroprevalences were inArkansas and Missouri Although high preva-lences were detected in local WTD populationsin most states considerable variation in preva-lence occurred within different regions of somestates (eg Florida Kansas North Carolina

Oklahoma and Virginia) A western boundaryfor the distribution of E chaffeensis was evidentspanning the states of Kansas Oklahoma andTexas Similarly a southern boundary wasnoted across peninsular Florida Clusters ofseronegative populations were detected insouthern Mississippi and Louisiana along thelower Mississippi River floodplain and in theAppalachian Mountains from western Mary-land to northern Alabama

PCR and culture validation of E chaffeensisserologic data

Eighty-six of 368 (234) WTD from 50seropositive populations were PCR positive forE chaffeensis by either 16S or VLPT PCR Of 101WTD previously tested by 16S PCR (Yabsley etal 2002) nine deer from seven populationswere positive by VLPT PCR Collectively forthis study and Yabsley et al (2002) 95 of 469(203) WTD from 56 of 122 (46) seropositivepopulations have been confirmed using PCR(Table 1) 70 of 95 deer with both gene targets2 of 95 for only the 16S rRNA gene and 23 of95 only for the VLPT gene In contrast 16S

199

TABLE 1 RESULTS OF SEROLOGIC AND POLYMERASE CHAIN REACTION TESTING

FOR EHRLICHIA CHAFFEENSIS IN WHITE-TAILED DEER FROM 18 STATESa

Counties Number IFA Number PCRtested by positive positive

State Years IFA test no tested () no tested ()

Alabama 1981ndash2001 33 59141 (42) NTb

Arkansas 1982ndash2002 29 111154 (72) 746 (15)Florida 1983ndash2002 15 43104 (40) 936 (25)Georgia 1973ndash2002 50 119243 (49) 2698 (27)Kansas 1998ndash2002 38 3976 (51) 326 (12)Kentucky 1983ndash2001 27 2699 (26) 223 (9)Louisiana 1981ndash2001 34 54166 (33) 118 (6)Maryland 1999ndash2002 5 1225 (48) 17 (14)Missouri 1994ndash2002 19 5779 (72) 323 (13)Mississippi 1995ndash2001 38 80152 (53) 119 (5)North Carolina 1982ndash2002 33 75131 (57) 1572 (21)New Jersey 2001 2 610 (60) 05Oklahoma 1985ndash2002 13 3050 (60) 03South Carolina 1991ndash2001 26 52124 (42) 460 (7)Tennessee 1981ndash2001 20 54100 (54) 010Texas 1992ndash2002 33 73123 (59) 013Virginia 1983ndash2002 37 94168 (56) 2357 (40)West Virginia 1977ndash2001 18 0156 014Total 471 9842101 (47) 95530 (18)

aIncludes PCR results for the VLPT antigen gene for 101 WTD previously tested by PCR for the 16S rRNA gene(Yabsley et al 2002)NT not tested

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

rRNA and VLPT PCR assays on 61 deer from20 seronegative populations were all negativePresence of PCR positive deer was strongly as-sociated with seropositive populations (x2 5116 p 0001)

Of 72 culture attempts 34 (442) were lostto contamination with bacteria andor Try-panosoma cervi Seven of the remaining 38 cul-tures (184) were positive for E chaffeensisPositive cultures were obtained from two WTDfrom Greene Co AR two from Woodruff CoAR and one each from Clarke Co GA AnsonCo NC and Georgetown Co SC All isolateswere confirmed as E chaffeensis by PCR of the16S rRNA and VLPT gene targets

Relationship of age class and gender to antibody prevalence and PCR detection

Age was available on 1882 deer tested sero-logically 1086 deer from this study 517 fromDawson et al (1994) 165 from Lockhart et al

(1996) five from Little et al (1997) and 109from Yabsley et al (2002) For this combineddata set differences were not noted (x2 5 565p 5 046) in seroprevalance among age classes(Fig 2A) Similarly seroprevalence did not dif-fer between males and females (693 vs696 x2 5 0016 p 005)

Age was available for 393 deer from seropos-itive populations that were tested by PCRHigher proportions of deer were PCR positivein the younger age classes (x2 5 717 p 0001) Among deer 075 years old 531were PCR positive within the next age group(076ndash15 years) the prevalence decreased sig-nificantly to 27 (x2 5 147 p 0001) and lessthan 8 of deer 15 years old were PCR pos-itive (Fig 2B) The mean age of PCR positiveWTD was 11 years (range 025ndash4 years) In-sufficient culture isolates were made to exam-ine age-related associations however themean age for culture positive deer was 32years (range 03ndash95 years) Similar to antibody

YABSLEY ET AL200

FIG 1 Indirect immunofluorescent antibody results for Ehrlichia chaffeensis in white-tailed deer Results are from thecurrent study and from previously published work obtained in collaboration with SCWDS Darkly shaded counties pos-itive and lightly shaded counties negative for antibodies ($1128) reactive for E chaffeensis Previous serologic resultsthat differed from the results of the current study are shown as small circles within the county d previously positivefor antibodies ($1128) reactive for E chaffeensis s previously negative for antibodies reactive for E chaffeensis

prevalence PCR positivity did not differ be-tween males and females (19 vs 228 x2 5089 p 005)

Evaluation of sample size adequacy

Regarding sample size for seropositive pop-ulations post hoc statistical analysis utilizing amean 73 seroprevalence indicated that test-ing of only three or four deer gives a high prob-ability (95 and 99 respectively) of detectinga positive population (Thrusfield 1995) Al-though validation of seronegative populationsis more problematic testing of larger numbersof deer per population and repeated testing ofselected populations both produced consistentnegative results In a combined dataset fromDawson et al (1994a) and this study 250 deerfrom 17 seronegative populations (9ndash30 deer

per population) were negative for antibodies toE chaffeensis with means of 208 and 104 deertested per seronegative population respec-tively Similarly none of the 131 deer from fourpopulations consistently negative for LST in-festation had antibodies reactive to E chaffeen-sis The mean number of WTD tested per pop-ulation in this study was 62 (SD 5 61 range1ndash58) and data for populations with samplesizes less than four deer were used only if aseropositive animal was detected

Association between LST and serologic status at population level

Of 117 WTD populations examined for ticks82 were infested with LST 29 with A macula-tum (Gulf Coast tick) four with Ixodes affinisand 13 with I scapularis (blacklegged tick)

201

FIG 2 (A) Prevalence of antibodies reactive to E chaffeensis in WTD among age classes Numbers in bars representnumber of deer tested (B) Prevalence of PCR-positive WTD among age classes Numbers in bars represent numberof deer tested and different letters represent significant differences at p 005

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

Eighty of the 82 (976) populations infestedwith LST contained at least one seropositiveWTD while only four of 35 (114) populationsnot infested with LST contained seropositivedeer One LST-negative population (Harris CoGA) was both seronegative and tick negativein 1986 however in 2002 the deer populationwas seropositive but the LST status was notevaluated The prevalence of antibodies reac-tive to E chaffeensis was significantly higheramong LST infested populations of WTD thanpopulations not infested by LST (x2 5 899 p 0001) None of the six populations infestedonly with A maculatum contained seropositivedeer there were insufficient populations withI affinis or I scapularis alone to permit evalua-tion

Stability and spread of E chaffeensis and LST among WTD populations

Fifty-seven populations which were testedduring previous studies (Dawson et al 1994aLockhart et al 1996) were retested during thisstudy The serologic status for 50 (88) of thesepopulations remained unchanged 35 seropos-itive populations remained seropositive and 15seronegative populations remained seronega-tive Six populations which previously testedseronegative were seropositive during this

study Three of these representing Anson CoNC Haywood Co TN and Stewart Co GAhad documented invasion of LST between thefirst reported testing and this study One Texaspopulation (Travis Co) was seropositive (n 57 714) when tested in 1992 (Dawson et al1994a) but was seronegative when sampled(n 5 10) in 2002 for this study The LST statusfor this county was not known for either sam-pling period

Among the 11 populations where LST infes-tation and serologic status were monitored repeatedly over time there was complete con-cordance between LST presence and seroposi-tive deer (Table 2) All three LST positive pop-ulations were consistently seropositive eachyear tested The mean seroprevalence for thesethree populations was 83 and two of thesepopulations (Jones Co GA and Ashley CoAR) were confirmed PCR positive for E chaf-feensis None of the four populations where LSTwas consistently absent contained seropositivedeer at any sampling time Among the fourpopulations where LST status changed fromabsent to present seropositive deer were onlydetected following the appearance of LST Themean prevalence of antibodies reactive to Echaffeensis in the most recently tested year wassignificantly higher for populations infestedwith LST for multiple years (mean 5 100)

YABSLEY ET AL202

TABLE 2 SEROLOGIC AND TICK SURVEY RESULTS FROM SEVEN POPULATIONS OF WTD COLLECTED DURING

MULTIPLE TIME PERIODS AND TESTED FOR ANTIBODIES REACTIVE FOR E CHAFFEENSIS

No of Tickyears status No of deer Mean IFA prevalence

Site Years sampled (years) tested (range)

Sites positive for ticksJones Co GA 1979ndash2001 6 1 (6) 41 92 (80ndash100)Ashley Co AR 1982ndash2001 7 1 (7) 37 84 (71ndash100)Franklin Co FL 1984ndash1996 5 1 (5) 25 68 (40ndash100)

Sites negative for ticksFloyd Co GA 1973ndash2001 6 2 (6) 58 0Tyler Co WV 1985ndash2000 4 2 (4) 21 0Hardy Co WV 1977ndash1999 7 2 (7) 37 0Monroe Co FL 1992ndash2001 4 2 (4) 15 0

Sites where tick status changedAnson Co NC 1987 2001 2 2 1a 10 0 40bConcordia Pa LA 1986 1991 1999 3 2 1 1 22 0 38 60Haywood Co TN 1989 1994 1998 3 2 2 1 16 0 0 20Stewart Co GA 1986 1998 2 2 1 10 0 20

aTick status at sequential sampling periodsbPrevalence of IFA seropositive deer at sequential sampling periods

compared to populations only recently infested(mean 5 538 x2 5 155 p 0001) One re-cently infested population (Anson Co NC)was confirmed by isolation of E chaffeensisfrom a single WTD in 2001

DISCUSSION

The overarching goal of this study was toimplement and evaluate a prototype surveil-lance system using WTD as natural sentinels todetermine the geographic distribution of Echaffeensis across most of the suspected HMEendemic region of the United States To thisend the data obtained demonstrate the fol-lowing critical attributes (1) serologic findingsreflect E chaffeensis infection (2) effective sur-veillance can be achieved with small samplesizes (3) any deer 6 month old is suitable forsurveillance purposes (4) enzootic and consis-tently negative locales were identified repeat-edly across time (5) surveillance of deer hasthe ability to detect spread to new locales and(6) serologic molecular and culture diagnosticfindings among deer could be related to pres-ence of the principal tick vector A americanumThese findings compliment and expand on arecent WTD serologic surveillance effort withinIowa (Mueller-Anneling et al 2000) howeverthat study exclusively used serologic assaysand did not include either confirmatory mi-crobiological diagnostics or contemporaneoustick vector monitoring Collectively these twostudies confirm the effectiveness of this proto-typical surveillance system for monitoring thedistribution of this emerging human pathogen

Although several studies have examined theprevalence and distribution of E chaffeensis an-tibodies among many southeastern and southcentral WTD populations (Dawson et al 1994aLockhart et al 1996 Little et al 1997 Yabsleyet al 2002) the extensive additional testingfilled many geographic gaps and provide amore complete fine scale distribution of E chaf-feensis among WTD throughout the regions ofthe United States with the highest HME risk(Paddock and Childs 2003) The combined dataevaluated herein disclosed a wide distributionof seropositive WTD populations across all ofthe 18 states represented except for West Vir-

ginia however some distributional limits alsowere discerned including a western boundaryextending across Kansas Oklahoma andTexas a southern boundary across peninsularFlorida and a large cluster of many seronega-tive populations centered along the Ap-palachian Mountains (Fig 1) Other prior stud-ies of WTD have documented an apparentnorthern range limit for E chaffeensis in south-ern portions of Iowa Illinois Indiana and Ohio(Dawson et al 1994a Irving et al 2000 Mueller-Anneling et al 2000) Along the East coast sur-veys have indicated that A americanum and Echaffeensis are present as far north as Connecti-cut Maine and Rhode Island (Keirans and La-combe 1998 Ijdo et al 2000)

The accuracy of diagnostic assays utilized iscritical to any pathogen surveillance system In-fection with E chaffeensis was confirmed innearly half (46) of 122 seropositive WTD pop-ulations tested by PCR and was supplementedby isolation from others In contrast PCR evi-dence of infection was not detected in any of20 seronegative WTD populations Thus PCRand culture testing applied at the populationlevel were effective at confirming infection inmany seropositive populations and demon-strating a strong association between the sero-logic status of populations and infection withE chaffeensis

Because WTD in these regions are known to be infected with three other related rickettsiae(E ewingii A phagocytophilum and an Ana-plasma sp [WTD-agent]) in addition to E chaf-feensis (Dawson et al 1996b Little et al 1998Brandsma et al 1999 Magnarelli et al 1999Yabsley et al 2002) the potential for serologiccross-reaction is an important considerationAlthough antibodies reactive with E chaffeensisin WTD from Maryland have been confirmedby immunoblotting using the 28- to 29-kDa anti-gens of E chaffeensis (Walls et al 1998) the ex-tent of serologic cross-reactions among thesefour species is not fully understood (Dumlerand Walker 2001) However data from otherstudies suggests that infection with these otherrickettsiae may not commonly result in pro-duction of antibodies that cross-react with Echaffeensis Cross-reactions between E chaffeen-sis and E ewingii have been noted in infectedhumans and dogs however not all E ewingii

203EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

infected dogs or humans develop antibodies toE chaffeensis antigens (Murphy et al 1998 Pad-dock et al 2001) Similarly a WTD fawn ex-perimentally infected with E ewingii did not de-velop antibodies reactive with E chaffeensis(Yabsley et al 2002) Cross-reactions between Echaffeensis and A phagocytophilum have been re-ported in humans and may complicate diagno-sis (Comer et al 1999) Far fewer WTD in thesoutheastern states have antibodies reactive toA phagocytophilum (25) than to E chaffeensisantigens however although some deer reactedto both E chaffeensis and A phagocytophilum ator above the 1128 cut-off many also reacted toonly one of these antigens at titers $1128 (Lit-tle et al 1998 Walls et al 1998 VG Dugan andMJ Yabsley unpublished data) The high per-centage of populations where seropositive orseronegative status was confirmed by PCR orculture together with data from these otherstudies provide considerable evidence that theseroreactivity reported herein largely repre-sents specific seroconversion to E chaffeensis

The distribution of antibodies within WTDage and gender categories was investigated be-cause these variables are an important consid-eration for a WTD natural sentinel systemOnly a limited investigation of geometric meantiters among age categories has been reportedfor WTD (Lockhart et al 1995) and any signif-icant influence of age or gender on the occur-rence of antibodies would require samplingstratified according to these host attributes Im-portantly differences in antibody prevalencewere not noted among age or gender categoriesindicating that all WTD particularly animals $6 mo of age are suitable for use in a WTD sur-veillance system In contrast to the high stableprevalence of antibodies among age classes(Fig 2A) PCR evidence of rickettsemia de-clined with age (Fig 2B) This pattern conformsto proposed infection dynamics in naturally in-fected WTD whereby animals from approxi-mately 6 months to 15 years of age are morelikely to be rickettsemic than older adults (Pad-dock and Childs 2003 Dawson et al 1994Lockhart et al 1997a Davidson et al 2001) Inthe present study the prevalence of PCR pos-itivity dropped dramatically from over half(53) of WTD younger than 075 years to lessthan 8 of WTD older than 15 years The ma-

jority of deer tested from some seropositivepopulations were 15 years and only deer15 years from other populations were avail-able for PCR testing which may explain whynot all seropositive populations were con-firmed by PCR Because the probability of aWTD being rickettsemic declines with ageserology represents a better surveillance toolthan PCR and under natural conditions of re-exposure to ticks titers likely do not declineover time as often occurred in single exposureexperimental infections of WTD (Dawson et al1994b Ewing et al 1995 Davidson et al 2001)

Previous serologic surveys of E chaffeensis inWTD (Dawson et al 1994a Lockhart et al 1996Little et al 1997 Yabsley et al 2002) have uti-lized relatively small sample sizes per popula-tion but except for one (Mueller-Anneling etal 2000) have not addressed the adequacy ofthese sample sizes to correctly classify infectedand uninfected populations Data from the cur-rent and prior studies (Dawson et al 1994aLockhart et al 1996 Little et al 1997) haveshown that infected WTD populations in thesoutheastern and southcentral United Statestypically have high (70) seroprevalencesCalculations (Thrusfield 1995) based on a mean73 seroprevalence indicated that use of a sam-ple size of five WTD per population should re-liably detect nearly all positive populations Al-though it was beyond the scope of this studyto test enough deer to detect very low popula-tion prevalences larger sample sizes and re-peated sampling of seronegative populationsdid not reveal seropositive populations thatwould have been misclassified as seronegativebased on smaller sample sizes A chance of mis-classifying newly infected populations of WTDexists because seroprevalence may be low ini-tially however field evidence shows that anti-body prevalence can increase rapidly once apopulation becomes infected with E chaffeensis(Lockhart et al 1995) Utilizing per county sam-ple sizes equivalent to this study statewideserotesting of WTD from Iowa demonstrated aclear north-south gradient of seroprevalencethat corresponded to key epidemiologic factors(eg distributions of ticks deer and habitats)(Mueller-Anneling et al 2000)

Because the distributions of pathogens oftenare reflected by some combination of both sta-

YABSLEY ET AL204

tic and changing geographic distributions animportant attribute of an effective surveillancesystem is the ability to identify both persis-tently enzootic locales and spread to new loca-tions Testing of WTD from LST infested pop-ulations over the span of 20 years showed thatonce this biologically important vector is pres-ent and WTD populations are infected with Echaffeensis there is a high probability that thepopulations will remain infected and that WTDsentinels reflect these enzootic conditions Con-versely populations with no history of LST in-festation remained consistently seronegativewhen tested multiple times These results arecompatible with a strong site-specific associa-tion between E chaffeensis antibodies and thepresence of LST (Lockhart et al 1996) Of in-terest each of two populations (Choctaw CoAL and Bolivar Co MS) that were earlier re-ported as seropositive but LST negative (Lock-hart et al 1996) were shown in this study to beboth LST positive and seropositive for E chaf-feensis The rare failure to find LST in seropos-itive WTD populations may have occurred be-cause many collections were made during thefall hunting season when LST infestations areundergoing rapid seasonal declines (Allan2001)

Four instances of E chaffeensis spread to newlocations were detected by monitoring WTDfor multiple years (Table 2) At each of these lo-cations the change in population serologic sta-tus from negative to positive coincided withdocumented appearance of LST These findingsare very similar to those from a 12-year-longmonitoring of a WTD population that demon-strated the introduction of E chaffeensis was at-tributable to the establishment of LST (Lock-hart et al 1995) Because all of these changes toseropositive status were linked to the principalvector they provide further evidence that thissurveillance system reflects important epi-demiologic factors and that it should discernchanging local HME risks

In summary the biological reasons a sentinelWTD surveillance system functions effectivelyare that the force of transmission of E chaffeen-sis is focused on WTD and that aspects of WTDbiology allow attribution of infection to a spe-cific geographic location From an operationalperspective the system has the desirable at-

tributes of applicability on a wide geographicscale and logistic feasibility by using ldquofreerdquo(hunter-harvested) samples The present workwas done in a research context however it ispossible to implement sentinel WTD serologicsurveillance for E chaffeensis in an operationalpublic health context as clearly demonstratedpreviously in Iowa (Mueller-Anneling et al2000)

ACKNOWLEDGMENTS

We are indebted to over 100 individuals rep-resenting state wildlife agency personnelSCWDS staff and white-tailed deer researchersfor help in field collection of samples but es-pecially the following individuals for coordi-nating specific collections Chris Cook (AL)John Morgan (FL) Lloyd Fox (KS) Jon Gassettand David Yancy (KY) Jeff Beringer (MO)Douglas Roscoe (NJ) Evin Stanford (NC) AAlan Kocan (OK) Charles Ruth Jr (SC) BenLayton (TN) Matt Knox (VA) and Jim Crum(WV) Sarah Cross Page Luttrell and DarrellKavanaugh from SCWDS provided excellentfield and laboratory help Work was supportedprimarily by the National Institutes of Allergyand Infectious Diseases (5 R01 AI044235-02)Further support was provided by the FederalAid to Wildlife Restoration Act (50 Stat 917)and through sponsorship from fish and wildlifeagencies in Alabama Arkansas Florida Geor-gia Kansas Kentucky Louisiana MarylandMississippi Missouri North Carolina Okla-homa Puerto Rico South Carolina TennesseeVirginia and West Virginia

REFERENCES

Allan SA Ticks (Class Arachnida Order Acarina) InSamuel WM Pybus MJ and Kocan AA ed ParasiticDiseases of Wild Mammals Ames IA Iowa State Uni-versity Press 2001 72ndash106

Anderson BE Sumner JW Dawson JE et al Detection ofthe etiologic agent of human ehrlichiosis by polymerasechain reaction J Clin Microbiol 1992 30775ndash780

Anderson BE Sims KG Olson JG et al Amblyommaamericanum a potential vector of human ehrlichiosisAm J Trop Med Hyg 1993 49239ndash244

Bloemer SR Zimmerman RH Fairbanks K Abundanceattachment sites and density estimators of lone star

205EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

ticks (Acari Ixodidae) infesting white-tailed deer JMed Entomol 1988 25295ndash300

Boromisa RD Grimstand PR Seroconversion rates toJamestown Canyon virus among six populations ofwhite-tailed deer (Odocoileus virginianus) in Indiana JWildl Dis 1987 2723ndash33

Brandsma AR Little SE Lockhart JM et al NovelEhrlichia organism (Rickettsiales Ehrlichieae) in white-tailed deer associated with lone star tick (Acari Ixodi-dae) parasitism J Med Entomol 1999 36190ndash194

Childs JE Paddock CD The ascendancy of Amblyommaamericanum as a vector of pathogens affecting humansin the United States Annu Rev Entomol 2003 48307ndash337

Comer JA Nicholson WL Olson JG et al Serologic test-ing for human granulocytic ehrlichiosis at a national re-ferral center J Clin Microbiol 1999 37558ndash564

Dawson JE Childs JE Biggie KL et al White-tailed deeras a potential reservoir of Ehrlichia spp J Wildl Dis1994a 30162ndash168

Dawson JE Stallknecht D Howerth EW et al Suscep-tibility of white-tailed deer (Odocoileus virginianus) toinfection with Ehrlichia chaffeensis the etiologic agent ofhuman ehrlichiosis J Clin Microbiol 1994b 322725ndash2728

Dawson JE Anderson BE Fishbein DB et al Poly-merase chain reaction evidence of Ehrlichia chaffeensisetiologic agent of human ehrlichiosis in dogs fromsoutheast Virginia Am J Vet Res 1996a 571175ndash1179

Dawson JE Warner CK Baker V et al Ehrlichia-like 16SrDNA sequence from wild white-tailed deer (Odocoileusvirginianus) J Parasitol 1996b 8252ndash58

Davidson WR Lockhart JM Stallknecht DE et al Per-sistent Ehrlichia chaffeensis infection in white-tailed deerJ Wildl Dis 2001 37538ndash546

Dumler JS Walker DH Tick-borne ehrlichiosis LancetInfect Dis 2001 021ndash28

Ewing SA Dawson JE Kocan AA et al Experimentaltransmission of Ehrlichia chaffeensis (RickettsialesEhrlichieae) among white-tailed deer by Amblyommaamericanum (Acari Ixodidae) J Med Entomol 199532368ndash374

Fletcher WO Stallknecht DE Kearney MT et al Anti-bodies to vesicular stomatitis New Jersey type virus inwhite-tailed deer on Ossabaw Island Georgia 1985ndash1989 J Wildl Dis 1991 27675ndash680

Gill JS McLean RG Neitzil DF et al Serologic analysisof white-tailed deer sera for antibodies to Borreliaburgdorferi by enzyme-linked immunosorbent assayand western blotting J Clin Microbiol 1993 31318ndash322

Ijdo JW Wu C Magnarelli LA et al Detection ofEhrlichia chaffeensis DNA in Amblyomma americanumticks in Connecticut and Rhode Island J Clin Microbiol2000 384655ndash4656

Irving RP Pinger RR Vann CN et al Distribution ofEhrlichia chaffeensis (Rickettsiales Rickettsiaeceae) inAmblyomma americanum in southern Indiana and preva-lence of E chaffeensis-reactive antibodies in white-taileddeer in Indiana and Ohio in 1998 J Med Entomol 200037595ndash600

Keirans JE Lacombe EH First records of Amblyommaamericanum Ixodes (Ixodes) dentatus and Ixodes (Cer-atixodes) uriae (Acari Ixodidae) from Maine J Parasitol1998 84629ndash631

Little SE Dawson JE Lockhart JM et al Developmentand use of specific polymerase chain reaction for thedetection of an Ehrlichia-like organism in white-taileddeer J Wildl Dis 1997 33246ndash253

Little SE Stallknecht DE Lockhart JM et al Naturalcoinfection of a white-tailed deer (Odocoileus virgini-anus) population with three Ehrlichia spp J Parasitol1998 84897ndash901

Lockhart JM Davidson WR Dawson JE et al Tempo-ral association of Amblyomma americanumwith the pres-ence of Ehrlichia chaffeensis reactive antibodies in white-tailed deer J Wildl Dis 1995 31119ndash124

Lockhart JM Davidson WR Stallknecht DE et al Site-specific geographic association between Amblyommaamericanum (Acari Ixodidae) infestations and Ehrlichiachaffeensis-reactive (Rickettsiales Ehrlichieae) antibod-ies in white-tailed deer J Med Entomol 1996 33153ndash158

Lockhart JM Davidson WR Stallknecht DE et al Iso-lation of Ehrlichia chaffeensis from wild white-tailed deer(Odocoileus virginianus) confirms their role as naturalreservoir hosts J Clin Microbiol 1997a 351681ndash1686

Lockhart JM Davidson WR Stallknecht DE et al Nat-ural history of Ehrlichia chaffeensis (RickettsialesEhrlichieae) in the Piedmont physiographic province ofGeorgia J Parasitol 1997b 83887ndash894

Magnarelli LA Anderson JF Stafford KC et al Anti-bodies to multiple tick-borne pathogens of babesiosisehrlichiosis and Lyme borreliosis in white-footed miceJ Wildl Dis 1997 33466ndash473

Mueller-Anneling L Gilchrist MJ Thorne PS Ehrlichiachaffeensis antibodies in white-tailed deer Iowa 1994and 1996 Emerg Infect Dis 2000 6397ndash400

Murphy GL Ewing SA Whitworth LC et al A molec-ular and serologic survey of Ehrlichia canis E chaffeen-sis and E ewingii in dogs and ticks from OklahomaVet Parasitol 1998 79325ndash339

Paddock CD Childs JE Ehrlichia chaffeensis a prototyp-ical emerging pathogen Clin Microbiol Rev 2003 1637ndash64

Paddock CD Folk SM Shore GM et al Infections withEhrlichia chaffeensisand Ehrlichia ewingii in persons coin-fected with human immunodeficiency virus Clin InfectDis 2001 31586ndash1594

Roland WE Everett ED Cyr TL et al Ehrlichia chaffeensisin Missouri ticks Am J Trop Med Hyg 1998 59641ndash643

Smart DL and Trainer DO Serologic evidence ofVenezuelan equine encephalitis in some wild and do-mestic populations of southern Texas J Wildl Dis 197511195ndash200

Stallknecht DE Blue JL Rollor EA et al Precipitatingantibodies to bluetongue and epizootic hemorrhagicdisease viruses in white-tailed deer in the southeasternUnited States J Wildl Dis 1991 27238ndash247

Steiert JG Gilfoy F Infection rates of Amblyomma ameri-canum and Dermacentor variabilis by Ehrlichia chaffeensis

YABSLEY ET AL206

lected from 2101 WTD from 471 populationsin 18 states (AL AR GA FL KS KY LA MDMO MS NC NJ OK SC TN TX VA andWV) during Southeastern Cooperative WildlifeDisease Study herd health evaluation activitiesother research projects or in cooperation withstate wildlife agency personnel from hunter-harvested animals Blood samples were col-lected either aseptically from the heart fromthe jugular vein or from the body cavity Oneexception was that blood samples from 40 of79 deer from Missouri were collected on Nob-uto strips and eluted in phosphate-bufferedsaline (PBS) pH 74 as described (Mueller-An-neling et al 2000) Serum plasma or elutedblood samples were frozen at 220degC until sero-logic testing From a subset of WTD (368 from112 seropositive populations and 61 from 20seronegative populations) whole blood sam-ples were collected and frozen at 220degC forPCR testing Ticks were collected from a sec-ond subset of 715 WTD (117 populations) andsubmitted to the National Veterinary ServicesLaboratories (US Department of AgricultureAmes IA) for identification Animals werehandled and samples collected by proceduresapproved by the Animal Care and Use Com-mittee at the University of Georgia (A3437-01)

Serologic assays

Serum plasma or eluted blood sampleswere tested for antibodies reactive to E chaf-feensis by the indirect immunofluorescent anti-body (IFA) test as previously described (Daw-son et al 1994 Lockhart et al 1996) Sampleswere screened at a dilution of 1128 using Echaffeensis antigen slides (Focus TechnologiesCypress CA) A 150 dilution of fluoresceinisothiocyanate-labeled rabbit anti-deer im-munoglobulin G (Kirkegaard and Perry Labo-ratories Gaithersburg MD) was used as con-jugate Geometric mean titer (GMT) wascomputed for 152 seropositive deer

Molecular assays

DNA from 300 ml of 429 whole blood sam-ples was extracted using the GFX GenomicBlood DNA Purification Kit (Amersham Phar-macia Biotech Piscataway NJ) following themanufacturerrsquos protocol Two gene targets the

16S rRNA gene and the variable length PCRtarget (VLPT) antigen gene were utilized forscreening of whole blood for E chaffeensis DNAusing a nested PCR format Primary outsideamplification for the 16S rRNA gene consistedof 5 mL of DNA (from whole blood) in a 25-mLreaction containing 10 mM Tris-Cl (pH 83) 50mM KCl 15 mM MgCl2 02 mM each dNTP(Promega Madison WI) 25 units Taq DNAPolymerase (Promega) and 08 mM of primersECC and ECB (Dawson et al 1994b 1996a) Forthe nested PCR 1 mL of primary product wasused as template in a 25-mL reaction contain-ing the same PCR components except E chaf-feensis specific primers HE1 and HE3 (Ander-son et al 1992) were used

Nested PCR for the VLPT antigen gene wasconducted on 429 blood samples collected dur-ing this study as previously described (Sumneret al 1999) In addition 101 blood samples froman additional 10 populations previously testedusing 16S rRNA primers (Yabsley et al 2002)were tested for E chaffeensis using the VLPTgene target Amplified products were sepa-rated in 2 agarose gels stained with ethidiumbromide and visualized with UV light To con-firm identity representative secondary VLPTgene amplicons were purified with a Microconspin filter (Amicon Inc Beverley MA) se-quenced at MWG-BIOTECH (High Point NC)and compared to published E chaffeensis VLPTsequences in the GenBank database

Stringent protocols and controls were uti-lized in all PCR assays to prevent and detectcontamination DNA extraction primary am-plification secondary amplification and prod-uct analysis were performed in separate dedi-cated laboratory areas Two negative watercontrols were included in each set of DNA ex-tractions and one water control was includedin each set of primary and secondary PCR re-actions

Culture isolation

During 2001ndash2003 aseptically collectedblood samples from 72 WTD from 15 LST pos-itive populations were cultured in DH82 cellsfor isolation of E chaffeensis as previously de-scribed (Lockhart et al 1997a) Cultures werefed biweekly with MEM medium (Sigma St

197EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

Louis MO) supplemented with 10 fetalbovine serum (Sigma) and monitored for evi-dence of cytopathic effect or for a maximum of45 days Cultures showing cytopathic effectand cultures completing the 45-day incubationperiod were harvested with a cell scraper andtested by direct fluorescent antibody assay aspreviously described (Lockhart et al 1997a)

Data analysis

Serologic data from this study (n 5 2101)were combined with data from previous stud-ies conducted independently by SCWDS re-searchers (n 5 425 Lockhart et al 1996 Littleet al 1997 Yabsley et al 2002) or collabora-tively with others (n 5 749 Dawson et al1994a) to construct a comprehensive map delineating the distribution of E chaffeensisseroreactive WTD in the southeastern andsouth central United States Some of the previ-ous studies used different cutoff values for IFAtesting (eg 164 or 1128) but in order to stan-dardize results from this study and previousstudies IFA assays were classified as positiveonly if a sample had a titer of $1128 To facil-itate graphic presentation data for each popu-lation were categorized by county or parish ifone or more deer with antibodies reactive to Echaffeensis was detected that county or parishwas classified as seropositive

To confirm serologic data PCR andor cul-ture isolation were conducted on a total of 541deer from 122 seropositive populations and on61 deer from 20 seronegative populations Chi-square analysis was used to test for differencesof PCR prevalence between seropositive andseronegative populations

Prevalence of antibodies reactive to E chaf-feensis among different age classes and gendercategories were only determined for WTD inseropositive populations Serologic data fromDawson et al (1994) Lockhart et al (1996) Lit-tle et al (1997) Yabsley et al (2002) and thisstudy were combined for determining preva-lence among age classes because analysis ofWTD age data had not been previously con-ducted PCR testing was not conducted byDawson et al (1994) and Lockhart et al (1996)therefore analysis of age and PCR positivitywas conducted only on deer from Little et al(1997) Yabsley et al (2002) and this study Chi

square analysis was used to test for differencesin both seroprevalence and PCR positivityamong age classes

To accurately classify the serostatus of a pop-ulation an appropriate sample size needs to betested In previous studies testing of five deerfrom a population consistently detected E chaf-feensisndashreactive antibodies in deer populationsinfested with LST (Lockhart et al 1997 Little etal 1997 1998) and based on Dawson et al(1994) testing a mean of 21 deer failed to detectseroreactive deer in seronegative populationsHowever additional evaluation of the samplesizes required to reliably classify the serologicstatus of populations was conducted in thisstudy To determine minimal number of sam-ples needed to detect the presence of a seropos-itive deer in a population the formula n 5(1 2 (1-a)1D) (N 2 (D 2 1)2) was used as de-scribed (Thrusfield 1995) Because seronegativepopulations are more difficult to accuratelyclassify two methods were used to examine ad-equacy of sample size Larger numbers of deer(n 5 8 2 16) were tested from nine seronega-tive populations to enhance detection of po-tentially low prevalences in those populationsAlso four populations with no history of LSTinfestation were repeatedly tested for severalyears

To assess the predicted epidemiologic asso-ciation between LST infestation and serologicstatus of WTD populations chi-square analy-sis was used to test for an association betweenLST presence and seroreactors at the popula-tion level Furthermore 11 populations forwhich both tick and serologic data were avail-able were evaluated during multiple yearsthree populations had LST infestation eachsample year four never had any evidence ofLST infestation and four were negative for LSTduring earlier sampling period (s) but becameLST positive during subsequent test years Chi-square analysis was used to test for differencesbetween groups

RESULTS

Regional serologic database for WTD

Among the 2101 WTD collected specificallyfor this study 984 (468 CI95 5 447

YABSLEY ET AL198

489) had antibodies reactive ($1128 titer) toE chaffeensis by IFA testing (Table 1) The meanprevalence of antibodies reactive to E chaffeen-sis in seropositive populations was 738(SD 5 258 range 20ndash100) White-tailed deerwith antibodies reactive to E chaffeensis weredetected in all states tested except West Vir-ginia The modal antibody titer for seropositivedeer was 1128 with a maximum titer of 14096The GMT of a subset of 152 seropositive WTDwas 355

The combined dataset of WTD serologicallytested during this and four prior studies (Daw-son et al 1994a Lockhart et al 1996 Little etal 1997 Yabsley et al 2002) contained a totalof 3275 WTD from 18 southeastern and southcentral states In the combined data set 1549(473) WTD examined had antibodies reactiveto E chaffeensis with seropositive populationsdetected in 17 of 18 states (Fig 1)

The highest overall seroprevalences were inArkansas and Missouri Although high preva-lences were detected in local WTD populationsin most states considerable variation in preva-lence occurred within different regions of somestates (eg Florida Kansas North Carolina

Oklahoma and Virginia) A western boundaryfor the distribution of E chaffeensis was evidentspanning the states of Kansas Oklahoma andTexas Similarly a southern boundary wasnoted across peninsular Florida Clusters ofseronegative populations were detected insouthern Mississippi and Louisiana along thelower Mississippi River floodplain and in theAppalachian Mountains from western Mary-land to northern Alabama

PCR and culture validation of E chaffeensisserologic data

Eighty-six of 368 (234) WTD from 50seropositive populations were PCR positive forE chaffeensis by either 16S or VLPT PCR Of 101WTD previously tested by 16S PCR (Yabsley etal 2002) nine deer from seven populationswere positive by VLPT PCR Collectively forthis study and Yabsley et al (2002) 95 of 469(203) WTD from 56 of 122 (46) seropositivepopulations have been confirmed using PCR(Table 1) 70 of 95 deer with both gene targets2 of 95 for only the 16S rRNA gene and 23 of95 only for the VLPT gene In contrast 16S

199

TABLE 1 RESULTS OF SEROLOGIC AND POLYMERASE CHAIN REACTION TESTING

FOR EHRLICHIA CHAFFEENSIS IN WHITE-TAILED DEER FROM 18 STATESa

Counties Number IFA Number PCRtested by positive positive

State Years IFA test no tested () no tested ()

Alabama 1981ndash2001 33 59141 (42) NTb

Arkansas 1982ndash2002 29 111154 (72) 746 (15)Florida 1983ndash2002 15 43104 (40) 936 (25)Georgia 1973ndash2002 50 119243 (49) 2698 (27)Kansas 1998ndash2002 38 3976 (51) 326 (12)Kentucky 1983ndash2001 27 2699 (26) 223 (9)Louisiana 1981ndash2001 34 54166 (33) 118 (6)Maryland 1999ndash2002 5 1225 (48) 17 (14)Missouri 1994ndash2002 19 5779 (72) 323 (13)Mississippi 1995ndash2001 38 80152 (53) 119 (5)North Carolina 1982ndash2002 33 75131 (57) 1572 (21)New Jersey 2001 2 610 (60) 05Oklahoma 1985ndash2002 13 3050 (60) 03South Carolina 1991ndash2001 26 52124 (42) 460 (7)Tennessee 1981ndash2001 20 54100 (54) 010Texas 1992ndash2002 33 73123 (59) 013Virginia 1983ndash2002 37 94168 (56) 2357 (40)West Virginia 1977ndash2001 18 0156 014Total 471 9842101 (47) 95530 (18)

aIncludes PCR results for the VLPT antigen gene for 101 WTD previously tested by PCR for the 16S rRNA gene(Yabsley et al 2002)NT not tested

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

rRNA and VLPT PCR assays on 61 deer from20 seronegative populations were all negativePresence of PCR positive deer was strongly as-sociated with seropositive populations (x2 5116 p 0001)

Of 72 culture attempts 34 (442) were lostto contamination with bacteria andor Try-panosoma cervi Seven of the remaining 38 cul-tures (184) were positive for E chaffeensisPositive cultures were obtained from two WTDfrom Greene Co AR two from Woodruff CoAR and one each from Clarke Co GA AnsonCo NC and Georgetown Co SC All isolateswere confirmed as E chaffeensis by PCR of the16S rRNA and VLPT gene targets

Relationship of age class and gender to antibody prevalence and PCR detection

Age was available on 1882 deer tested sero-logically 1086 deer from this study 517 fromDawson et al (1994) 165 from Lockhart et al

(1996) five from Little et al (1997) and 109from Yabsley et al (2002) For this combineddata set differences were not noted (x2 5 565p 5 046) in seroprevalance among age classes(Fig 2A) Similarly seroprevalence did not dif-fer between males and females (693 vs696 x2 5 0016 p 005)

Age was available for 393 deer from seropos-itive populations that were tested by PCRHigher proportions of deer were PCR positivein the younger age classes (x2 5 717 p 0001) Among deer 075 years old 531were PCR positive within the next age group(076ndash15 years) the prevalence decreased sig-nificantly to 27 (x2 5 147 p 0001) and lessthan 8 of deer 15 years old were PCR pos-itive (Fig 2B) The mean age of PCR positiveWTD was 11 years (range 025ndash4 years) In-sufficient culture isolates were made to exam-ine age-related associations however themean age for culture positive deer was 32years (range 03ndash95 years) Similar to antibody

YABSLEY ET AL200

FIG 1 Indirect immunofluorescent antibody results for Ehrlichia chaffeensis in white-tailed deer Results are from thecurrent study and from previously published work obtained in collaboration with SCWDS Darkly shaded counties pos-itive and lightly shaded counties negative for antibodies ($1128) reactive for E chaffeensis Previous serologic resultsthat differed from the results of the current study are shown as small circles within the county d previously positivefor antibodies ($1128) reactive for E chaffeensis s previously negative for antibodies reactive for E chaffeensis

prevalence PCR positivity did not differ be-tween males and females (19 vs 228 x2 5089 p 005)

Evaluation of sample size adequacy

Regarding sample size for seropositive pop-ulations post hoc statistical analysis utilizing amean 73 seroprevalence indicated that test-ing of only three or four deer gives a high prob-ability (95 and 99 respectively) of detectinga positive population (Thrusfield 1995) Al-though validation of seronegative populationsis more problematic testing of larger numbersof deer per population and repeated testing ofselected populations both produced consistentnegative results In a combined dataset fromDawson et al (1994a) and this study 250 deerfrom 17 seronegative populations (9ndash30 deer

per population) were negative for antibodies toE chaffeensis with means of 208 and 104 deertested per seronegative population respec-tively Similarly none of the 131 deer from fourpopulations consistently negative for LST in-festation had antibodies reactive to E chaffeen-sis The mean number of WTD tested per pop-ulation in this study was 62 (SD 5 61 range1ndash58) and data for populations with samplesizes less than four deer were used only if aseropositive animal was detected

Association between LST and serologic status at population level

Of 117 WTD populations examined for ticks82 were infested with LST 29 with A macula-tum (Gulf Coast tick) four with Ixodes affinisand 13 with I scapularis (blacklegged tick)

201

FIG 2 (A) Prevalence of antibodies reactive to E chaffeensis in WTD among age classes Numbers in bars representnumber of deer tested (B) Prevalence of PCR-positive WTD among age classes Numbers in bars represent numberof deer tested and different letters represent significant differences at p 005

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

Eighty of the 82 (976) populations infestedwith LST contained at least one seropositiveWTD while only four of 35 (114) populationsnot infested with LST contained seropositivedeer One LST-negative population (Harris CoGA) was both seronegative and tick negativein 1986 however in 2002 the deer populationwas seropositive but the LST status was notevaluated The prevalence of antibodies reac-tive to E chaffeensis was significantly higheramong LST infested populations of WTD thanpopulations not infested by LST (x2 5 899 p 0001) None of the six populations infestedonly with A maculatum contained seropositivedeer there were insufficient populations withI affinis or I scapularis alone to permit evalua-tion

Stability and spread of E chaffeensis and LST among WTD populations

Fifty-seven populations which were testedduring previous studies (Dawson et al 1994aLockhart et al 1996) were retested during thisstudy The serologic status for 50 (88) of thesepopulations remained unchanged 35 seropos-itive populations remained seropositive and 15seronegative populations remained seronega-tive Six populations which previously testedseronegative were seropositive during this

study Three of these representing Anson CoNC Haywood Co TN and Stewart Co GAhad documented invasion of LST between thefirst reported testing and this study One Texaspopulation (Travis Co) was seropositive (n 57 714) when tested in 1992 (Dawson et al1994a) but was seronegative when sampled(n 5 10) in 2002 for this study The LST statusfor this county was not known for either sam-pling period

Among the 11 populations where LST infes-tation and serologic status were monitored repeatedly over time there was complete con-cordance between LST presence and seroposi-tive deer (Table 2) All three LST positive pop-ulations were consistently seropositive eachyear tested The mean seroprevalence for thesethree populations was 83 and two of thesepopulations (Jones Co GA and Ashley CoAR) were confirmed PCR positive for E chaf-feensis None of the four populations where LSTwas consistently absent contained seropositivedeer at any sampling time Among the fourpopulations where LST status changed fromabsent to present seropositive deer were onlydetected following the appearance of LST Themean prevalence of antibodies reactive to Echaffeensis in the most recently tested year wassignificantly higher for populations infestedwith LST for multiple years (mean 5 100)

YABSLEY ET AL202

TABLE 2 SEROLOGIC AND TICK SURVEY RESULTS FROM SEVEN POPULATIONS OF WTD COLLECTED DURING

MULTIPLE TIME PERIODS AND TESTED FOR ANTIBODIES REACTIVE FOR E CHAFFEENSIS

No of Tickyears status No of deer Mean IFA prevalence

Site Years sampled (years) tested (range)

Sites positive for ticksJones Co GA 1979ndash2001 6 1 (6) 41 92 (80ndash100)Ashley Co AR 1982ndash2001 7 1 (7) 37 84 (71ndash100)Franklin Co FL 1984ndash1996 5 1 (5) 25 68 (40ndash100)

Sites negative for ticksFloyd Co GA 1973ndash2001 6 2 (6) 58 0Tyler Co WV 1985ndash2000 4 2 (4) 21 0Hardy Co WV 1977ndash1999 7 2 (7) 37 0Monroe Co FL 1992ndash2001 4 2 (4) 15 0

Sites where tick status changedAnson Co NC 1987 2001 2 2 1a 10 0 40bConcordia Pa LA 1986 1991 1999 3 2 1 1 22 0 38 60Haywood Co TN 1989 1994 1998 3 2 2 1 16 0 0 20Stewart Co GA 1986 1998 2 2 1 10 0 20

aTick status at sequential sampling periodsbPrevalence of IFA seropositive deer at sequential sampling periods

compared to populations only recently infested(mean 5 538 x2 5 155 p 0001) One re-cently infested population (Anson Co NC)was confirmed by isolation of E chaffeensisfrom a single WTD in 2001

DISCUSSION

The overarching goal of this study was toimplement and evaluate a prototype surveil-lance system using WTD as natural sentinels todetermine the geographic distribution of Echaffeensis across most of the suspected HMEendemic region of the United States To thisend the data obtained demonstrate the fol-lowing critical attributes (1) serologic findingsreflect E chaffeensis infection (2) effective sur-veillance can be achieved with small samplesizes (3) any deer 6 month old is suitable forsurveillance purposes (4) enzootic and consis-tently negative locales were identified repeat-edly across time (5) surveillance of deer hasthe ability to detect spread to new locales and(6) serologic molecular and culture diagnosticfindings among deer could be related to pres-ence of the principal tick vector A americanumThese findings compliment and expand on arecent WTD serologic surveillance effort withinIowa (Mueller-Anneling et al 2000) howeverthat study exclusively used serologic assaysand did not include either confirmatory mi-crobiological diagnostics or contemporaneoustick vector monitoring Collectively these twostudies confirm the effectiveness of this proto-typical surveillance system for monitoring thedistribution of this emerging human pathogen

Although several studies have examined theprevalence and distribution of E chaffeensis an-tibodies among many southeastern and southcentral WTD populations (Dawson et al 1994aLockhart et al 1996 Little et al 1997 Yabsleyet al 2002) the extensive additional testingfilled many geographic gaps and provide amore complete fine scale distribution of E chaf-feensis among WTD throughout the regions ofthe United States with the highest HME risk(Paddock and Childs 2003) The combined dataevaluated herein disclosed a wide distributionof seropositive WTD populations across all ofthe 18 states represented except for West Vir-

ginia however some distributional limits alsowere discerned including a western boundaryextending across Kansas Oklahoma andTexas a southern boundary across peninsularFlorida and a large cluster of many seronega-tive populations centered along the Ap-palachian Mountains (Fig 1) Other prior stud-ies of WTD have documented an apparentnorthern range limit for E chaffeensis in south-ern portions of Iowa Illinois Indiana and Ohio(Dawson et al 1994a Irving et al 2000 Mueller-Anneling et al 2000) Along the East coast sur-veys have indicated that A americanum and Echaffeensis are present as far north as Connecti-cut Maine and Rhode Island (Keirans and La-combe 1998 Ijdo et al 2000)

The accuracy of diagnostic assays utilized iscritical to any pathogen surveillance system In-fection with E chaffeensis was confirmed innearly half (46) of 122 seropositive WTD pop-ulations tested by PCR and was supplementedby isolation from others In contrast PCR evi-dence of infection was not detected in any of20 seronegative WTD populations Thus PCRand culture testing applied at the populationlevel were effective at confirming infection inmany seropositive populations and demon-strating a strong association between the sero-logic status of populations and infection withE chaffeensis

Because WTD in these regions are known to be infected with three other related rickettsiae(E ewingii A phagocytophilum and an Ana-plasma sp [WTD-agent]) in addition to E chaf-feensis (Dawson et al 1996b Little et al 1998Brandsma et al 1999 Magnarelli et al 1999Yabsley et al 2002) the potential for serologiccross-reaction is an important considerationAlthough antibodies reactive with E chaffeensisin WTD from Maryland have been confirmedby immunoblotting using the 28- to 29-kDa anti-gens of E chaffeensis (Walls et al 1998) the ex-tent of serologic cross-reactions among thesefour species is not fully understood (Dumlerand Walker 2001) However data from otherstudies suggests that infection with these otherrickettsiae may not commonly result in pro-duction of antibodies that cross-react with Echaffeensis Cross-reactions between E chaffeen-sis and E ewingii have been noted in infectedhumans and dogs however not all E ewingii

203EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

infected dogs or humans develop antibodies toE chaffeensis antigens (Murphy et al 1998 Pad-dock et al 2001) Similarly a WTD fawn ex-perimentally infected with E ewingii did not de-velop antibodies reactive with E chaffeensis(Yabsley et al 2002) Cross-reactions between Echaffeensis and A phagocytophilum have been re-ported in humans and may complicate diagno-sis (Comer et al 1999) Far fewer WTD in thesoutheastern states have antibodies reactive toA phagocytophilum (25) than to E chaffeensisantigens however although some deer reactedto both E chaffeensis and A phagocytophilum ator above the 1128 cut-off many also reacted toonly one of these antigens at titers $1128 (Lit-tle et al 1998 Walls et al 1998 VG Dugan andMJ Yabsley unpublished data) The high per-centage of populations where seropositive orseronegative status was confirmed by PCR orculture together with data from these otherstudies provide considerable evidence that theseroreactivity reported herein largely repre-sents specific seroconversion to E chaffeensis

The distribution of antibodies within WTDage and gender categories was investigated be-cause these variables are an important consid-eration for a WTD natural sentinel systemOnly a limited investigation of geometric meantiters among age categories has been reportedfor WTD (Lockhart et al 1995) and any signif-icant influence of age or gender on the occur-rence of antibodies would require samplingstratified according to these host attributes Im-portantly differences in antibody prevalencewere not noted among age or gender categoriesindicating that all WTD particularly animals $6 mo of age are suitable for use in a WTD sur-veillance system In contrast to the high stableprevalence of antibodies among age classes(Fig 2A) PCR evidence of rickettsemia de-clined with age (Fig 2B) This pattern conformsto proposed infection dynamics in naturally in-fected WTD whereby animals from approxi-mately 6 months to 15 years of age are morelikely to be rickettsemic than older adults (Pad-dock and Childs 2003 Dawson et al 1994Lockhart et al 1997a Davidson et al 2001) Inthe present study the prevalence of PCR pos-itivity dropped dramatically from over half(53) of WTD younger than 075 years to lessthan 8 of WTD older than 15 years The ma-

jority of deer tested from some seropositivepopulations were 15 years and only deer15 years from other populations were avail-able for PCR testing which may explain whynot all seropositive populations were con-firmed by PCR Because the probability of aWTD being rickettsemic declines with ageserology represents a better surveillance toolthan PCR and under natural conditions of re-exposure to ticks titers likely do not declineover time as often occurred in single exposureexperimental infections of WTD (Dawson et al1994b Ewing et al 1995 Davidson et al 2001)

Previous serologic surveys of E chaffeensis inWTD (Dawson et al 1994a Lockhart et al 1996Little et al 1997 Yabsley et al 2002) have uti-lized relatively small sample sizes per popula-tion but except for one (Mueller-Anneling etal 2000) have not addressed the adequacy ofthese sample sizes to correctly classify infectedand uninfected populations Data from the cur-rent and prior studies (Dawson et al 1994aLockhart et al 1996 Little et al 1997) haveshown that infected WTD populations in thesoutheastern and southcentral United Statestypically have high (70) seroprevalencesCalculations (Thrusfield 1995) based on a mean73 seroprevalence indicated that use of a sam-ple size of five WTD per population should re-liably detect nearly all positive populations Al-though it was beyond the scope of this studyto test enough deer to detect very low popula-tion prevalences larger sample sizes and re-peated sampling of seronegative populationsdid not reveal seropositive populations thatwould have been misclassified as seronegativebased on smaller sample sizes A chance of mis-classifying newly infected populations of WTDexists because seroprevalence may be low ini-tially however field evidence shows that anti-body prevalence can increase rapidly once apopulation becomes infected with E chaffeensis(Lockhart et al 1995) Utilizing per county sam-ple sizes equivalent to this study statewideserotesting of WTD from Iowa demonstrated aclear north-south gradient of seroprevalencethat corresponded to key epidemiologic factors(eg distributions of ticks deer and habitats)(Mueller-Anneling et al 2000)

Because the distributions of pathogens oftenare reflected by some combination of both sta-

YABSLEY ET AL204

tic and changing geographic distributions animportant attribute of an effective surveillancesystem is the ability to identify both persis-tently enzootic locales and spread to new loca-tions Testing of WTD from LST infested pop-ulations over the span of 20 years showed thatonce this biologically important vector is pres-ent and WTD populations are infected with Echaffeensis there is a high probability that thepopulations will remain infected and that WTDsentinels reflect these enzootic conditions Con-versely populations with no history of LST in-festation remained consistently seronegativewhen tested multiple times These results arecompatible with a strong site-specific associa-tion between E chaffeensis antibodies and thepresence of LST (Lockhart et al 1996) Of in-terest each of two populations (Choctaw CoAL and Bolivar Co MS) that were earlier re-ported as seropositive but LST negative (Lock-hart et al 1996) were shown in this study to beboth LST positive and seropositive for E chaf-feensis The rare failure to find LST in seropos-itive WTD populations may have occurred be-cause many collections were made during thefall hunting season when LST infestations areundergoing rapid seasonal declines (Allan2001)

Four instances of E chaffeensis spread to newlocations were detected by monitoring WTDfor multiple years (Table 2) At each of these lo-cations the change in population serologic sta-tus from negative to positive coincided withdocumented appearance of LST These findingsare very similar to those from a 12-year-longmonitoring of a WTD population that demon-strated the introduction of E chaffeensis was at-tributable to the establishment of LST (Lock-hart et al 1995) Because all of these changes toseropositive status were linked to the principalvector they provide further evidence that thissurveillance system reflects important epi-demiologic factors and that it should discernchanging local HME risks

In summary the biological reasons a sentinelWTD surveillance system functions effectivelyare that the force of transmission of E chaffeen-sis is focused on WTD and that aspects of WTDbiology allow attribution of infection to a spe-cific geographic location From an operationalperspective the system has the desirable at-

tributes of applicability on a wide geographicscale and logistic feasibility by using ldquofreerdquo(hunter-harvested) samples The present workwas done in a research context however it ispossible to implement sentinel WTD serologicsurveillance for E chaffeensis in an operationalpublic health context as clearly demonstratedpreviously in Iowa (Mueller-Anneling et al2000)

ACKNOWLEDGMENTS

We are indebted to over 100 individuals rep-resenting state wildlife agency personnelSCWDS staff and white-tailed deer researchersfor help in field collection of samples but es-pecially the following individuals for coordi-nating specific collections Chris Cook (AL)John Morgan (FL) Lloyd Fox (KS) Jon Gassettand David Yancy (KY) Jeff Beringer (MO)Douglas Roscoe (NJ) Evin Stanford (NC) AAlan Kocan (OK) Charles Ruth Jr (SC) BenLayton (TN) Matt Knox (VA) and Jim Crum(WV) Sarah Cross Page Luttrell and DarrellKavanaugh from SCWDS provided excellentfield and laboratory help Work was supportedprimarily by the National Institutes of Allergyand Infectious Diseases (5 R01 AI044235-02)Further support was provided by the FederalAid to Wildlife Restoration Act (50 Stat 917)and through sponsorship from fish and wildlifeagencies in Alabama Arkansas Florida Geor-gia Kansas Kentucky Louisiana MarylandMississippi Missouri North Carolina Okla-homa Puerto Rico South Carolina TennesseeVirginia and West Virginia

REFERENCES

Allan SA Ticks (Class Arachnida Order Acarina) InSamuel WM Pybus MJ and Kocan AA ed ParasiticDiseases of Wild Mammals Ames IA Iowa State Uni-versity Press 2001 72ndash106

Anderson BE Sumner JW Dawson JE et al Detection ofthe etiologic agent of human ehrlichiosis by polymerasechain reaction J Clin Microbiol 1992 30775ndash780

Anderson BE Sims KG Olson JG et al Amblyommaamericanum a potential vector of human ehrlichiosisAm J Trop Med Hyg 1993 49239ndash244

Bloemer SR Zimmerman RH Fairbanks K Abundanceattachment sites and density estimators of lone star

205EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

ticks (Acari Ixodidae) infesting white-tailed deer JMed Entomol 1988 25295ndash300

Boromisa RD Grimstand PR Seroconversion rates toJamestown Canyon virus among six populations ofwhite-tailed deer (Odocoileus virginianus) in Indiana JWildl Dis 1987 2723ndash33

Brandsma AR Little SE Lockhart JM et al NovelEhrlichia organism (Rickettsiales Ehrlichieae) in white-tailed deer associated with lone star tick (Acari Ixodi-dae) parasitism J Med Entomol 1999 36190ndash194

Childs JE Paddock CD The ascendancy of Amblyommaamericanum as a vector of pathogens affecting humansin the United States Annu Rev Entomol 2003 48307ndash337

Comer JA Nicholson WL Olson JG et al Serologic test-ing for human granulocytic ehrlichiosis at a national re-ferral center J Clin Microbiol 1999 37558ndash564

Dawson JE Childs JE Biggie KL et al White-tailed deeras a potential reservoir of Ehrlichia spp J Wildl Dis1994a 30162ndash168

Dawson JE Stallknecht D Howerth EW et al Suscep-tibility of white-tailed deer (Odocoileus virginianus) toinfection with Ehrlichia chaffeensis the etiologic agent ofhuman ehrlichiosis J Clin Microbiol 1994b 322725ndash2728

Dawson JE Anderson BE Fishbein DB et al Poly-merase chain reaction evidence of Ehrlichia chaffeensisetiologic agent of human ehrlichiosis in dogs fromsoutheast Virginia Am J Vet Res 1996a 571175ndash1179

Dawson JE Warner CK Baker V et al Ehrlichia-like 16SrDNA sequence from wild white-tailed deer (Odocoileusvirginianus) J Parasitol 1996b 8252ndash58

Davidson WR Lockhart JM Stallknecht DE et al Per-sistent Ehrlichia chaffeensis infection in white-tailed deerJ Wildl Dis 2001 37538ndash546

Dumler JS Walker DH Tick-borne ehrlichiosis LancetInfect Dis 2001 021ndash28

Ewing SA Dawson JE Kocan AA et al Experimentaltransmission of Ehrlichia chaffeensis (RickettsialesEhrlichieae) among white-tailed deer by Amblyommaamericanum (Acari Ixodidae) J Med Entomol 199532368ndash374

Fletcher WO Stallknecht DE Kearney MT et al Anti-bodies to vesicular stomatitis New Jersey type virus inwhite-tailed deer on Ossabaw Island Georgia 1985ndash1989 J Wildl Dis 1991 27675ndash680

Gill JS McLean RG Neitzil DF et al Serologic analysisof white-tailed deer sera for antibodies to Borreliaburgdorferi by enzyme-linked immunosorbent assayand western blotting J Clin Microbiol 1993 31318ndash322

Ijdo JW Wu C Magnarelli LA et al Detection ofEhrlichia chaffeensis DNA in Amblyomma americanumticks in Connecticut and Rhode Island J Clin Microbiol2000 384655ndash4656

Irving RP Pinger RR Vann CN et al Distribution ofEhrlichia chaffeensis (Rickettsiales Rickettsiaeceae) inAmblyomma americanum in southern Indiana and preva-lence of E chaffeensis-reactive antibodies in white-taileddeer in Indiana and Ohio in 1998 J Med Entomol 200037595ndash600

Keirans JE Lacombe EH First records of Amblyommaamericanum Ixodes (Ixodes) dentatus and Ixodes (Cer-atixodes) uriae (Acari Ixodidae) from Maine J Parasitol1998 84629ndash631

Little SE Dawson JE Lockhart JM et al Developmentand use of specific polymerase chain reaction for thedetection of an Ehrlichia-like organism in white-taileddeer J Wildl Dis 1997 33246ndash253

Little SE Stallknecht DE Lockhart JM et al Naturalcoinfection of a white-tailed deer (Odocoileus virgini-anus) population with three Ehrlichia spp J Parasitol1998 84897ndash901

Lockhart JM Davidson WR Dawson JE et al Tempo-ral association of Amblyomma americanumwith the pres-ence of Ehrlichia chaffeensis reactive antibodies in white-tailed deer J Wildl Dis 1995 31119ndash124

Lockhart JM Davidson WR Stallknecht DE et al Site-specific geographic association between Amblyommaamericanum (Acari Ixodidae) infestations and Ehrlichiachaffeensis-reactive (Rickettsiales Ehrlichieae) antibod-ies in white-tailed deer J Med Entomol 1996 33153ndash158

Lockhart JM Davidson WR Stallknecht DE et al Iso-lation of Ehrlichia chaffeensis from wild white-tailed deer(Odocoileus virginianus) confirms their role as naturalreservoir hosts J Clin Microbiol 1997a 351681ndash1686

Lockhart JM Davidson WR Stallknecht DE et al Nat-ural history of Ehrlichia chaffeensis (RickettsialesEhrlichieae) in the Piedmont physiographic province ofGeorgia J Parasitol 1997b 83887ndash894

Magnarelli LA Anderson JF Stafford KC et al Anti-bodies to multiple tick-borne pathogens of babesiosisehrlichiosis and Lyme borreliosis in white-footed miceJ Wildl Dis 1997 33466ndash473

Mueller-Anneling L Gilchrist MJ Thorne PS Ehrlichiachaffeensis antibodies in white-tailed deer Iowa 1994and 1996 Emerg Infect Dis 2000 6397ndash400

Murphy GL Ewing SA Whitworth LC et al A molec-ular and serologic survey of Ehrlichia canis E chaffeen-sis and E ewingii in dogs and ticks from OklahomaVet Parasitol 1998 79325ndash339

Paddock CD Childs JE Ehrlichia chaffeensis a prototyp-ical emerging pathogen Clin Microbiol Rev 2003 1637ndash64

Paddock CD Folk SM Shore GM et al Infections withEhrlichia chaffeensisand Ehrlichia ewingii in persons coin-fected with human immunodeficiency virus Clin InfectDis 2001 31586ndash1594

Roland WE Everett ED Cyr TL et al Ehrlichia chaffeensisin Missouri ticks Am J Trop Med Hyg 1998 59641ndash643

Smart DL and Trainer DO Serologic evidence ofVenezuelan equine encephalitis in some wild and do-mestic populations of southern Texas J Wildl Dis 197511195ndash200

Stallknecht DE Blue JL Rollor EA et al Precipitatingantibodies to bluetongue and epizootic hemorrhagicdisease viruses in white-tailed deer in the southeasternUnited States J Wildl Dis 1991 27238ndash247

Steiert JG Gilfoy F Infection rates of Amblyomma ameri-canum and Dermacentor variabilis by Ehrlichia chaffeensis

YABSLEY ET AL206

Louis MO) supplemented with 10 fetalbovine serum (Sigma) and monitored for evi-dence of cytopathic effect or for a maximum of45 days Cultures showing cytopathic effectand cultures completing the 45-day incubationperiod were harvested with a cell scraper andtested by direct fluorescent antibody assay aspreviously described (Lockhart et al 1997a)

Data analysis

Serologic data from this study (n 5 2101)were combined with data from previous stud-ies conducted independently by SCWDS re-searchers (n 5 425 Lockhart et al 1996 Littleet al 1997 Yabsley et al 2002) or collabora-tively with others (n 5 749 Dawson et al1994a) to construct a comprehensive map delineating the distribution of E chaffeensisseroreactive WTD in the southeastern andsouth central United States Some of the previ-ous studies used different cutoff values for IFAtesting (eg 164 or 1128) but in order to stan-dardize results from this study and previousstudies IFA assays were classified as positiveonly if a sample had a titer of $1128 To facil-itate graphic presentation data for each popu-lation were categorized by county or parish ifone or more deer with antibodies reactive to Echaffeensis was detected that county or parishwas classified as seropositive

To confirm serologic data PCR andor cul-ture isolation were conducted on a total of 541deer from 122 seropositive populations and on61 deer from 20 seronegative populations Chi-square analysis was used to test for differencesof PCR prevalence between seropositive andseronegative populations

Prevalence of antibodies reactive to E chaf-feensis among different age classes and gendercategories were only determined for WTD inseropositive populations Serologic data fromDawson et al (1994) Lockhart et al (1996) Lit-tle et al (1997) Yabsley et al (2002) and thisstudy were combined for determining preva-lence among age classes because analysis ofWTD age data had not been previously con-ducted PCR testing was not conducted byDawson et al (1994) and Lockhart et al (1996)therefore analysis of age and PCR positivitywas conducted only on deer from Little et al(1997) Yabsley et al (2002) and this study Chi

square analysis was used to test for differencesin both seroprevalence and PCR positivityamong age classes

To accurately classify the serostatus of a pop-ulation an appropriate sample size needs to betested In previous studies testing of five deerfrom a population consistently detected E chaf-feensisndashreactive antibodies in deer populationsinfested with LST (Lockhart et al 1997 Little etal 1997 1998) and based on Dawson et al(1994) testing a mean of 21 deer failed to detectseroreactive deer in seronegative populationsHowever additional evaluation of the samplesizes required to reliably classify the serologicstatus of populations was conducted in thisstudy To determine minimal number of sam-ples needed to detect the presence of a seropos-itive deer in a population the formula n 5(1 2 (1-a)1D) (N 2 (D 2 1)2) was used as de-scribed (Thrusfield 1995) Because seronegativepopulations are more difficult to accuratelyclassify two methods were used to examine ad-equacy of sample size Larger numbers of deer(n 5 8 2 16) were tested from nine seronega-tive populations to enhance detection of po-tentially low prevalences in those populationsAlso four populations with no history of LSTinfestation were repeatedly tested for severalyears

To assess the predicted epidemiologic asso-ciation between LST infestation and serologicstatus of WTD populations chi-square analy-sis was used to test for an association betweenLST presence and seroreactors at the popula-tion level Furthermore 11 populations forwhich both tick and serologic data were avail-able were evaluated during multiple yearsthree populations had LST infestation eachsample year four never had any evidence ofLST infestation and four were negative for LSTduring earlier sampling period (s) but becameLST positive during subsequent test years Chi-square analysis was used to test for differencesbetween groups

RESULTS

Regional serologic database for WTD

Among the 2101 WTD collected specificallyfor this study 984 (468 CI95 5 447

YABSLEY ET AL198

489) had antibodies reactive ($1128 titer) toE chaffeensis by IFA testing (Table 1) The meanprevalence of antibodies reactive to E chaffeen-sis in seropositive populations was 738(SD 5 258 range 20ndash100) White-tailed deerwith antibodies reactive to E chaffeensis weredetected in all states tested except West Vir-ginia The modal antibody titer for seropositivedeer was 1128 with a maximum titer of 14096The GMT of a subset of 152 seropositive WTDwas 355

The combined dataset of WTD serologicallytested during this and four prior studies (Daw-son et al 1994a Lockhart et al 1996 Little etal 1997 Yabsley et al 2002) contained a totalof 3275 WTD from 18 southeastern and southcentral states In the combined data set 1549(473) WTD examined had antibodies reactiveto E chaffeensis with seropositive populationsdetected in 17 of 18 states (Fig 1)

The highest overall seroprevalences were inArkansas and Missouri Although high preva-lences were detected in local WTD populationsin most states considerable variation in preva-lence occurred within different regions of somestates (eg Florida Kansas North Carolina

Oklahoma and Virginia) A western boundaryfor the distribution of E chaffeensis was evidentspanning the states of Kansas Oklahoma andTexas Similarly a southern boundary wasnoted across peninsular Florida Clusters ofseronegative populations were detected insouthern Mississippi and Louisiana along thelower Mississippi River floodplain and in theAppalachian Mountains from western Mary-land to northern Alabama

PCR and culture validation of E chaffeensisserologic data

Eighty-six of 368 (234) WTD from 50seropositive populations were PCR positive forE chaffeensis by either 16S or VLPT PCR Of 101WTD previously tested by 16S PCR (Yabsley etal 2002) nine deer from seven populationswere positive by VLPT PCR Collectively forthis study and Yabsley et al (2002) 95 of 469(203) WTD from 56 of 122 (46) seropositivepopulations have been confirmed using PCR(Table 1) 70 of 95 deer with both gene targets2 of 95 for only the 16S rRNA gene and 23 of95 only for the VLPT gene In contrast 16S

199

TABLE 1 RESULTS OF SEROLOGIC AND POLYMERASE CHAIN REACTION TESTING

FOR EHRLICHIA CHAFFEENSIS IN WHITE-TAILED DEER FROM 18 STATESa

Counties Number IFA Number PCRtested by positive positive

State Years IFA test no tested () no tested ()

Alabama 1981ndash2001 33 59141 (42) NTb

Arkansas 1982ndash2002 29 111154 (72) 746 (15)Florida 1983ndash2002 15 43104 (40) 936 (25)Georgia 1973ndash2002 50 119243 (49) 2698 (27)Kansas 1998ndash2002 38 3976 (51) 326 (12)Kentucky 1983ndash2001 27 2699 (26) 223 (9)Louisiana 1981ndash2001 34 54166 (33) 118 (6)Maryland 1999ndash2002 5 1225 (48) 17 (14)Missouri 1994ndash2002 19 5779 (72) 323 (13)Mississippi 1995ndash2001 38 80152 (53) 119 (5)North Carolina 1982ndash2002 33 75131 (57) 1572 (21)New Jersey 2001 2 610 (60) 05Oklahoma 1985ndash2002 13 3050 (60) 03South Carolina 1991ndash2001 26 52124 (42) 460 (7)Tennessee 1981ndash2001 20 54100 (54) 010Texas 1992ndash2002 33 73123 (59) 013Virginia 1983ndash2002 37 94168 (56) 2357 (40)West Virginia 1977ndash2001 18 0156 014Total 471 9842101 (47) 95530 (18)

aIncludes PCR results for the VLPT antigen gene for 101 WTD previously tested by PCR for the 16S rRNA gene(Yabsley et al 2002)NT not tested

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

rRNA and VLPT PCR assays on 61 deer from20 seronegative populations were all negativePresence of PCR positive deer was strongly as-sociated with seropositive populations (x2 5116 p 0001)

Of 72 culture attempts 34 (442) were lostto contamination with bacteria andor Try-panosoma cervi Seven of the remaining 38 cul-tures (184) were positive for E chaffeensisPositive cultures were obtained from two WTDfrom Greene Co AR two from Woodruff CoAR and one each from Clarke Co GA AnsonCo NC and Georgetown Co SC All isolateswere confirmed as E chaffeensis by PCR of the16S rRNA and VLPT gene targets

Relationship of age class and gender to antibody prevalence and PCR detection

Age was available on 1882 deer tested sero-logically 1086 deer from this study 517 fromDawson et al (1994) 165 from Lockhart et al

(1996) five from Little et al (1997) and 109from Yabsley et al (2002) For this combineddata set differences were not noted (x2 5 565p 5 046) in seroprevalance among age classes(Fig 2A) Similarly seroprevalence did not dif-fer between males and females (693 vs696 x2 5 0016 p 005)

Age was available for 393 deer from seropos-itive populations that were tested by PCRHigher proportions of deer were PCR positivein the younger age classes (x2 5 717 p 0001) Among deer 075 years old 531were PCR positive within the next age group(076ndash15 years) the prevalence decreased sig-nificantly to 27 (x2 5 147 p 0001) and lessthan 8 of deer 15 years old were PCR pos-itive (Fig 2B) The mean age of PCR positiveWTD was 11 years (range 025ndash4 years) In-sufficient culture isolates were made to exam-ine age-related associations however themean age for culture positive deer was 32years (range 03ndash95 years) Similar to antibody

YABSLEY ET AL200

FIG 1 Indirect immunofluorescent antibody results for Ehrlichia chaffeensis in white-tailed deer Results are from thecurrent study and from previously published work obtained in collaboration with SCWDS Darkly shaded counties pos-itive and lightly shaded counties negative for antibodies ($1128) reactive for E chaffeensis Previous serologic resultsthat differed from the results of the current study are shown as small circles within the county d previously positivefor antibodies ($1128) reactive for E chaffeensis s previously negative for antibodies reactive for E chaffeensis

prevalence PCR positivity did not differ be-tween males and females (19 vs 228 x2 5089 p 005)

Evaluation of sample size adequacy

Regarding sample size for seropositive pop-ulations post hoc statistical analysis utilizing amean 73 seroprevalence indicated that test-ing of only three or four deer gives a high prob-ability (95 and 99 respectively) of detectinga positive population (Thrusfield 1995) Al-though validation of seronegative populationsis more problematic testing of larger numbersof deer per population and repeated testing ofselected populations both produced consistentnegative results In a combined dataset fromDawson et al (1994a) and this study 250 deerfrom 17 seronegative populations (9ndash30 deer

per population) were negative for antibodies toE chaffeensis with means of 208 and 104 deertested per seronegative population respec-tively Similarly none of the 131 deer from fourpopulations consistently negative for LST in-festation had antibodies reactive to E chaffeen-sis The mean number of WTD tested per pop-ulation in this study was 62 (SD 5 61 range1ndash58) and data for populations with samplesizes less than four deer were used only if aseropositive animal was detected

Association between LST and serologic status at population level

Of 117 WTD populations examined for ticks82 were infested with LST 29 with A macula-tum (Gulf Coast tick) four with Ixodes affinisand 13 with I scapularis (blacklegged tick)

201

FIG 2 (A) Prevalence of antibodies reactive to E chaffeensis in WTD among age classes Numbers in bars representnumber of deer tested (B) Prevalence of PCR-positive WTD among age classes Numbers in bars represent numberof deer tested and different letters represent significant differences at p 005

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

Eighty of the 82 (976) populations infestedwith LST contained at least one seropositiveWTD while only four of 35 (114) populationsnot infested with LST contained seropositivedeer One LST-negative population (Harris CoGA) was both seronegative and tick negativein 1986 however in 2002 the deer populationwas seropositive but the LST status was notevaluated The prevalence of antibodies reac-tive to E chaffeensis was significantly higheramong LST infested populations of WTD thanpopulations not infested by LST (x2 5 899 p 0001) None of the six populations infestedonly with A maculatum contained seropositivedeer there were insufficient populations withI affinis or I scapularis alone to permit evalua-tion

Stability and spread of E chaffeensis and LST among WTD populations

Fifty-seven populations which were testedduring previous studies (Dawson et al 1994aLockhart et al 1996) were retested during thisstudy The serologic status for 50 (88) of thesepopulations remained unchanged 35 seropos-itive populations remained seropositive and 15seronegative populations remained seronega-tive Six populations which previously testedseronegative were seropositive during this

study Three of these representing Anson CoNC Haywood Co TN and Stewart Co GAhad documented invasion of LST between thefirst reported testing and this study One Texaspopulation (Travis Co) was seropositive (n 57 714) when tested in 1992 (Dawson et al1994a) but was seronegative when sampled(n 5 10) in 2002 for this study The LST statusfor this county was not known for either sam-pling period

Among the 11 populations where LST infes-tation and serologic status were monitored repeatedly over time there was complete con-cordance between LST presence and seroposi-tive deer (Table 2) All three LST positive pop-ulations were consistently seropositive eachyear tested The mean seroprevalence for thesethree populations was 83 and two of thesepopulations (Jones Co GA and Ashley CoAR) were confirmed PCR positive for E chaf-feensis None of the four populations where LSTwas consistently absent contained seropositivedeer at any sampling time Among the fourpopulations where LST status changed fromabsent to present seropositive deer were onlydetected following the appearance of LST Themean prevalence of antibodies reactive to Echaffeensis in the most recently tested year wassignificantly higher for populations infestedwith LST for multiple years (mean 5 100)

YABSLEY ET AL202

TABLE 2 SEROLOGIC AND TICK SURVEY RESULTS FROM SEVEN POPULATIONS OF WTD COLLECTED DURING

MULTIPLE TIME PERIODS AND TESTED FOR ANTIBODIES REACTIVE FOR E CHAFFEENSIS

No of Tickyears status No of deer Mean IFA prevalence

Site Years sampled (years) tested (range)

Sites positive for ticksJones Co GA 1979ndash2001 6 1 (6) 41 92 (80ndash100)Ashley Co AR 1982ndash2001 7 1 (7) 37 84 (71ndash100)Franklin Co FL 1984ndash1996 5 1 (5) 25 68 (40ndash100)

Sites negative for ticksFloyd Co GA 1973ndash2001 6 2 (6) 58 0Tyler Co WV 1985ndash2000 4 2 (4) 21 0Hardy Co WV 1977ndash1999 7 2 (7) 37 0Monroe Co FL 1992ndash2001 4 2 (4) 15 0

Sites where tick status changedAnson Co NC 1987 2001 2 2 1a 10 0 40bConcordia Pa LA 1986 1991 1999 3 2 1 1 22 0 38 60Haywood Co TN 1989 1994 1998 3 2 2 1 16 0 0 20Stewart Co GA 1986 1998 2 2 1 10 0 20

aTick status at sequential sampling periodsbPrevalence of IFA seropositive deer at sequential sampling periods

compared to populations only recently infested(mean 5 538 x2 5 155 p 0001) One re-cently infested population (Anson Co NC)was confirmed by isolation of E chaffeensisfrom a single WTD in 2001

DISCUSSION

The overarching goal of this study was toimplement and evaluate a prototype surveil-lance system using WTD as natural sentinels todetermine the geographic distribution of Echaffeensis across most of the suspected HMEendemic region of the United States To thisend the data obtained demonstrate the fol-lowing critical attributes (1) serologic findingsreflect E chaffeensis infection (2) effective sur-veillance can be achieved with small samplesizes (3) any deer 6 month old is suitable forsurveillance purposes (4) enzootic and consis-tently negative locales were identified repeat-edly across time (5) surveillance of deer hasthe ability to detect spread to new locales and(6) serologic molecular and culture diagnosticfindings among deer could be related to pres-ence of the principal tick vector A americanumThese findings compliment and expand on arecent WTD serologic surveillance effort withinIowa (Mueller-Anneling et al 2000) howeverthat study exclusively used serologic assaysand did not include either confirmatory mi-crobiological diagnostics or contemporaneoustick vector monitoring Collectively these twostudies confirm the effectiveness of this proto-typical surveillance system for monitoring thedistribution of this emerging human pathogen

Although several studies have examined theprevalence and distribution of E chaffeensis an-tibodies among many southeastern and southcentral WTD populations (Dawson et al 1994aLockhart et al 1996 Little et al 1997 Yabsleyet al 2002) the extensive additional testingfilled many geographic gaps and provide amore complete fine scale distribution of E chaf-feensis among WTD throughout the regions ofthe United States with the highest HME risk(Paddock and Childs 2003) The combined dataevaluated herein disclosed a wide distributionof seropositive WTD populations across all ofthe 18 states represented except for West Vir-

ginia however some distributional limits alsowere discerned including a western boundaryextending across Kansas Oklahoma andTexas a southern boundary across peninsularFlorida and a large cluster of many seronega-tive populations centered along the Ap-palachian Mountains (Fig 1) Other prior stud-ies of WTD have documented an apparentnorthern range limit for E chaffeensis in south-ern portions of Iowa Illinois Indiana and Ohio(Dawson et al 1994a Irving et al 2000 Mueller-Anneling et al 2000) Along the East coast sur-veys have indicated that A americanum and Echaffeensis are present as far north as Connecti-cut Maine and Rhode Island (Keirans and La-combe 1998 Ijdo et al 2000)

The accuracy of diagnostic assays utilized iscritical to any pathogen surveillance system In-fection with E chaffeensis was confirmed innearly half (46) of 122 seropositive WTD pop-ulations tested by PCR and was supplementedby isolation from others In contrast PCR evi-dence of infection was not detected in any of20 seronegative WTD populations Thus PCRand culture testing applied at the populationlevel were effective at confirming infection inmany seropositive populations and demon-strating a strong association between the sero-logic status of populations and infection withE chaffeensis

Because WTD in these regions are known to be infected with three other related rickettsiae(E ewingii A phagocytophilum and an Ana-plasma sp [WTD-agent]) in addition to E chaf-feensis (Dawson et al 1996b Little et al 1998Brandsma et al 1999 Magnarelli et al 1999Yabsley et al 2002) the potential for serologiccross-reaction is an important considerationAlthough antibodies reactive with E chaffeensisin WTD from Maryland have been confirmedby immunoblotting using the 28- to 29-kDa anti-gens of E chaffeensis (Walls et al 1998) the ex-tent of serologic cross-reactions among thesefour species is not fully understood (Dumlerand Walker 2001) However data from otherstudies suggests that infection with these otherrickettsiae may not commonly result in pro-duction of antibodies that cross-react with Echaffeensis Cross-reactions between E chaffeen-sis and E ewingii have been noted in infectedhumans and dogs however not all E ewingii

203EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

infected dogs or humans develop antibodies toE chaffeensis antigens (Murphy et al 1998 Pad-dock et al 2001) Similarly a WTD fawn ex-perimentally infected with E ewingii did not de-velop antibodies reactive with E chaffeensis(Yabsley et al 2002) Cross-reactions between Echaffeensis and A phagocytophilum have been re-ported in humans and may complicate diagno-sis (Comer et al 1999) Far fewer WTD in thesoutheastern states have antibodies reactive toA phagocytophilum (25) than to E chaffeensisantigens however although some deer reactedto both E chaffeensis and A phagocytophilum ator above the 1128 cut-off many also reacted toonly one of these antigens at titers $1128 (Lit-tle et al 1998 Walls et al 1998 VG Dugan andMJ Yabsley unpublished data) The high per-centage of populations where seropositive orseronegative status was confirmed by PCR orculture together with data from these otherstudies provide considerable evidence that theseroreactivity reported herein largely repre-sents specific seroconversion to E chaffeensis

The distribution of antibodies within WTDage and gender categories was investigated be-cause these variables are an important consid-eration for a WTD natural sentinel systemOnly a limited investigation of geometric meantiters among age categories has been reportedfor WTD (Lockhart et al 1995) and any signif-icant influence of age or gender on the occur-rence of antibodies would require samplingstratified according to these host attributes Im-portantly differences in antibody prevalencewere not noted among age or gender categoriesindicating that all WTD particularly animals $6 mo of age are suitable for use in a WTD sur-veillance system In contrast to the high stableprevalence of antibodies among age classes(Fig 2A) PCR evidence of rickettsemia de-clined with age (Fig 2B) This pattern conformsto proposed infection dynamics in naturally in-fected WTD whereby animals from approxi-mately 6 months to 15 years of age are morelikely to be rickettsemic than older adults (Pad-dock and Childs 2003 Dawson et al 1994Lockhart et al 1997a Davidson et al 2001) Inthe present study the prevalence of PCR pos-itivity dropped dramatically from over half(53) of WTD younger than 075 years to lessthan 8 of WTD older than 15 years The ma-

jority of deer tested from some seropositivepopulations were 15 years and only deer15 years from other populations were avail-able for PCR testing which may explain whynot all seropositive populations were con-firmed by PCR Because the probability of aWTD being rickettsemic declines with ageserology represents a better surveillance toolthan PCR and under natural conditions of re-exposure to ticks titers likely do not declineover time as often occurred in single exposureexperimental infections of WTD (Dawson et al1994b Ewing et al 1995 Davidson et al 2001)

Previous serologic surveys of E chaffeensis inWTD (Dawson et al 1994a Lockhart et al 1996Little et al 1997 Yabsley et al 2002) have uti-lized relatively small sample sizes per popula-tion but except for one (Mueller-Anneling etal 2000) have not addressed the adequacy ofthese sample sizes to correctly classify infectedand uninfected populations Data from the cur-rent and prior studies (Dawson et al 1994aLockhart et al 1996 Little et al 1997) haveshown that infected WTD populations in thesoutheastern and southcentral United Statestypically have high (70) seroprevalencesCalculations (Thrusfield 1995) based on a mean73 seroprevalence indicated that use of a sam-ple size of five WTD per population should re-liably detect nearly all positive populations Al-though it was beyond the scope of this studyto test enough deer to detect very low popula-tion prevalences larger sample sizes and re-peated sampling of seronegative populationsdid not reveal seropositive populations thatwould have been misclassified as seronegativebased on smaller sample sizes A chance of mis-classifying newly infected populations of WTDexists because seroprevalence may be low ini-tially however field evidence shows that anti-body prevalence can increase rapidly once apopulation becomes infected with E chaffeensis(Lockhart et al 1995) Utilizing per county sam-ple sizes equivalent to this study statewideserotesting of WTD from Iowa demonstrated aclear north-south gradient of seroprevalencethat corresponded to key epidemiologic factors(eg distributions of ticks deer and habitats)(Mueller-Anneling et al 2000)

Because the distributions of pathogens oftenare reflected by some combination of both sta-

YABSLEY ET AL204

tic and changing geographic distributions animportant attribute of an effective surveillancesystem is the ability to identify both persis-tently enzootic locales and spread to new loca-tions Testing of WTD from LST infested pop-ulations over the span of 20 years showed thatonce this biologically important vector is pres-ent and WTD populations are infected with Echaffeensis there is a high probability that thepopulations will remain infected and that WTDsentinels reflect these enzootic conditions Con-versely populations with no history of LST in-festation remained consistently seronegativewhen tested multiple times These results arecompatible with a strong site-specific associa-tion between E chaffeensis antibodies and thepresence of LST (Lockhart et al 1996) Of in-terest each of two populations (Choctaw CoAL and Bolivar Co MS) that were earlier re-ported as seropositive but LST negative (Lock-hart et al 1996) were shown in this study to beboth LST positive and seropositive for E chaf-feensis The rare failure to find LST in seropos-itive WTD populations may have occurred be-cause many collections were made during thefall hunting season when LST infestations areundergoing rapid seasonal declines (Allan2001)

Four instances of E chaffeensis spread to newlocations were detected by monitoring WTDfor multiple years (Table 2) At each of these lo-cations the change in population serologic sta-tus from negative to positive coincided withdocumented appearance of LST These findingsare very similar to those from a 12-year-longmonitoring of a WTD population that demon-strated the introduction of E chaffeensis was at-tributable to the establishment of LST (Lock-hart et al 1995) Because all of these changes toseropositive status were linked to the principalvector they provide further evidence that thissurveillance system reflects important epi-demiologic factors and that it should discernchanging local HME risks

In summary the biological reasons a sentinelWTD surveillance system functions effectivelyare that the force of transmission of E chaffeen-sis is focused on WTD and that aspects of WTDbiology allow attribution of infection to a spe-cific geographic location From an operationalperspective the system has the desirable at-

tributes of applicability on a wide geographicscale and logistic feasibility by using ldquofreerdquo(hunter-harvested) samples The present workwas done in a research context however it ispossible to implement sentinel WTD serologicsurveillance for E chaffeensis in an operationalpublic health context as clearly demonstratedpreviously in Iowa (Mueller-Anneling et al2000)

ACKNOWLEDGMENTS

We are indebted to over 100 individuals rep-resenting state wildlife agency personnelSCWDS staff and white-tailed deer researchersfor help in field collection of samples but es-pecially the following individuals for coordi-nating specific collections Chris Cook (AL)John Morgan (FL) Lloyd Fox (KS) Jon Gassettand David Yancy (KY) Jeff Beringer (MO)Douglas Roscoe (NJ) Evin Stanford (NC) AAlan Kocan (OK) Charles Ruth Jr (SC) BenLayton (TN) Matt Knox (VA) and Jim Crum(WV) Sarah Cross Page Luttrell and DarrellKavanaugh from SCWDS provided excellentfield and laboratory help Work was supportedprimarily by the National Institutes of Allergyand Infectious Diseases (5 R01 AI044235-02)Further support was provided by the FederalAid to Wildlife Restoration Act (50 Stat 917)and through sponsorship from fish and wildlifeagencies in Alabama Arkansas Florida Geor-gia Kansas Kentucky Louisiana MarylandMississippi Missouri North Carolina Okla-homa Puerto Rico South Carolina TennesseeVirginia and West Virginia

REFERENCES

Allan SA Ticks (Class Arachnida Order Acarina) InSamuel WM Pybus MJ and Kocan AA ed ParasiticDiseases of Wild Mammals Ames IA Iowa State Uni-versity Press 2001 72ndash106

Anderson BE Sumner JW Dawson JE et al Detection ofthe etiologic agent of human ehrlichiosis by polymerasechain reaction J Clin Microbiol 1992 30775ndash780

Anderson BE Sims KG Olson JG et al Amblyommaamericanum a potential vector of human ehrlichiosisAm J Trop Med Hyg 1993 49239ndash244

Bloemer SR Zimmerman RH Fairbanks K Abundanceattachment sites and density estimators of lone star

205EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

ticks (Acari Ixodidae) infesting white-tailed deer JMed Entomol 1988 25295ndash300

Boromisa RD Grimstand PR Seroconversion rates toJamestown Canyon virus among six populations ofwhite-tailed deer (Odocoileus virginianus) in Indiana JWildl Dis 1987 2723ndash33

Brandsma AR Little SE Lockhart JM et al NovelEhrlichia organism (Rickettsiales Ehrlichieae) in white-tailed deer associated with lone star tick (Acari Ixodi-dae) parasitism J Med Entomol 1999 36190ndash194

Childs JE Paddock CD The ascendancy of Amblyommaamericanum as a vector of pathogens affecting humansin the United States Annu Rev Entomol 2003 48307ndash337

Comer JA Nicholson WL Olson JG et al Serologic test-ing for human granulocytic ehrlichiosis at a national re-ferral center J Clin Microbiol 1999 37558ndash564

Dawson JE Childs JE Biggie KL et al White-tailed deeras a potential reservoir of Ehrlichia spp J Wildl Dis1994a 30162ndash168

Dawson JE Stallknecht D Howerth EW et al Suscep-tibility of white-tailed deer (Odocoileus virginianus) toinfection with Ehrlichia chaffeensis the etiologic agent ofhuman ehrlichiosis J Clin Microbiol 1994b 322725ndash2728

Dawson JE Anderson BE Fishbein DB et al Poly-merase chain reaction evidence of Ehrlichia chaffeensisetiologic agent of human ehrlichiosis in dogs fromsoutheast Virginia Am J Vet Res 1996a 571175ndash1179

Dawson JE Warner CK Baker V et al Ehrlichia-like 16SrDNA sequence from wild white-tailed deer (Odocoileusvirginianus) J Parasitol 1996b 8252ndash58

Davidson WR Lockhart JM Stallknecht DE et al Per-sistent Ehrlichia chaffeensis infection in white-tailed deerJ Wildl Dis 2001 37538ndash546

Dumler JS Walker DH Tick-borne ehrlichiosis LancetInfect Dis 2001 021ndash28

Ewing SA Dawson JE Kocan AA et al Experimentaltransmission of Ehrlichia chaffeensis (RickettsialesEhrlichieae) among white-tailed deer by Amblyommaamericanum (Acari Ixodidae) J Med Entomol 199532368ndash374

Fletcher WO Stallknecht DE Kearney MT et al Anti-bodies to vesicular stomatitis New Jersey type virus inwhite-tailed deer on Ossabaw Island Georgia 1985ndash1989 J Wildl Dis 1991 27675ndash680

Gill JS McLean RG Neitzil DF et al Serologic analysisof white-tailed deer sera for antibodies to Borreliaburgdorferi by enzyme-linked immunosorbent assayand western blotting J Clin Microbiol 1993 31318ndash322

Ijdo JW Wu C Magnarelli LA et al Detection ofEhrlichia chaffeensis DNA in Amblyomma americanumticks in Connecticut and Rhode Island J Clin Microbiol2000 384655ndash4656

Irving RP Pinger RR Vann CN et al Distribution ofEhrlichia chaffeensis (Rickettsiales Rickettsiaeceae) inAmblyomma americanum in southern Indiana and preva-lence of E chaffeensis-reactive antibodies in white-taileddeer in Indiana and Ohio in 1998 J Med Entomol 200037595ndash600

Keirans JE Lacombe EH First records of Amblyommaamericanum Ixodes (Ixodes) dentatus and Ixodes (Cer-atixodes) uriae (Acari Ixodidae) from Maine J Parasitol1998 84629ndash631

Little SE Dawson JE Lockhart JM et al Developmentand use of specific polymerase chain reaction for thedetection of an Ehrlichia-like organism in white-taileddeer J Wildl Dis 1997 33246ndash253

Little SE Stallknecht DE Lockhart JM et al Naturalcoinfection of a white-tailed deer (Odocoileus virgini-anus) population with three Ehrlichia spp J Parasitol1998 84897ndash901

Lockhart JM Davidson WR Dawson JE et al Tempo-ral association of Amblyomma americanumwith the pres-ence of Ehrlichia chaffeensis reactive antibodies in white-tailed deer J Wildl Dis 1995 31119ndash124

Lockhart JM Davidson WR Stallknecht DE et al Site-specific geographic association between Amblyommaamericanum (Acari Ixodidae) infestations and Ehrlichiachaffeensis-reactive (Rickettsiales Ehrlichieae) antibod-ies in white-tailed deer J Med Entomol 1996 33153ndash158

Lockhart JM Davidson WR Stallknecht DE et al Iso-lation of Ehrlichia chaffeensis from wild white-tailed deer(Odocoileus virginianus) confirms their role as naturalreservoir hosts J Clin Microbiol 1997a 351681ndash1686

Lockhart JM Davidson WR Stallknecht DE et al Nat-ural history of Ehrlichia chaffeensis (RickettsialesEhrlichieae) in the Piedmont physiographic province ofGeorgia J Parasitol 1997b 83887ndash894

Magnarelli LA Anderson JF Stafford KC et al Anti-bodies to multiple tick-borne pathogens of babesiosisehrlichiosis and Lyme borreliosis in white-footed miceJ Wildl Dis 1997 33466ndash473

Mueller-Anneling L Gilchrist MJ Thorne PS Ehrlichiachaffeensis antibodies in white-tailed deer Iowa 1994and 1996 Emerg Infect Dis 2000 6397ndash400

Murphy GL Ewing SA Whitworth LC et al A molec-ular and serologic survey of Ehrlichia canis E chaffeen-sis and E ewingii in dogs and ticks from OklahomaVet Parasitol 1998 79325ndash339

Paddock CD Childs JE Ehrlichia chaffeensis a prototyp-ical emerging pathogen Clin Microbiol Rev 2003 1637ndash64

Paddock CD Folk SM Shore GM et al Infections withEhrlichia chaffeensisand Ehrlichia ewingii in persons coin-fected with human immunodeficiency virus Clin InfectDis 2001 31586ndash1594

Roland WE Everett ED Cyr TL et al Ehrlichia chaffeensisin Missouri ticks Am J Trop Med Hyg 1998 59641ndash643

Smart DL and Trainer DO Serologic evidence ofVenezuelan equine encephalitis in some wild and do-mestic populations of southern Texas J Wildl Dis 197511195ndash200

Stallknecht DE Blue JL Rollor EA et al Precipitatingantibodies to bluetongue and epizootic hemorrhagicdisease viruses in white-tailed deer in the southeasternUnited States J Wildl Dis 1991 27238ndash247

Steiert JG Gilfoy F Infection rates of Amblyomma ameri-canum and Dermacentor variabilis by Ehrlichia chaffeensis

YABSLEY ET AL206

489) had antibodies reactive ($1128 titer) toE chaffeensis by IFA testing (Table 1) The meanprevalence of antibodies reactive to E chaffeen-sis in seropositive populations was 738(SD 5 258 range 20ndash100) White-tailed deerwith antibodies reactive to E chaffeensis weredetected in all states tested except West Vir-ginia The modal antibody titer for seropositivedeer was 1128 with a maximum titer of 14096The GMT of a subset of 152 seropositive WTDwas 355

The combined dataset of WTD serologicallytested during this and four prior studies (Daw-son et al 1994a Lockhart et al 1996 Little etal 1997 Yabsley et al 2002) contained a totalof 3275 WTD from 18 southeastern and southcentral states In the combined data set 1549(473) WTD examined had antibodies reactiveto E chaffeensis with seropositive populationsdetected in 17 of 18 states (Fig 1)

The highest overall seroprevalences were inArkansas and Missouri Although high preva-lences were detected in local WTD populationsin most states considerable variation in preva-lence occurred within different regions of somestates (eg Florida Kansas North Carolina

Oklahoma and Virginia) A western boundaryfor the distribution of E chaffeensis was evidentspanning the states of Kansas Oklahoma andTexas Similarly a southern boundary wasnoted across peninsular Florida Clusters ofseronegative populations were detected insouthern Mississippi and Louisiana along thelower Mississippi River floodplain and in theAppalachian Mountains from western Mary-land to northern Alabama

PCR and culture validation of E chaffeensisserologic data

Eighty-six of 368 (234) WTD from 50seropositive populations were PCR positive forE chaffeensis by either 16S or VLPT PCR Of 101WTD previously tested by 16S PCR (Yabsley etal 2002) nine deer from seven populationswere positive by VLPT PCR Collectively forthis study and Yabsley et al (2002) 95 of 469(203) WTD from 56 of 122 (46) seropositivepopulations have been confirmed using PCR(Table 1) 70 of 95 deer with both gene targets2 of 95 for only the 16S rRNA gene and 23 of95 only for the VLPT gene In contrast 16S

199

TABLE 1 RESULTS OF SEROLOGIC AND POLYMERASE CHAIN REACTION TESTING

FOR EHRLICHIA CHAFFEENSIS IN WHITE-TAILED DEER FROM 18 STATESa

Counties Number IFA Number PCRtested by positive positive

State Years IFA test no tested () no tested ()

Alabama 1981ndash2001 33 59141 (42) NTb

Arkansas 1982ndash2002 29 111154 (72) 746 (15)Florida 1983ndash2002 15 43104 (40) 936 (25)Georgia 1973ndash2002 50 119243 (49) 2698 (27)Kansas 1998ndash2002 38 3976 (51) 326 (12)Kentucky 1983ndash2001 27 2699 (26) 223 (9)Louisiana 1981ndash2001 34 54166 (33) 118 (6)Maryland 1999ndash2002 5 1225 (48) 17 (14)Missouri 1994ndash2002 19 5779 (72) 323 (13)Mississippi 1995ndash2001 38 80152 (53) 119 (5)North Carolina 1982ndash2002 33 75131 (57) 1572 (21)New Jersey 2001 2 610 (60) 05Oklahoma 1985ndash2002 13 3050 (60) 03South Carolina 1991ndash2001 26 52124 (42) 460 (7)Tennessee 1981ndash2001 20 54100 (54) 010Texas 1992ndash2002 33 73123 (59) 013Virginia 1983ndash2002 37 94168 (56) 2357 (40)West Virginia 1977ndash2001 18 0156 014Total 471 9842101 (47) 95530 (18)

aIncludes PCR results for the VLPT antigen gene for 101 WTD previously tested by PCR for the 16S rRNA gene(Yabsley et al 2002)NT not tested

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

rRNA and VLPT PCR assays on 61 deer from20 seronegative populations were all negativePresence of PCR positive deer was strongly as-sociated with seropositive populations (x2 5116 p 0001)

Of 72 culture attempts 34 (442) were lostto contamination with bacteria andor Try-panosoma cervi Seven of the remaining 38 cul-tures (184) were positive for E chaffeensisPositive cultures were obtained from two WTDfrom Greene Co AR two from Woodruff CoAR and one each from Clarke Co GA AnsonCo NC and Georgetown Co SC All isolateswere confirmed as E chaffeensis by PCR of the16S rRNA and VLPT gene targets

Relationship of age class and gender to antibody prevalence and PCR detection

Age was available on 1882 deer tested sero-logically 1086 deer from this study 517 fromDawson et al (1994) 165 from Lockhart et al

(1996) five from Little et al (1997) and 109from Yabsley et al (2002) For this combineddata set differences were not noted (x2 5 565p 5 046) in seroprevalance among age classes(Fig 2A) Similarly seroprevalence did not dif-fer between males and females (693 vs696 x2 5 0016 p 005)

Age was available for 393 deer from seropos-itive populations that were tested by PCRHigher proportions of deer were PCR positivein the younger age classes (x2 5 717 p 0001) Among deer 075 years old 531were PCR positive within the next age group(076ndash15 years) the prevalence decreased sig-nificantly to 27 (x2 5 147 p 0001) and lessthan 8 of deer 15 years old were PCR pos-itive (Fig 2B) The mean age of PCR positiveWTD was 11 years (range 025ndash4 years) In-sufficient culture isolates were made to exam-ine age-related associations however themean age for culture positive deer was 32years (range 03ndash95 years) Similar to antibody

YABSLEY ET AL200

FIG 1 Indirect immunofluorescent antibody results for Ehrlichia chaffeensis in white-tailed deer Results are from thecurrent study and from previously published work obtained in collaboration with SCWDS Darkly shaded counties pos-itive and lightly shaded counties negative for antibodies ($1128) reactive for E chaffeensis Previous serologic resultsthat differed from the results of the current study are shown as small circles within the county d previously positivefor antibodies ($1128) reactive for E chaffeensis s previously negative for antibodies reactive for E chaffeensis

prevalence PCR positivity did not differ be-tween males and females (19 vs 228 x2 5089 p 005)

Evaluation of sample size adequacy

Regarding sample size for seropositive pop-ulations post hoc statistical analysis utilizing amean 73 seroprevalence indicated that test-ing of only three or four deer gives a high prob-ability (95 and 99 respectively) of detectinga positive population (Thrusfield 1995) Al-though validation of seronegative populationsis more problematic testing of larger numbersof deer per population and repeated testing ofselected populations both produced consistentnegative results In a combined dataset fromDawson et al (1994a) and this study 250 deerfrom 17 seronegative populations (9ndash30 deer

per population) were negative for antibodies toE chaffeensis with means of 208 and 104 deertested per seronegative population respec-tively Similarly none of the 131 deer from fourpopulations consistently negative for LST in-festation had antibodies reactive to E chaffeen-sis The mean number of WTD tested per pop-ulation in this study was 62 (SD 5 61 range1ndash58) and data for populations with samplesizes less than four deer were used only if aseropositive animal was detected

Association between LST and serologic status at population level

Of 117 WTD populations examined for ticks82 were infested with LST 29 with A macula-tum (Gulf Coast tick) four with Ixodes affinisand 13 with I scapularis (blacklegged tick)

201

FIG 2 (A) Prevalence of antibodies reactive to E chaffeensis in WTD among age classes Numbers in bars representnumber of deer tested (B) Prevalence of PCR-positive WTD among age classes Numbers in bars represent numberof deer tested and different letters represent significant differences at p 005

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

Eighty of the 82 (976) populations infestedwith LST contained at least one seropositiveWTD while only four of 35 (114) populationsnot infested with LST contained seropositivedeer One LST-negative population (Harris CoGA) was both seronegative and tick negativein 1986 however in 2002 the deer populationwas seropositive but the LST status was notevaluated The prevalence of antibodies reac-tive to E chaffeensis was significantly higheramong LST infested populations of WTD thanpopulations not infested by LST (x2 5 899 p 0001) None of the six populations infestedonly with A maculatum contained seropositivedeer there were insufficient populations withI affinis or I scapularis alone to permit evalua-tion

Stability and spread of E chaffeensis and LST among WTD populations

Fifty-seven populations which were testedduring previous studies (Dawson et al 1994aLockhart et al 1996) were retested during thisstudy The serologic status for 50 (88) of thesepopulations remained unchanged 35 seropos-itive populations remained seropositive and 15seronegative populations remained seronega-tive Six populations which previously testedseronegative were seropositive during this

study Three of these representing Anson CoNC Haywood Co TN and Stewart Co GAhad documented invasion of LST between thefirst reported testing and this study One Texaspopulation (Travis Co) was seropositive (n 57 714) when tested in 1992 (Dawson et al1994a) but was seronegative when sampled(n 5 10) in 2002 for this study The LST statusfor this county was not known for either sam-pling period

Among the 11 populations where LST infes-tation and serologic status were monitored repeatedly over time there was complete con-cordance between LST presence and seroposi-tive deer (Table 2) All three LST positive pop-ulations were consistently seropositive eachyear tested The mean seroprevalence for thesethree populations was 83 and two of thesepopulations (Jones Co GA and Ashley CoAR) were confirmed PCR positive for E chaf-feensis None of the four populations where LSTwas consistently absent contained seropositivedeer at any sampling time Among the fourpopulations where LST status changed fromabsent to present seropositive deer were onlydetected following the appearance of LST Themean prevalence of antibodies reactive to Echaffeensis in the most recently tested year wassignificantly higher for populations infestedwith LST for multiple years (mean 5 100)

YABSLEY ET AL202

TABLE 2 SEROLOGIC AND TICK SURVEY RESULTS FROM SEVEN POPULATIONS OF WTD COLLECTED DURING

MULTIPLE TIME PERIODS AND TESTED FOR ANTIBODIES REACTIVE FOR E CHAFFEENSIS

No of Tickyears status No of deer Mean IFA prevalence

Site Years sampled (years) tested (range)

Sites positive for ticksJones Co GA 1979ndash2001 6 1 (6) 41 92 (80ndash100)Ashley Co AR 1982ndash2001 7 1 (7) 37 84 (71ndash100)Franklin Co FL 1984ndash1996 5 1 (5) 25 68 (40ndash100)

Sites negative for ticksFloyd Co GA 1973ndash2001 6 2 (6) 58 0Tyler Co WV 1985ndash2000 4 2 (4) 21 0Hardy Co WV 1977ndash1999 7 2 (7) 37 0Monroe Co FL 1992ndash2001 4 2 (4) 15 0

Sites where tick status changedAnson Co NC 1987 2001 2 2 1a 10 0 40bConcordia Pa LA 1986 1991 1999 3 2 1 1 22 0 38 60Haywood Co TN 1989 1994 1998 3 2 2 1 16 0 0 20Stewart Co GA 1986 1998 2 2 1 10 0 20

aTick status at sequential sampling periodsbPrevalence of IFA seropositive deer at sequential sampling periods

compared to populations only recently infested(mean 5 538 x2 5 155 p 0001) One re-cently infested population (Anson Co NC)was confirmed by isolation of E chaffeensisfrom a single WTD in 2001

DISCUSSION

The overarching goal of this study was toimplement and evaluate a prototype surveil-lance system using WTD as natural sentinels todetermine the geographic distribution of Echaffeensis across most of the suspected HMEendemic region of the United States To thisend the data obtained demonstrate the fol-lowing critical attributes (1) serologic findingsreflect E chaffeensis infection (2) effective sur-veillance can be achieved with small samplesizes (3) any deer 6 month old is suitable forsurveillance purposes (4) enzootic and consis-tently negative locales were identified repeat-edly across time (5) surveillance of deer hasthe ability to detect spread to new locales and(6) serologic molecular and culture diagnosticfindings among deer could be related to pres-ence of the principal tick vector A americanumThese findings compliment and expand on arecent WTD serologic surveillance effort withinIowa (Mueller-Anneling et al 2000) howeverthat study exclusively used serologic assaysand did not include either confirmatory mi-crobiological diagnostics or contemporaneoustick vector monitoring Collectively these twostudies confirm the effectiveness of this proto-typical surveillance system for monitoring thedistribution of this emerging human pathogen

Although several studies have examined theprevalence and distribution of E chaffeensis an-tibodies among many southeastern and southcentral WTD populations (Dawson et al 1994aLockhart et al 1996 Little et al 1997 Yabsleyet al 2002) the extensive additional testingfilled many geographic gaps and provide amore complete fine scale distribution of E chaf-feensis among WTD throughout the regions ofthe United States with the highest HME risk(Paddock and Childs 2003) The combined dataevaluated herein disclosed a wide distributionof seropositive WTD populations across all ofthe 18 states represented except for West Vir-

ginia however some distributional limits alsowere discerned including a western boundaryextending across Kansas Oklahoma andTexas a southern boundary across peninsularFlorida and a large cluster of many seronega-tive populations centered along the Ap-palachian Mountains (Fig 1) Other prior stud-ies of WTD have documented an apparentnorthern range limit for E chaffeensis in south-ern portions of Iowa Illinois Indiana and Ohio(Dawson et al 1994a Irving et al 2000 Mueller-Anneling et al 2000) Along the East coast sur-veys have indicated that A americanum and Echaffeensis are present as far north as Connecti-cut Maine and Rhode Island (Keirans and La-combe 1998 Ijdo et al 2000)

The accuracy of diagnostic assays utilized iscritical to any pathogen surveillance system In-fection with E chaffeensis was confirmed innearly half (46) of 122 seropositive WTD pop-ulations tested by PCR and was supplementedby isolation from others In contrast PCR evi-dence of infection was not detected in any of20 seronegative WTD populations Thus PCRand culture testing applied at the populationlevel were effective at confirming infection inmany seropositive populations and demon-strating a strong association between the sero-logic status of populations and infection withE chaffeensis

Because WTD in these regions are known to be infected with three other related rickettsiae(E ewingii A phagocytophilum and an Ana-plasma sp [WTD-agent]) in addition to E chaf-feensis (Dawson et al 1996b Little et al 1998Brandsma et al 1999 Magnarelli et al 1999Yabsley et al 2002) the potential for serologiccross-reaction is an important considerationAlthough antibodies reactive with E chaffeensisin WTD from Maryland have been confirmedby immunoblotting using the 28- to 29-kDa anti-gens of E chaffeensis (Walls et al 1998) the ex-tent of serologic cross-reactions among thesefour species is not fully understood (Dumlerand Walker 2001) However data from otherstudies suggests that infection with these otherrickettsiae may not commonly result in pro-duction of antibodies that cross-react with Echaffeensis Cross-reactions between E chaffeen-sis and E ewingii have been noted in infectedhumans and dogs however not all E ewingii

203EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

infected dogs or humans develop antibodies toE chaffeensis antigens (Murphy et al 1998 Pad-dock et al 2001) Similarly a WTD fawn ex-perimentally infected with E ewingii did not de-velop antibodies reactive with E chaffeensis(Yabsley et al 2002) Cross-reactions between Echaffeensis and A phagocytophilum have been re-ported in humans and may complicate diagno-sis (Comer et al 1999) Far fewer WTD in thesoutheastern states have antibodies reactive toA phagocytophilum (25) than to E chaffeensisantigens however although some deer reactedto both E chaffeensis and A phagocytophilum ator above the 1128 cut-off many also reacted toonly one of these antigens at titers $1128 (Lit-tle et al 1998 Walls et al 1998 VG Dugan andMJ Yabsley unpublished data) The high per-centage of populations where seropositive orseronegative status was confirmed by PCR orculture together with data from these otherstudies provide considerable evidence that theseroreactivity reported herein largely repre-sents specific seroconversion to E chaffeensis

The distribution of antibodies within WTDage and gender categories was investigated be-cause these variables are an important consid-eration for a WTD natural sentinel systemOnly a limited investigation of geometric meantiters among age categories has been reportedfor WTD (Lockhart et al 1995) and any signif-icant influence of age or gender on the occur-rence of antibodies would require samplingstratified according to these host attributes Im-portantly differences in antibody prevalencewere not noted among age or gender categoriesindicating that all WTD particularly animals $6 mo of age are suitable for use in a WTD sur-veillance system In contrast to the high stableprevalence of antibodies among age classes(Fig 2A) PCR evidence of rickettsemia de-clined with age (Fig 2B) This pattern conformsto proposed infection dynamics in naturally in-fected WTD whereby animals from approxi-mately 6 months to 15 years of age are morelikely to be rickettsemic than older adults (Pad-dock and Childs 2003 Dawson et al 1994Lockhart et al 1997a Davidson et al 2001) Inthe present study the prevalence of PCR pos-itivity dropped dramatically from over half(53) of WTD younger than 075 years to lessthan 8 of WTD older than 15 years The ma-

jority of deer tested from some seropositivepopulations were 15 years and only deer15 years from other populations were avail-able for PCR testing which may explain whynot all seropositive populations were con-firmed by PCR Because the probability of aWTD being rickettsemic declines with ageserology represents a better surveillance toolthan PCR and under natural conditions of re-exposure to ticks titers likely do not declineover time as often occurred in single exposureexperimental infections of WTD (Dawson et al1994b Ewing et al 1995 Davidson et al 2001)

Previous serologic surveys of E chaffeensis inWTD (Dawson et al 1994a Lockhart et al 1996Little et al 1997 Yabsley et al 2002) have uti-lized relatively small sample sizes per popula-tion but except for one (Mueller-Anneling etal 2000) have not addressed the adequacy ofthese sample sizes to correctly classify infectedand uninfected populations Data from the cur-rent and prior studies (Dawson et al 1994aLockhart et al 1996 Little et al 1997) haveshown that infected WTD populations in thesoutheastern and southcentral United Statestypically have high (70) seroprevalencesCalculations (Thrusfield 1995) based on a mean73 seroprevalence indicated that use of a sam-ple size of five WTD per population should re-liably detect nearly all positive populations Al-though it was beyond the scope of this studyto test enough deer to detect very low popula-tion prevalences larger sample sizes and re-peated sampling of seronegative populationsdid not reveal seropositive populations thatwould have been misclassified as seronegativebased on smaller sample sizes A chance of mis-classifying newly infected populations of WTDexists because seroprevalence may be low ini-tially however field evidence shows that anti-body prevalence can increase rapidly once apopulation becomes infected with E chaffeensis(Lockhart et al 1995) Utilizing per county sam-ple sizes equivalent to this study statewideserotesting of WTD from Iowa demonstrated aclear north-south gradient of seroprevalencethat corresponded to key epidemiologic factors(eg distributions of ticks deer and habitats)(Mueller-Anneling et al 2000)

Because the distributions of pathogens oftenare reflected by some combination of both sta-

YABSLEY ET AL204

tic and changing geographic distributions animportant attribute of an effective surveillancesystem is the ability to identify both persis-tently enzootic locales and spread to new loca-tions Testing of WTD from LST infested pop-ulations over the span of 20 years showed thatonce this biologically important vector is pres-ent and WTD populations are infected with Echaffeensis there is a high probability that thepopulations will remain infected and that WTDsentinels reflect these enzootic conditions Con-versely populations with no history of LST in-festation remained consistently seronegativewhen tested multiple times These results arecompatible with a strong site-specific associa-tion between E chaffeensis antibodies and thepresence of LST (Lockhart et al 1996) Of in-terest each of two populations (Choctaw CoAL and Bolivar Co MS) that were earlier re-ported as seropositive but LST negative (Lock-hart et al 1996) were shown in this study to beboth LST positive and seropositive for E chaf-feensis The rare failure to find LST in seropos-itive WTD populations may have occurred be-cause many collections were made during thefall hunting season when LST infestations areundergoing rapid seasonal declines (Allan2001)

Four instances of E chaffeensis spread to newlocations were detected by monitoring WTDfor multiple years (Table 2) At each of these lo-cations the change in population serologic sta-tus from negative to positive coincided withdocumented appearance of LST These findingsare very similar to those from a 12-year-longmonitoring of a WTD population that demon-strated the introduction of E chaffeensis was at-tributable to the establishment of LST (Lock-hart et al 1995) Because all of these changes toseropositive status were linked to the principalvector they provide further evidence that thissurveillance system reflects important epi-demiologic factors and that it should discernchanging local HME risks

In summary the biological reasons a sentinelWTD surveillance system functions effectivelyare that the force of transmission of E chaffeen-sis is focused on WTD and that aspects of WTDbiology allow attribution of infection to a spe-cific geographic location From an operationalperspective the system has the desirable at-

tributes of applicability on a wide geographicscale and logistic feasibility by using ldquofreerdquo(hunter-harvested) samples The present workwas done in a research context however it ispossible to implement sentinel WTD serologicsurveillance for E chaffeensis in an operationalpublic health context as clearly demonstratedpreviously in Iowa (Mueller-Anneling et al2000)

ACKNOWLEDGMENTS

We are indebted to over 100 individuals rep-resenting state wildlife agency personnelSCWDS staff and white-tailed deer researchersfor help in field collection of samples but es-pecially the following individuals for coordi-nating specific collections Chris Cook (AL)John Morgan (FL) Lloyd Fox (KS) Jon Gassettand David Yancy (KY) Jeff Beringer (MO)Douglas Roscoe (NJ) Evin Stanford (NC) AAlan Kocan (OK) Charles Ruth Jr (SC) BenLayton (TN) Matt Knox (VA) and Jim Crum(WV) Sarah Cross Page Luttrell and DarrellKavanaugh from SCWDS provided excellentfield and laboratory help Work was supportedprimarily by the National Institutes of Allergyand Infectious Diseases (5 R01 AI044235-02)Further support was provided by the FederalAid to Wildlife Restoration Act (50 Stat 917)and through sponsorship from fish and wildlifeagencies in Alabama Arkansas Florida Geor-gia Kansas Kentucky Louisiana MarylandMississippi Missouri North Carolina Okla-homa Puerto Rico South Carolina TennesseeVirginia and West Virginia

REFERENCES

Allan SA Ticks (Class Arachnida Order Acarina) InSamuel WM Pybus MJ and Kocan AA ed ParasiticDiseases of Wild Mammals Ames IA Iowa State Uni-versity Press 2001 72ndash106

Anderson BE Sumner JW Dawson JE et al Detection ofthe etiologic agent of human ehrlichiosis by polymerasechain reaction J Clin Microbiol 1992 30775ndash780

Anderson BE Sims KG Olson JG et al Amblyommaamericanum a potential vector of human ehrlichiosisAm J Trop Med Hyg 1993 49239ndash244

Bloemer SR Zimmerman RH Fairbanks K Abundanceattachment sites and density estimators of lone star

205EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

ticks (Acari Ixodidae) infesting white-tailed deer JMed Entomol 1988 25295ndash300

Boromisa RD Grimstand PR Seroconversion rates toJamestown Canyon virus among six populations ofwhite-tailed deer (Odocoileus virginianus) in Indiana JWildl Dis 1987 2723ndash33

Brandsma AR Little SE Lockhart JM et al NovelEhrlichia organism (Rickettsiales Ehrlichieae) in white-tailed deer associated with lone star tick (Acari Ixodi-dae) parasitism J Med Entomol 1999 36190ndash194

Childs JE Paddock CD The ascendancy of Amblyommaamericanum as a vector of pathogens affecting humansin the United States Annu Rev Entomol 2003 48307ndash337

Comer JA Nicholson WL Olson JG et al Serologic test-ing for human granulocytic ehrlichiosis at a national re-ferral center J Clin Microbiol 1999 37558ndash564

Dawson JE Childs JE Biggie KL et al White-tailed deeras a potential reservoir of Ehrlichia spp J Wildl Dis1994a 30162ndash168

Dawson JE Stallknecht D Howerth EW et al Suscep-tibility of white-tailed deer (Odocoileus virginianus) toinfection with Ehrlichia chaffeensis the etiologic agent ofhuman ehrlichiosis J Clin Microbiol 1994b 322725ndash2728

Dawson JE Anderson BE Fishbein DB et al Poly-merase chain reaction evidence of Ehrlichia chaffeensisetiologic agent of human ehrlichiosis in dogs fromsoutheast Virginia Am J Vet Res 1996a 571175ndash1179

Dawson JE Warner CK Baker V et al Ehrlichia-like 16SrDNA sequence from wild white-tailed deer (Odocoileusvirginianus) J Parasitol 1996b 8252ndash58

Davidson WR Lockhart JM Stallknecht DE et al Per-sistent Ehrlichia chaffeensis infection in white-tailed deerJ Wildl Dis 2001 37538ndash546

Dumler JS Walker DH Tick-borne ehrlichiosis LancetInfect Dis 2001 021ndash28

Ewing SA Dawson JE Kocan AA et al Experimentaltransmission of Ehrlichia chaffeensis (RickettsialesEhrlichieae) among white-tailed deer by Amblyommaamericanum (Acari Ixodidae) J Med Entomol 199532368ndash374

Fletcher WO Stallknecht DE Kearney MT et al Anti-bodies to vesicular stomatitis New Jersey type virus inwhite-tailed deer on Ossabaw Island Georgia 1985ndash1989 J Wildl Dis 1991 27675ndash680

Gill JS McLean RG Neitzil DF et al Serologic analysisof white-tailed deer sera for antibodies to Borreliaburgdorferi by enzyme-linked immunosorbent assayand western blotting J Clin Microbiol 1993 31318ndash322

Ijdo JW Wu C Magnarelli LA et al Detection ofEhrlichia chaffeensis DNA in Amblyomma americanumticks in Connecticut and Rhode Island J Clin Microbiol2000 384655ndash4656

Irving RP Pinger RR Vann CN et al Distribution ofEhrlichia chaffeensis (Rickettsiales Rickettsiaeceae) inAmblyomma americanum in southern Indiana and preva-lence of E chaffeensis-reactive antibodies in white-taileddeer in Indiana and Ohio in 1998 J Med Entomol 200037595ndash600

Keirans JE Lacombe EH First records of Amblyommaamericanum Ixodes (Ixodes) dentatus and Ixodes (Cer-atixodes) uriae (Acari Ixodidae) from Maine J Parasitol1998 84629ndash631

Little SE Dawson JE Lockhart JM et al Developmentand use of specific polymerase chain reaction for thedetection of an Ehrlichia-like organism in white-taileddeer J Wildl Dis 1997 33246ndash253

Little SE Stallknecht DE Lockhart JM et al Naturalcoinfection of a white-tailed deer (Odocoileus virgini-anus) population with three Ehrlichia spp J Parasitol1998 84897ndash901

Lockhart JM Davidson WR Dawson JE et al Tempo-ral association of Amblyomma americanumwith the pres-ence of Ehrlichia chaffeensis reactive antibodies in white-tailed deer J Wildl Dis 1995 31119ndash124

Lockhart JM Davidson WR Stallknecht DE et al Site-specific geographic association between Amblyommaamericanum (Acari Ixodidae) infestations and Ehrlichiachaffeensis-reactive (Rickettsiales Ehrlichieae) antibod-ies in white-tailed deer J Med Entomol 1996 33153ndash158

Lockhart JM Davidson WR Stallknecht DE et al Iso-lation of Ehrlichia chaffeensis from wild white-tailed deer(Odocoileus virginianus) confirms their role as naturalreservoir hosts J Clin Microbiol 1997a 351681ndash1686

Lockhart JM Davidson WR Stallknecht DE et al Nat-ural history of Ehrlichia chaffeensis (RickettsialesEhrlichieae) in the Piedmont physiographic province ofGeorgia J Parasitol 1997b 83887ndash894

Magnarelli LA Anderson JF Stafford KC et al Anti-bodies to multiple tick-borne pathogens of babesiosisehrlichiosis and Lyme borreliosis in white-footed miceJ Wildl Dis 1997 33466ndash473

Mueller-Anneling L Gilchrist MJ Thorne PS Ehrlichiachaffeensis antibodies in white-tailed deer Iowa 1994and 1996 Emerg Infect Dis 2000 6397ndash400

Murphy GL Ewing SA Whitworth LC et al A molec-ular and serologic survey of Ehrlichia canis E chaffeen-sis and E ewingii in dogs and ticks from OklahomaVet Parasitol 1998 79325ndash339

Paddock CD Childs JE Ehrlichia chaffeensis a prototyp-ical emerging pathogen Clin Microbiol Rev 2003 1637ndash64

Paddock CD Folk SM Shore GM et al Infections withEhrlichia chaffeensisand Ehrlichia ewingii in persons coin-fected with human immunodeficiency virus Clin InfectDis 2001 31586ndash1594

Roland WE Everett ED Cyr TL et al Ehrlichia chaffeensisin Missouri ticks Am J Trop Med Hyg 1998 59641ndash643

Smart DL and Trainer DO Serologic evidence ofVenezuelan equine encephalitis in some wild and do-mestic populations of southern Texas J Wildl Dis 197511195ndash200

Stallknecht DE Blue JL Rollor EA et al Precipitatingantibodies to bluetongue and epizootic hemorrhagicdisease viruses in white-tailed deer in the southeasternUnited States J Wildl Dis 1991 27238ndash247

Steiert JG Gilfoy F Infection rates of Amblyomma ameri-canum and Dermacentor variabilis by Ehrlichia chaffeensis

YABSLEY ET AL206

rRNA and VLPT PCR assays on 61 deer from20 seronegative populations were all negativePresence of PCR positive deer was strongly as-sociated with seropositive populations (x2 5116 p 0001)

Of 72 culture attempts 34 (442) were lostto contamination with bacteria andor Try-panosoma cervi Seven of the remaining 38 cul-tures (184) were positive for E chaffeensisPositive cultures were obtained from two WTDfrom Greene Co AR two from Woodruff CoAR and one each from Clarke Co GA AnsonCo NC and Georgetown Co SC All isolateswere confirmed as E chaffeensis by PCR of the16S rRNA and VLPT gene targets

Relationship of age class and gender to antibody prevalence and PCR detection

Age was available on 1882 deer tested sero-logically 1086 deer from this study 517 fromDawson et al (1994) 165 from Lockhart et al

(1996) five from Little et al (1997) and 109from Yabsley et al (2002) For this combineddata set differences were not noted (x2 5 565p 5 046) in seroprevalance among age classes(Fig 2A) Similarly seroprevalence did not dif-fer between males and females (693 vs696 x2 5 0016 p 005)

Age was available for 393 deer from seropos-itive populations that were tested by PCRHigher proportions of deer were PCR positivein the younger age classes (x2 5 717 p 0001) Among deer 075 years old 531were PCR positive within the next age group(076ndash15 years) the prevalence decreased sig-nificantly to 27 (x2 5 147 p 0001) and lessthan 8 of deer 15 years old were PCR pos-itive (Fig 2B) The mean age of PCR positiveWTD was 11 years (range 025ndash4 years) In-sufficient culture isolates were made to exam-ine age-related associations however themean age for culture positive deer was 32years (range 03ndash95 years) Similar to antibody

YABSLEY ET AL200

FIG 1 Indirect immunofluorescent antibody results for Ehrlichia chaffeensis in white-tailed deer Results are from thecurrent study and from previously published work obtained in collaboration with SCWDS Darkly shaded counties pos-itive and lightly shaded counties negative for antibodies ($1128) reactive for E chaffeensis Previous serologic resultsthat differed from the results of the current study are shown as small circles within the county d previously positivefor antibodies ($1128) reactive for E chaffeensis s previously negative for antibodies reactive for E chaffeensis

prevalence PCR positivity did not differ be-tween males and females (19 vs 228 x2 5089 p 005)

Evaluation of sample size adequacy

Regarding sample size for seropositive pop-ulations post hoc statistical analysis utilizing amean 73 seroprevalence indicated that test-ing of only three or four deer gives a high prob-ability (95 and 99 respectively) of detectinga positive population (Thrusfield 1995) Al-though validation of seronegative populationsis more problematic testing of larger numbersof deer per population and repeated testing ofselected populations both produced consistentnegative results In a combined dataset fromDawson et al (1994a) and this study 250 deerfrom 17 seronegative populations (9ndash30 deer

per population) were negative for antibodies toE chaffeensis with means of 208 and 104 deertested per seronegative population respec-tively Similarly none of the 131 deer from fourpopulations consistently negative for LST in-festation had antibodies reactive to E chaffeen-sis The mean number of WTD tested per pop-ulation in this study was 62 (SD 5 61 range1ndash58) and data for populations with samplesizes less than four deer were used only if aseropositive animal was detected

Association between LST and serologic status at population level

Of 117 WTD populations examined for ticks82 were infested with LST 29 with A macula-tum (Gulf Coast tick) four with Ixodes affinisand 13 with I scapularis (blacklegged tick)

201

FIG 2 (A) Prevalence of antibodies reactive to E chaffeensis in WTD among age classes Numbers in bars representnumber of deer tested (B) Prevalence of PCR-positive WTD among age classes Numbers in bars represent numberof deer tested and different letters represent significant differences at p 005

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

Eighty of the 82 (976) populations infestedwith LST contained at least one seropositiveWTD while only four of 35 (114) populationsnot infested with LST contained seropositivedeer One LST-negative population (Harris CoGA) was both seronegative and tick negativein 1986 however in 2002 the deer populationwas seropositive but the LST status was notevaluated The prevalence of antibodies reac-tive to E chaffeensis was significantly higheramong LST infested populations of WTD thanpopulations not infested by LST (x2 5 899 p 0001) None of the six populations infestedonly with A maculatum contained seropositivedeer there were insufficient populations withI affinis or I scapularis alone to permit evalua-tion

Stability and spread of E chaffeensis and LST among WTD populations

Fifty-seven populations which were testedduring previous studies (Dawson et al 1994aLockhart et al 1996) were retested during thisstudy The serologic status for 50 (88) of thesepopulations remained unchanged 35 seropos-itive populations remained seropositive and 15seronegative populations remained seronega-tive Six populations which previously testedseronegative were seropositive during this

study Three of these representing Anson CoNC Haywood Co TN and Stewart Co GAhad documented invasion of LST between thefirst reported testing and this study One Texaspopulation (Travis Co) was seropositive (n 57 714) when tested in 1992 (Dawson et al1994a) but was seronegative when sampled(n 5 10) in 2002 for this study The LST statusfor this county was not known for either sam-pling period

Among the 11 populations where LST infes-tation and serologic status were monitored repeatedly over time there was complete con-cordance between LST presence and seroposi-tive deer (Table 2) All three LST positive pop-ulations were consistently seropositive eachyear tested The mean seroprevalence for thesethree populations was 83 and two of thesepopulations (Jones Co GA and Ashley CoAR) were confirmed PCR positive for E chaf-feensis None of the four populations where LSTwas consistently absent contained seropositivedeer at any sampling time Among the fourpopulations where LST status changed fromabsent to present seropositive deer were onlydetected following the appearance of LST Themean prevalence of antibodies reactive to Echaffeensis in the most recently tested year wassignificantly higher for populations infestedwith LST for multiple years (mean 5 100)

YABSLEY ET AL202

TABLE 2 SEROLOGIC AND TICK SURVEY RESULTS FROM SEVEN POPULATIONS OF WTD COLLECTED DURING

MULTIPLE TIME PERIODS AND TESTED FOR ANTIBODIES REACTIVE FOR E CHAFFEENSIS

No of Tickyears status No of deer Mean IFA prevalence

Site Years sampled (years) tested (range)

Sites positive for ticksJones Co GA 1979ndash2001 6 1 (6) 41 92 (80ndash100)Ashley Co AR 1982ndash2001 7 1 (7) 37 84 (71ndash100)Franklin Co FL 1984ndash1996 5 1 (5) 25 68 (40ndash100)

Sites negative for ticksFloyd Co GA 1973ndash2001 6 2 (6) 58 0Tyler Co WV 1985ndash2000 4 2 (4) 21 0Hardy Co WV 1977ndash1999 7 2 (7) 37 0Monroe Co FL 1992ndash2001 4 2 (4) 15 0

Sites where tick status changedAnson Co NC 1987 2001 2 2 1a 10 0 40bConcordia Pa LA 1986 1991 1999 3 2 1 1 22 0 38 60Haywood Co TN 1989 1994 1998 3 2 2 1 16 0 0 20Stewart Co GA 1986 1998 2 2 1 10 0 20

aTick status at sequential sampling periodsbPrevalence of IFA seropositive deer at sequential sampling periods

compared to populations only recently infested(mean 5 538 x2 5 155 p 0001) One re-cently infested population (Anson Co NC)was confirmed by isolation of E chaffeensisfrom a single WTD in 2001

DISCUSSION

The overarching goal of this study was toimplement and evaluate a prototype surveil-lance system using WTD as natural sentinels todetermine the geographic distribution of Echaffeensis across most of the suspected HMEendemic region of the United States To thisend the data obtained demonstrate the fol-lowing critical attributes (1) serologic findingsreflect E chaffeensis infection (2) effective sur-veillance can be achieved with small samplesizes (3) any deer 6 month old is suitable forsurveillance purposes (4) enzootic and consis-tently negative locales were identified repeat-edly across time (5) surveillance of deer hasthe ability to detect spread to new locales and(6) serologic molecular and culture diagnosticfindings among deer could be related to pres-ence of the principal tick vector A americanumThese findings compliment and expand on arecent WTD serologic surveillance effort withinIowa (Mueller-Anneling et al 2000) howeverthat study exclusively used serologic assaysand did not include either confirmatory mi-crobiological diagnostics or contemporaneoustick vector monitoring Collectively these twostudies confirm the effectiveness of this proto-typical surveillance system for monitoring thedistribution of this emerging human pathogen

Although several studies have examined theprevalence and distribution of E chaffeensis an-tibodies among many southeastern and southcentral WTD populations (Dawson et al 1994aLockhart et al 1996 Little et al 1997 Yabsleyet al 2002) the extensive additional testingfilled many geographic gaps and provide amore complete fine scale distribution of E chaf-feensis among WTD throughout the regions ofthe United States with the highest HME risk(Paddock and Childs 2003) The combined dataevaluated herein disclosed a wide distributionof seropositive WTD populations across all ofthe 18 states represented except for West Vir-

ginia however some distributional limits alsowere discerned including a western boundaryextending across Kansas Oklahoma andTexas a southern boundary across peninsularFlorida and a large cluster of many seronega-tive populations centered along the Ap-palachian Mountains (Fig 1) Other prior stud-ies of WTD have documented an apparentnorthern range limit for E chaffeensis in south-ern portions of Iowa Illinois Indiana and Ohio(Dawson et al 1994a Irving et al 2000 Mueller-Anneling et al 2000) Along the East coast sur-veys have indicated that A americanum and Echaffeensis are present as far north as Connecti-cut Maine and Rhode Island (Keirans and La-combe 1998 Ijdo et al 2000)

The accuracy of diagnostic assays utilized iscritical to any pathogen surveillance system In-fection with E chaffeensis was confirmed innearly half (46) of 122 seropositive WTD pop-ulations tested by PCR and was supplementedby isolation from others In contrast PCR evi-dence of infection was not detected in any of20 seronegative WTD populations Thus PCRand culture testing applied at the populationlevel were effective at confirming infection inmany seropositive populations and demon-strating a strong association between the sero-logic status of populations and infection withE chaffeensis

Because WTD in these regions are known to be infected with three other related rickettsiae(E ewingii A phagocytophilum and an Ana-plasma sp [WTD-agent]) in addition to E chaf-feensis (Dawson et al 1996b Little et al 1998Brandsma et al 1999 Magnarelli et al 1999Yabsley et al 2002) the potential for serologiccross-reaction is an important considerationAlthough antibodies reactive with E chaffeensisin WTD from Maryland have been confirmedby immunoblotting using the 28- to 29-kDa anti-gens of E chaffeensis (Walls et al 1998) the ex-tent of serologic cross-reactions among thesefour species is not fully understood (Dumlerand Walker 2001) However data from otherstudies suggests that infection with these otherrickettsiae may not commonly result in pro-duction of antibodies that cross-react with Echaffeensis Cross-reactions between E chaffeen-sis and E ewingii have been noted in infectedhumans and dogs however not all E ewingii

203EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

infected dogs or humans develop antibodies toE chaffeensis antigens (Murphy et al 1998 Pad-dock et al 2001) Similarly a WTD fawn ex-perimentally infected with E ewingii did not de-velop antibodies reactive with E chaffeensis(Yabsley et al 2002) Cross-reactions between Echaffeensis and A phagocytophilum have been re-ported in humans and may complicate diagno-sis (Comer et al 1999) Far fewer WTD in thesoutheastern states have antibodies reactive toA phagocytophilum (25) than to E chaffeensisantigens however although some deer reactedto both E chaffeensis and A phagocytophilum ator above the 1128 cut-off many also reacted toonly one of these antigens at titers $1128 (Lit-tle et al 1998 Walls et al 1998 VG Dugan andMJ Yabsley unpublished data) The high per-centage of populations where seropositive orseronegative status was confirmed by PCR orculture together with data from these otherstudies provide considerable evidence that theseroreactivity reported herein largely repre-sents specific seroconversion to E chaffeensis

The distribution of antibodies within WTDage and gender categories was investigated be-cause these variables are an important consid-eration for a WTD natural sentinel systemOnly a limited investigation of geometric meantiters among age categories has been reportedfor WTD (Lockhart et al 1995) and any signif-icant influence of age or gender on the occur-rence of antibodies would require samplingstratified according to these host attributes Im-portantly differences in antibody prevalencewere not noted among age or gender categoriesindicating that all WTD particularly animals $6 mo of age are suitable for use in a WTD sur-veillance system In contrast to the high stableprevalence of antibodies among age classes(Fig 2A) PCR evidence of rickettsemia de-clined with age (Fig 2B) This pattern conformsto proposed infection dynamics in naturally in-fected WTD whereby animals from approxi-mately 6 months to 15 years of age are morelikely to be rickettsemic than older adults (Pad-dock and Childs 2003 Dawson et al 1994Lockhart et al 1997a Davidson et al 2001) Inthe present study the prevalence of PCR pos-itivity dropped dramatically from over half(53) of WTD younger than 075 years to lessthan 8 of WTD older than 15 years The ma-

jority of deer tested from some seropositivepopulations were 15 years and only deer15 years from other populations were avail-able for PCR testing which may explain whynot all seropositive populations were con-firmed by PCR Because the probability of aWTD being rickettsemic declines with ageserology represents a better surveillance toolthan PCR and under natural conditions of re-exposure to ticks titers likely do not declineover time as often occurred in single exposureexperimental infections of WTD (Dawson et al1994b Ewing et al 1995 Davidson et al 2001)

Previous serologic surveys of E chaffeensis inWTD (Dawson et al 1994a Lockhart et al 1996Little et al 1997 Yabsley et al 2002) have uti-lized relatively small sample sizes per popula-tion but except for one (Mueller-Anneling etal 2000) have not addressed the adequacy ofthese sample sizes to correctly classify infectedand uninfected populations Data from the cur-rent and prior studies (Dawson et al 1994aLockhart et al 1996 Little et al 1997) haveshown that infected WTD populations in thesoutheastern and southcentral United Statestypically have high (70) seroprevalencesCalculations (Thrusfield 1995) based on a mean73 seroprevalence indicated that use of a sam-ple size of five WTD per population should re-liably detect nearly all positive populations Al-though it was beyond the scope of this studyto test enough deer to detect very low popula-tion prevalences larger sample sizes and re-peated sampling of seronegative populationsdid not reveal seropositive populations thatwould have been misclassified as seronegativebased on smaller sample sizes A chance of mis-classifying newly infected populations of WTDexists because seroprevalence may be low ini-tially however field evidence shows that anti-body prevalence can increase rapidly once apopulation becomes infected with E chaffeensis(Lockhart et al 1995) Utilizing per county sam-ple sizes equivalent to this study statewideserotesting of WTD from Iowa demonstrated aclear north-south gradient of seroprevalencethat corresponded to key epidemiologic factors(eg distributions of ticks deer and habitats)(Mueller-Anneling et al 2000)

Because the distributions of pathogens oftenare reflected by some combination of both sta-

YABSLEY ET AL204

tic and changing geographic distributions animportant attribute of an effective surveillancesystem is the ability to identify both persis-tently enzootic locales and spread to new loca-tions Testing of WTD from LST infested pop-ulations over the span of 20 years showed thatonce this biologically important vector is pres-ent and WTD populations are infected with Echaffeensis there is a high probability that thepopulations will remain infected and that WTDsentinels reflect these enzootic conditions Con-versely populations with no history of LST in-festation remained consistently seronegativewhen tested multiple times These results arecompatible with a strong site-specific associa-tion between E chaffeensis antibodies and thepresence of LST (Lockhart et al 1996) Of in-terest each of two populations (Choctaw CoAL and Bolivar Co MS) that were earlier re-ported as seropositive but LST negative (Lock-hart et al 1996) were shown in this study to beboth LST positive and seropositive for E chaf-feensis The rare failure to find LST in seropos-itive WTD populations may have occurred be-cause many collections were made during thefall hunting season when LST infestations areundergoing rapid seasonal declines (Allan2001)

Four instances of E chaffeensis spread to newlocations were detected by monitoring WTDfor multiple years (Table 2) At each of these lo-cations the change in population serologic sta-tus from negative to positive coincided withdocumented appearance of LST These findingsare very similar to those from a 12-year-longmonitoring of a WTD population that demon-strated the introduction of E chaffeensis was at-tributable to the establishment of LST (Lock-hart et al 1995) Because all of these changes toseropositive status were linked to the principalvector they provide further evidence that thissurveillance system reflects important epi-demiologic factors and that it should discernchanging local HME risks

In summary the biological reasons a sentinelWTD surveillance system functions effectivelyare that the force of transmission of E chaffeen-sis is focused on WTD and that aspects of WTDbiology allow attribution of infection to a spe-cific geographic location From an operationalperspective the system has the desirable at-

tributes of applicability on a wide geographicscale and logistic feasibility by using ldquofreerdquo(hunter-harvested) samples The present workwas done in a research context however it ispossible to implement sentinel WTD serologicsurveillance for E chaffeensis in an operationalpublic health context as clearly demonstratedpreviously in Iowa (Mueller-Anneling et al2000)

ACKNOWLEDGMENTS

We are indebted to over 100 individuals rep-resenting state wildlife agency personnelSCWDS staff and white-tailed deer researchersfor help in field collection of samples but es-pecially the following individuals for coordi-nating specific collections Chris Cook (AL)John Morgan (FL) Lloyd Fox (KS) Jon Gassettand David Yancy (KY) Jeff Beringer (MO)Douglas Roscoe (NJ) Evin Stanford (NC) AAlan Kocan (OK) Charles Ruth Jr (SC) BenLayton (TN) Matt Knox (VA) and Jim Crum(WV) Sarah Cross Page Luttrell and DarrellKavanaugh from SCWDS provided excellentfield and laboratory help Work was supportedprimarily by the National Institutes of Allergyand Infectious Diseases (5 R01 AI044235-02)Further support was provided by the FederalAid to Wildlife Restoration Act (50 Stat 917)and through sponsorship from fish and wildlifeagencies in Alabama Arkansas Florida Geor-gia Kansas Kentucky Louisiana MarylandMississippi Missouri North Carolina Okla-homa Puerto Rico South Carolina TennesseeVirginia and West Virginia

REFERENCES

Allan SA Ticks (Class Arachnida Order Acarina) InSamuel WM Pybus MJ and Kocan AA ed ParasiticDiseases of Wild Mammals Ames IA Iowa State Uni-versity Press 2001 72ndash106

Anderson BE Sumner JW Dawson JE et al Detection ofthe etiologic agent of human ehrlichiosis by polymerasechain reaction J Clin Microbiol 1992 30775ndash780

Anderson BE Sims KG Olson JG et al Amblyommaamericanum a potential vector of human ehrlichiosisAm J Trop Med Hyg 1993 49239ndash244

Bloemer SR Zimmerman RH Fairbanks K Abundanceattachment sites and density estimators of lone star

205EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

ticks (Acari Ixodidae) infesting white-tailed deer JMed Entomol 1988 25295ndash300

Boromisa RD Grimstand PR Seroconversion rates toJamestown Canyon virus among six populations ofwhite-tailed deer (Odocoileus virginianus) in Indiana JWildl Dis 1987 2723ndash33

Brandsma AR Little SE Lockhart JM et al NovelEhrlichia organism (Rickettsiales Ehrlichieae) in white-tailed deer associated with lone star tick (Acari Ixodi-dae) parasitism J Med Entomol 1999 36190ndash194

Childs JE Paddock CD The ascendancy of Amblyommaamericanum as a vector of pathogens affecting humansin the United States Annu Rev Entomol 2003 48307ndash337

Comer JA Nicholson WL Olson JG et al Serologic test-ing for human granulocytic ehrlichiosis at a national re-ferral center J Clin Microbiol 1999 37558ndash564

Dawson JE Childs JE Biggie KL et al White-tailed deeras a potential reservoir of Ehrlichia spp J Wildl Dis1994a 30162ndash168

Dawson JE Stallknecht D Howerth EW et al Suscep-tibility of white-tailed deer (Odocoileus virginianus) toinfection with Ehrlichia chaffeensis the etiologic agent ofhuman ehrlichiosis J Clin Microbiol 1994b 322725ndash2728

Dawson JE Anderson BE Fishbein DB et al Poly-merase chain reaction evidence of Ehrlichia chaffeensisetiologic agent of human ehrlichiosis in dogs fromsoutheast Virginia Am J Vet Res 1996a 571175ndash1179

Dawson JE Warner CK Baker V et al Ehrlichia-like 16SrDNA sequence from wild white-tailed deer (Odocoileusvirginianus) J Parasitol 1996b 8252ndash58

Davidson WR Lockhart JM Stallknecht DE et al Per-sistent Ehrlichia chaffeensis infection in white-tailed deerJ Wildl Dis 2001 37538ndash546

Dumler JS Walker DH Tick-borne ehrlichiosis LancetInfect Dis 2001 021ndash28

Ewing SA Dawson JE Kocan AA et al Experimentaltransmission of Ehrlichia chaffeensis (RickettsialesEhrlichieae) among white-tailed deer by Amblyommaamericanum (Acari Ixodidae) J Med Entomol 199532368ndash374

Fletcher WO Stallknecht DE Kearney MT et al Anti-bodies to vesicular stomatitis New Jersey type virus inwhite-tailed deer on Ossabaw Island Georgia 1985ndash1989 J Wildl Dis 1991 27675ndash680

Gill JS McLean RG Neitzil DF et al Serologic analysisof white-tailed deer sera for antibodies to Borreliaburgdorferi by enzyme-linked immunosorbent assayand western blotting J Clin Microbiol 1993 31318ndash322

Ijdo JW Wu C Magnarelli LA et al Detection ofEhrlichia chaffeensis DNA in Amblyomma americanumticks in Connecticut and Rhode Island J Clin Microbiol2000 384655ndash4656

Irving RP Pinger RR Vann CN et al Distribution ofEhrlichia chaffeensis (Rickettsiales Rickettsiaeceae) inAmblyomma americanum in southern Indiana and preva-lence of E chaffeensis-reactive antibodies in white-taileddeer in Indiana and Ohio in 1998 J Med Entomol 200037595ndash600

Keirans JE Lacombe EH First records of Amblyommaamericanum Ixodes (Ixodes) dentatus and Ixodes (Cer-atixodes) uriae (Acari Ixodidae) from Maine J Parasitol1998 84629ndash631

Little SE Dawson JE Lockhart JM et al Developmentand use of specific polymerase chain reaction for thedetection of an Ehrlichia-like organism in white-taileddeer J Wildl Dis 1997 33246ndash253

Little SE Stallknecht DE Lockhart JM et al Naturalcoinfection of a white-tailed deer (Odocoileus virgini-anus) population with three Ehrlichia spp J Parasitol1998 84897ndash901

Lockhart JM Davidson WR Dawson JE et al Tempo-ral association of Amblyomma americanumwith the pres-ence of Ehrlichia chaffeensis reactive antibodies in white-tailed deer J Wildl Dis 1995 31119ndash124

Lockhart JM Davidson WR Stallknecht DE et al Site-specific geographic association between Amblyommaamericanum (Acari Ixodidae) infestations and Ehrlichiachaffeensis-reactive (Rickettsiales Ehrlichieae) antibod-ies in white-tailed deer J Med Entomol 1996 33153ndash158

Lockhart JM Davidson WR Stallknecht DE et al Iso-lation of Ehrlichia chaffeensis from wild white-tailed deer(Odocoileus virginianus) confirms their role as naturalreservoir hosts J Clin Microbiol 1997a 351681ndash1686

Lockhart JM Davidson WR Stallknecht DE et al Nat-ural history of Ehrlichia chaffeensis (RickettsialesEhrlichieae) in the Piedmont physiographic province ofGeorgia J Parasitol 1997b 83887ndash894

Magnarelli LA Anderson JF Stafford KC et al Anti-bodies to multiple tick-borne pathogens of babesiosisehrlichiosis and Lyme borreliosis in white-footed miceJ Wildl Dis 1997 33466ndash473

Mueller-Anneling L Gilchrist MJ Thorne PS Ehrlichiachaffeensis antibodies in white-tailed deer Iowa 1994and 1996 Emerg Infect Dis 2000 6397ndash400

Murphy GL Ewing SA Whitworth LC et al A molec-ular and serologic survey of Ehrlichia canis E chaffeen-sis and E ewingii in dogs and ticks from OklahomaVet Parasitol 1998 79325ndash339

Paddock CD Childs JE Ehrlichia chaffeensis a prototyp-ical emerging pathogen Clin Microbiol Rev 2003 1637ndash64

Paddock CD Folk SM Shore GM et al Infections withEhrlichia chaffeensisand Ehrlichia ewingii in persons coin-fected with human immunodeficiency virus Clin InfectDis 2001 31586ndash1594

Roland WE Everett ED Cyr TL et al Ehrlichia chaffeensisin Missouri ticks Am J Trop Med Hyg 1998 59641ndash643

Smart DL and Trainer DO Serologic evidence ofVenezuelan equine encephalitis in some wild and do-mestic populations of southern Texas J Wildl Dis 197511195ndash200

Stallknecht DE Blue JL Rollor EA et al Precipitatingantibodies to bluetongue and epizootic hemorrhagicdisease viruses in white-tailed deer in the southeasternUnited States J Wildl Dis 1991 27238ndash247

Steiert JG Gilfoy F Infection rates of Amblyomma ameri-canum and Dermacentor variabilis by Ehrlichia chaffeensis

YABSLEY ET AL206

prevalence PCR positivity did not differ be-tween males and females (19 vs 228 x2 5089 p 005)

Evaluation of sample size adequacy

Regarding sample size for seropositive pop-ulations post hoc statistical analysis utilizing amean 73 seroprevalence indicated that test-ing of only three or four deer gives a high prob-ability (95 and 99 respectively) of detectinga positive population (Thrusfield 1995) Al-though validation of seronegative populationsis more problematic testing of larger numbersof deer per population and repeated testing ofselected populations both produced consistentnegative results In a combined dataset fromDawson et al (1994a) and this study 250 deerfrom 17 seronegative populations (9ndash30 deer

per population) were negative for antibodies toE chaffeensis with means of 208 and 104 deertested per seronegative population respec-tively Similarly none of the 131 deer from fourpopulations consistently negative for LST in-festation had antibodies reactive to E chaffeen-sis The mean number of WTD tested per pop-ulation in this study was 62 (SD 5 61 range1ndash58) and data for populations with samplesizes less than four deer were used only if aseropositive animal was detected

Association between LST and serologic status at population level

Of 117 WTD populations examined for ticks82 were infested with LST 29 with A macula-tum (Gulf Coast tick) four with Ixodes affinisand 13 with I scapularis (blacklegged tick)

201

FIG 2 (A) Prevalence of antibodies reactive to E chaffeensis in WTD among age classes Numbers in bars representnumber of deer tested (B) Prevalence of PCR-positive WTD among age classes Numbers in bars represent numberof deer tested and different letters represent significant differences at p 005

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

Eighty of the 82 (976) populations infestedwith LST contained at least one seropositiveWTD while only four of 35 (114) populationsnot infested with LST contained seropositivedeer One LST-negative population (Harris CoGA) was both seronegative and tick negativein 1986 however in 2002 the deer populationwas seropositive but the LST status was notevaluated The prevalence of antibodies reac-tive to E chaffeensis was significantly higheramong LST infested populations of WTD thanpopulations not infested by LST (x2 5 899 p 0001) None of the six populations infestedonly with A maculatum contained seropositivedeer there were insufficient populations withI affinis or I scapularis alone to permit evalua-tion

Stability and spread of E chaffeensis and LST among WTD populations

Fifty-seven populations which were testedduring previous studies (Dawson et al 1994aLockhart et al 1996) were retested during thisstudy The serologic status for 50 (88) of thesepopulations remained unchanged 35 seropos-itive populations remained seropositive and 15seronegative populations remained seronega-tive Six populations which previously testedseronegative were seropositive during this

study Three of these representing Anson CoNC Haywood Co TN and Stewart Co GAhad documented invasion of LST between thefirst reported testing and this study One Texaspopulation (Travis Co) was seropositive (n 57 714) when tested in 1992 (Dawson et al1994a) but was seronegative when sampled(n 5 10) in 2002 for this study The LST statusfor this county was not known for either sam-pling period

Among the 11 populations where LST infes-tation and serologic status were monitored repeatedly over time there was complete con-cordance between LST presence and seroposi-tive deer (Table 2) All three LST positive pop-ulations were consistently seropositive eachyear tested The mean seroprevalence for thesethree populations was 83 and two of thesepopulations (Jones Co GA and Ashley CoAR) were confirmed PCR positive for E chaf-feensis None of the four populations where LSTwas consistently absent contained seropositivedeer at any sampling time Among the fourpopulations where LST status changed fromabsent to present seropositive deer were onlydetected following the appearance of LST Themean prevalence of antibodies reactive to Echaffeensis in the most recently tested year wassignificantly higher for populations infestedwith LST for multiple years (mean 5 100)

YABSLEY ET AL202

TABLE 2 SEROLOGIC AND TICK SURVEY RESULTS FROM SEVEN POPULATIONS OF WTD COLLECTED DURING

MULTIPLE TIME PERIODS AND TESTED FOR ANTIBODIES REACTIVE FOR E CHAFFEENSIS

No of Tickyears status No of deer Mean IFA prevalence

Site Years sampled (years) tested (range)

Sites positive for ticksJones Co GA 1979ndash2001 6 1 (6) 41 92 (80ndash100)Ashley Co AR 1982ndash2001 7 1 (7) 37 84 (71ndash100)Franklin Co FL 1984ndash1996 5 1 (5) 25 68 (40ndash100)

Sites negative for ticksFloyd Co GA 1973ndash2001 6 2 (6) 58 0Tyler Co WV 1985ndash2000 4 2 (4) 21 0Hardy Co WV 1977ndash1999 7 2 (7) 37 0Monroe Co FL 1992ndash2001 4 2 (4) 15 0

Sites where tick status changedAnson Co NC 1987 2001 2 2 1a 10 0 40bConcordia Pa LA 1986 1991 1999 3 2 1 1 22 0 38 60Haywood Co TN 1989 1994 1998 3 2 2 1 16 0 0 20Stewart Co GA 1986 1998 2 2 1 10 0 20

aTick status at sequential sampling periodsbPrevalence of IFA seropositive deer at sequential sampling periods

compared to populations only recently infested(mean 5 538 x2 5 155 p 0001) One re-cently infested population (Anson Co NC)was confirmed by isolation of E chaffeensisfrom a single WTD in 2001

DISCUSSION

The overarching goal of this study was toimplement and evaluate a prototype surveil-lance system using WTD as natural sentinels todetermine the geographic distribution of Echaffeensis across most of the suspected HMEendemic region of the United States To thisend the data obtained demonstrate the fol-lowing critical attributes (1) serologic findingsreflect E chaffeensis infection (2) effective sur-veillance can be achieved with small samplesizes (3) any deer 6 month old is suitable forsurveillance purposes (4) enzootic and consis-tently negative locales were identified repeat-edly across time (5) surveillance of deer hasthe ability to detect spread to new locales and(6) serologic molecular and culture diagnosticfindings among deer could be related to pres-ence of the principal tick vector A americanumThese findings compliment and expand on arecent WTD serologic surveillance effort withinIowa (Mueller-Anneling et al 2000) howeverthat study exclusively used serologic assaysand did not include either confirmatory mi-crobiological diagnostics or contemporaneoustick vector monitoring Collectively these twostudies confirm the effectiveness of this proto-typical surveillance system for monitoring thedistribution of this emerging human pathogen

Although several studies have examined theprevalence and distribution of E chaffeensis an-tibodies among many southeastern and southcentral WTD populations (Dawson et al 1994aLockhart et al 1996 Little et al 1997 Yabsleyet al 2002) the extensive additional testingfilled many geographic gaps and provide amore complete fine scale distribution of E chaf-feensis among WTD throughout the regions ofthe United States with the highest HME risk(Paddock and Childs 2003) The combined dataevaluated herein disclosed a wide distributionof seropositive WTD populations across all ofthe 18 states represented except for West Vir-

ginia however some distributional limits alsowere discerned including a western boundaryextending across Kansas Oklahoma andTexas a southern boundary across peninsularFlorida and a large cluster of many seronega-tive populations centered along the Ap-palachian Mountains (Fig 1) Other prior stud-ies of WTD have documented an apparentnorthern range limit for E chaffeensis in south-ern portions of Iowa Illinois Indiana and Ohio(Dawson et al 1994a Irving et al 2000 Mueller-Anneling et al 2000) Along the East coast sur-veys have indicated that A americanum and Echaffeensis are present as far north as Connecti-cut Maine and Rhode Island (Keirans and La-combe 1998 Ijdo et al 2000)

The accuracy of diagnostic assays utilized iscritical to any pathogen surveillance system In-fection with E chaffeensis was confirmed innearly half (46) of 122 seropositive WTD pop-ulations tested by PCR and was supplementedby isolation from others In contrast PCR evi-dence of infection was not detected in any of20 seronegative WTD populations Thus PCRand culture testing applied at the populationlevel were effective at confirming infection inmany seropositive populations and demon-strating a strong association between the sero-logic status of populations and infection withE chaffeensis

Because WTD in these regions are known to be infected with three other related rickettsiae(E ewingii A phagocytophilum and an Ana-plasma sp [WTD-agent]) in addition to E chaf-feensis (Dawson et al 1996b Little et al 1998Brandsma et al 1999 Magnarelli et al 1999Yabsley et al 2002) the potential for serologiccross-reaction is an important considerationAlthough antibodies reactive with E chaffeensisin WTD from Maryland have been confirmedby immunoblotting using the 28- to 29-kDa anti-gens of E chaffeensis (Walls et al 1998) the ex-tent of serologic cross-reactions among thesefour species is not fully understood (Dumlerand Walker 2001) However data from otherstudies suggests that infection with these otherrickettsiae may not commonly result in pro-duction of antibodies that cross-react with Echaffeensis Cross-reactions between E chaffeen-sis and E ewingii have been noted in infectedhumans and dogs however not all E ewingii

203EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

infected dogs or humans develop antibodies toE chaffeensis antigens (Murphy et al 1998 Pad-dock et al 2001) Similarly a WTD fawn ex-perimentally infected with E ewingii did not de-velop antibodies reactive with E chaffeensis(Yabsley et al 2002) Cross-reactions between Echaffeensis and A phagocytophilum have been re-ported in humans and may complicate diagno-sis (Comer et al 1999) Far fewer WTD in thesoutheastern states have antibodies reactive toA phagocytophilum (25) than to E chaffeensisantigens however although some deer reactedto both E chaffeensis and A phagocytophilum ator above the 1128 cut-off many also reacted toonly one of these antigens at titers $1128 (Lit-tle et al 1998 Walls et al 1998 VG Dugan andMJ Yabsley unpublished data) The high per-centage of populations where seropositive orseronegative status was confirmed by PCR orculture together with data from these otherstudies provide considerable evidence that theseroreactivity reported herein largely repre-sents specific seroconversion to E chaffeensis

The distribution of antibodies within WTDage and gender categories was investigated be-cause these variables are an important consid-eration for a WTD natural sentinel systemOnly a limited investigation of geometric meantiters among age categories has been reportedfor WTD (Lockhart et al 1995) and any signif-icant influence of age or gender on the occur-rence of antibodies would require samplingstratified according to these host attributes Im-portantly differences in antibody prevalencewere not noted among age or gender categoriesindicating that all WTD particularly animals $6 mo of age are suitable for use in a WTD sur-veillance system In contrast to the high stableprevalence of antibodies among age classes(Fig 2A) PCR evidence of rickettsemia de-clined with age (Fig 2B) This pattern conformsto proposed infection dynamics in naturally in-fected WTD whereby animals from approxi-mately 6 months to 15 years of age are morelikely to be rickettsemic than older adults (Pad-dock and Childs 2003 Dawson et al 1994Lockhart et al 1997a Davidson et al 2001) Inthe present study the prevalence of PCR pos-itivity dropped dramatically from over half(53) of WTD younger than 075 years to lessthan 8 of WTD older than 15 years The ma-

jority of deer tested from some seropositivepopulations were 15 years and only deer15 years from other populations were avail-able for PCR testing which may explain whynot all seropositive populations were con-firmed by PCR Because the probability of aWTD being rickettsemic declines with ageserology represents a better surveillance toolthan PCR and under natural conditions of re-exposure to ticks titers likely do not declineover time as often occurred in single exposureexperimental infections of WTD (Dawson et al1994b Ewing et al 1995 Davidson et al 2001)

Previous serologic surveys of E chaffeensis inWTD (Dawson et al 1994a Lockhart et al 1996Little et al 1997 Yabsley et al 2002) have uti-lized relatively small sample sizes per popula-tion but except for one (Mueller-Anneling etal 2000) have not addressed the adequacy ofthese sample sizes to correctly classify infectedand uninfected populations Data from the cur-rent and prior studies (Dawson et al 1994aLockhart et al 1996 Little et al 1997) haveshown that infected WTD populations in thesoutheastern and southcentral United Statestypically have high (70) seroprevalencesCalculations (Thrusfield 1995) based on a mean73 seroprevalence indicated that use of a sam-ple size of five WTD per population should re-liably detect nearly all positive populations Al-though it was beyond the scope of this studyto test enough deer to detect very low popula-tion prevalences larger sample sizes and re-peated sampling of seronegative populationsdid not reveal seropositive populations thatwould have been misclassified as seronegativebased on smaller sample sizes A chance of mis-classifying newly infected populations of WTDexists because seroprevalence may be low ini-tially however field evidence shows that anti-body prevalence can increase rapidly once apopulation becomes infected with E chaffeensis(Lockhart et al 1995) Utilizing per county sam-ple sizes equivalent to this study statewideserotesting of WTD from Iowa demonstrated aclear north-south gradient of seroprevalencethat corresponded to key epidemiologic factors(eg distributions of ticks deer and habitats)(Mueller-Anneling et al 2000)

Because the distributions of pathogens oftenare reflected by some combination of both sta-

YABSLEY ET AL204

tic and changing geographic distributions animportant attribute of an effective surveillancesystem is the ability to identify both persis-tently enzootic locales and spread to new loca-tions Testing of WTD from LST infested pop-ulations over the span of 20 years showed thatonce this biologically important vector is pres-ent and WTD populations are infected with Echaffeensis there is a high probability that thepopulations will remain infected and that WTDsentinels reflect these enzootic conditions Con-versely populations with no history of LST in-festation remained consistently seronegativewhen tested multiple times These results arecompatible with a strong site-specific associa-tion between E chaffeensis antibodies and thepresence of LST (Lockhart et al 1996) Of in-terest each of two populations (Choctaw CoAL and Bolivar Co MS) that were earlier re-ported as seropositive but LST negative (Lock-hart et al 1996) were shown in this study to beboth LST positive and seropositive for E chaf-feensis The rare failure to find LST in seropos-itive WTD populations may have occurred be-cause many collections were made during thefall hunting season when LST infestations areundergoing rapid seasonal declines (Allan2001)

Four instances of E chaffeensis spread to newlocations were detected by monitoring WTDfor multiple years (Table 2) At each of these lo-cations the change in population serologic sta-tus from negative to positive coincided withdocumented appearance of LST These findingsare very similar to those from a 12-year-longmonitoring of a WTD population that demon-strated the introduction of E chaffeensis was at-tributable to the establishment of LST (Lock-hart et al 1995) Because all of these changes toseropositive status were linked to the principalvector they provide further evidence that thissurveillance system reflects important epi-demiologic factors and that it should discernchanging local HME risks

In summary the biological reasons a sentinelWTD surveillance system functions effectivelyare that the force of transmission of E chaffeen-sis is focused on WTD and that aspects of WTDbiology allow attribution of infection to a spe-cific geographic location From an operationalperspective the system has the desirable at-

tributes of applicability on a wide geographicscale and logistic feasibility by using ldquofreerdquo(hunter-harvested) samples The present workwas done in a research context however it ispossible to implement sentinel WTD serologicsurveillance for E chaffeensis in an operationalpublic health context as clearly demonstratedpreviously in Iowa (Mueller-Anneling et al2000)

ACKNOWLEDGMENTS

We are indebted to over 100 individuals rep-resenting state wildlife agency personnelSCWDS staff and white-tailed deer researchersfor help in field collection of samples but es-pecially the following individuals for coordi-nating specific collections Chris Cook (AL)John Morgan (FL) Lloyd Fox (KS) Jon Gassettand David Yancy (KY) Jeff Beringer (MO)Douglas Roscoe (NJ) Evin Stanford (NC) AAlan Kocan (OK) Charles Ruth Jr (SC) BenLayton (TN) Matt Knox (VA) and Jim Crum(WV) Sarah Cross Page Luttrell and DarrellKavanaugh from SCWDS provided excellentfield and laboratory help Work was supportedprimarily by the National Institutes of Allergyand Infectious Diseases (5 R01 AI044235-02)Further support was provided by the FederalAid to Wildlife Restoration Act (50 Stat 917)and through sponsorship from fish and wildlifeagencies in Alabama Arkansas Florida Geor-gia Kansas Kentucky Louisiana MarylandMississippi Missouri North Carolina Okla-homa Puerto Rico South Carolina TennesseeVirginia and West Virginia

REFERENCES

Allan SA Ticks (Class Arachnida Order Acarina) InSamuel WM Pybus MJ and Kocan AA ed ParasiticDiseases of Wild Mammals Ames IA Iowa State Uni-versity Press 2001 72ndash106

Anderson BE Sumner JW Dawson JE et al Detection ofthe etiologic agent of human ehrlichiosis by polymerasechain reaction J Clin Microbiol 1992 30775ndash780

Anderson BE Sims KG Olson JG et al Amblyommaamericanum a potential vector of human ehrlichiosisAm J Trop Med Hyg 1993 49239ndash244

Bloemer SR Zimmerman RH Fairbanks K Abundanceattachment sites and density estimators of lone star

205EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

ticks (Acari Ixodidae) infesting white-tailed deer JMed Entomol 1988 25295ndash300

Boromisa RD Grimstand PR Seroconversion rates toJamestown Canyon virus among six populations ofwhite-tailed deer (Odocoileus virginianus) in Indiana JWildl Dis 1987 2723ndash33

Brandsma AR Little SE Lockhart JM et al NovelEhrlichia organism (Rickettsiales Ehrlichieae) in white-tailed deer associated with lone star tick (Acari Ixodi-dae) parasitism J Med Entomol 1999 36190ndash194

Childs JE Paddock CD The ascendancy of Amblyommaamericanum as a vector of pathogens affecting humansin the United States Annu Rev Entomol 2003 48307ndash337

Comer JA Nicholson WL Olson JG et al Serologic test-ing for human granulocytic ehrlichiosis at a national re-ferral center J Clin Microbiol 1999 37558ndash564

Dawson JE Childs JE Biggie KL et al White-tailed deeras a potential reservoir of Ehrlichia spp J Wildl Dis1994a 30162ndash168

Dawson JE Stallknecht D Howerth EW et al Suscep-tibility of white-tailed deer (Odocoileus virginianus) toinfection with Ehrlichia chaffeensis the etiologic agent ofhuman ehrlichiosis J Clin Microbiol 1994b 322725ndash2728

Dawson JE Anderson BE Fishbein DB et al Poly-merase chain reaction evidence of Ehrlichia chaffeensisetiologic agent of human ehrlichiosis in dogs fromsoutheast Virginia Am J Vet Res 1996a 571175ndash1179

Dawson JE Warner CK Baker V et al Ehrlichia-like 16SrDNA sequence from wild white-tailed deer (Odocoileusvirginianus) J Parasitol 1996b 8252ndash58

Davidson WR Lockhart JM Stallknecht DE et al Per-sistent Ehrlichia chaffeensis infection in white-tailed deerJ Wildl Dis 2001 37538ndash546

Dumler JS Walker DH Tick-borne ehrlichiosis LancetInfect Dis 2001 021ndash28

Ewing SA Dawson JE Kocan AA et al Experimentaltransmission of Ehrlichia chaffeensis (RickettsialesEhrlichieae) among white-tailed deer by Amblyommaamericanum (Acari Ixodidae) J Med Entomol 199532368ndash374

Fletcher WO Stallknecht DE Kearney MT et al Anti-bodies to vesicular stomatitis New Jersey type virus inwhite-tailed deer on Ossabaw Island Georgia 1985ndash1989 J Wildl Dis 1991 27675ndash680

Gill JS McLean RG Neitzil DF et al Serologic analysisof white-tailed deer sera for antibodies to Borreliaburgdorferi by enzyme-linked immunosorbent assayand western blotting J Clin Microbiol 1993 31318ndash322

Ijdo JW Wu C Magnarelli LA et al Detection ofEhrlichia chaffeensis DNA in Amblyomma americanumticks in Connecticut and Rhode Island J Clin Microbiol2000 384655ndash4656

Irving RP Pinger RR Vann CN et al Distribution ofEhrlichia chaffeensis (Rickettsiales Rickettsiaeceae) inAmblyomma americanum in southern Indiana and preva-lence of E chaffeensis-reactive antibodies in white-taileddeer in Indiana and Ohio in 1998 J Med Entomol 200037595ndash600

Keirans JE Lacombe EH First records of Amblyommaamericanum Ixodes (Ixodes) dentatus and Ixodes (Cer-atixodes) uriae (Acari Ixodidae) from Maine J Parasitol1998 84629ndash631

Little SE Dawson JE Lockhart JM et al Developmentand use of specific polymerase chain reaction for thedetection of an Ehrlichia-like organism in white-taileddeer J Wildl Dis 1997 33246ndash253

Little SE Stallknecht DE Lockhart JM et al Naturalcoinfection of a white-tailed deer (Odocoileus virgini-anus) population with three Ehrlichia spp J Parasitol1998 84897ndash901

Lockhart JM Davidson WR Dawson JE et al Tempo-ral association of Amblyomma americanumwith the pres-ence of Ehrlichia chaffeensis reactive antibodies in white-tailed deer J Wildl Dis 1995 31119ndash124

Lockhart JM Davidson WR Stallknecht DE et al Site-specific geographic association between Amblyommaamericanum (Acari Ixodidae) infestations and Ehrlichiachaffeensis-reactive (Rickettsiales Ehrlichieae) antibod-ies in white-tailed deer J Med Entomol 1996 33153ndash158

Lockhart JM Davidson WR Stallknecht DE et al Iso-lation of Ehrlichia chaffeensis from wild white-tailed deer(Odocoileus virginianus) confirms their role as naturalreservoir hosts J Clin Microbiol 1997a 351681ndash1686

Lockhart JM Davidson WR Stallknecht DE et al Nat-ural history of Ehrlichia chaffeensis (RickettsialesEhrlichieae) in the Piedmont physiographic province ofGeorgia J Parasitol 1997b 83887ndash894

Magnarelli LA Anderson JF Stafford KC et al Anti-bodies to multiple tick-borne pathogens of babesiosisehrlichiosis and Lyme borreliosis in white-footed miceJ Wildl Dis 1997 33466ndash473

Mueller-Anneling L Gilchrist MJ Thorne PS Ehrlichiachaffeensis antibodies in white-tailed deer Iowa 1994and 1996 Emerg Infect Dis 2000 6397ndash400

Murphy GL Ewing SA Whitworth LC et al A molec-ular and serologic survey of Ehrlichia canis E chaffeen-sis and E ewingii in dogs and ticks from OklahomaVet Parasitol 1998 79325ndash339

Paddock CD Childs JE Ehrlichia chaffeensis a prototyp-ical emerging pathogen Clin Microbiol Rev 2003 1637ndash64

Paddock CD Folk SM Shore GM et al Infections withEhrlichia chaffeensisand Ehrlichia ewingii in persons coin-fected with human immunodeficiency virus Clin InfectDis 2001 31586ndash1594

Roland WE Everett ED Cyr TL et al Ehrlichia chaffeensisin Missouri ticks Am J Trop Med Hyg 1998 59641ndash643

Smart DL and Trainer DO Serologic evidence ofVenezuelan equine encephalitis in some wild and do-mestic populations of southern Texas J Wildl Dis 197511195ndash200

Stallknecht DE Blue JL Rollor EA et al Precipitatingantibodies to bluetongue and epizootic hemorrhagicdisease viruses in white-tailed deer in the southeasternUnited States J Wildl Dis 1991 27238ndash247

Steiert JG Gilfoy F Infection rates of Amblyomma ameri-canum and Dermacentor variabilis by Ehrlichia chaffeensis

YABSLEY ET AL206

Eighty of the 82 (976) populations infestedwith LST contained at least one seropositiveWTD while only four of 35 (114) populationsnot infested with LST contained seropositivedeer One LST-negative population (Harris CoGA) was both seronegative and tick negativein 1986 however in 2002 the deer populationwas seropositive but the LST status was notevaluated The prevalence of antibodies reac-tive to E chaffeensis was significantly higheramong LST infested populations of WTD thanpopulations not infested by LST (x2 5 899 p 0001) None of the six populations infestedonly with A maculatum contained seropositivedeer there were insufficient populations withI affinis or I scapularis alone to permit evalua-tion

Stability and spread of E chaffeensis and LST among WTD populations

Fifty-seven populations which were testedduring previous studies (Dawson et al 1994aLockhart et al 1996) were retested during thisstudy The serologic status for 50 (88) of thesepopulations remained unchanged 35 seropos-itive populations remained seropositive and 15seronegative populations remained seronega-tive Six populations which previously testedseronegative were seropositive during this

study Three of these representing Anson CoNC Haywood Co TN and Stewart Co GAhad documented invasion of LST between thefirst reported testing and this study One Texaspopulation (Travis Co) was seropositive (n 57 714) when tested in 1992 (Dawson et al1994a) but was seronegative when sampled(n 5 10) in 2002 for this study The LST statusfor this county was not known for either sam-pling period

Among the 11 populations where LST infes-tation and serologic status were monitored repeatedly over time there was complete con-cordance between LST presence and seroposi-tive deer (Table 2) All three LST positive pop-ulations were consistently seropositive eachyear tested The mean seroprevalence for thesethree populations was 83 and two of thesepopulations (Jones Co GA and Ashley CoAR) were confirmed PCR positive for E chaf-feensis None of the four populations where LSTwas consistently absent contained seropositivedeer at any sampling time Among the fourpopulations where LST status changed fromabsent to present seropositive deer were onlydetected following the appearance of LST Themean prevalence of antibodies reactive to Echaffeensis in the most recently tested year wassignificantly higher for populations infestedwith LST for multiple years (mean 5 100)

YABSLEY ET AL202

TABLE 2 SEROLOGIC AND TICK SURVEY RESULTS FROM SEVEN POPULATIONS OF WTD COLLECTED DURING

MULTIPLE TIME PERIODS AND TESTED FOR ANTIBODIES REACTIVE FOR E CHAFFEENSIS

No of Tickyears status No of deer Mean IFA prevalence

Site Years sampled (years) tested (range)

Sites positive for ticksJones Co GA 1979ndash2001 6 1 (6) 41 92 (80ndash100)Ashley Co AR 1982ndash2001 7 1 (7) 37 84 (71ndash100)Franklin Co FL 1984ndash1996 5 1 (5) 25 68 (40ndash100)

Sites negative for ticksFloyd Co GA 1973ndash2001 6 2 (6) 58 0Tyler Co WV 1985ndash2000 4 2 (4) 21 0Hardy Co WV 1977ndash1999 7 2 (7) 37 0Monroe Co FL 1992ndash2001 4 2 (4) 15 0

Sites where tick status changedAnson Co NC 1987 2001 2 2 1a 10 0 40bConcordia Pa LA 1986 1991 1999 3 2 1 1 22 0 38 60Haywood Co TN 1989 1994 1998 3 2 2 1 16 0 0 20Stewart Co GA 1986 1998 2 2 1 10 0 20

aTick status at sequential sampling periodsbPrevalence of IFA seropositive deer at sequential sampling periods

compared to populations only recently infested(mean 5 538 x2 5 155 p 0001) One re-cently infested population (Anson Co NC)was confirmed by isolation of E chaffeensisfrom a single WTD in 2001

DISCUSSION

The overarching goal of this study was toimplement and evaluate a prototype surveil-lance system using WTD as natural sentinels todetermine the geographic distribution of Echaffeensis across most of the suspected HMEendemic region of the United States To thisend the data obtained demonstrate the fol-lowing critical attributes (1) serologic findingsreflect E chaffeensis infection (2) effective sur-veillance can be achieved with small samplesizes (3) any deer 6 month old is suitable forsurveillance purposes (4) enzootic and consis-tently negative locales were identified repeat-edly across time (5) surveillance of deer hasthe ability to detect spread to new locales and(6) serologic molecular and culture diagnosticfindings among deer could be related to pres-ence of the principal tick vector A americanumThese findings compliment and expand on arecent WTD serologic surveillance effort withinIowa (Mueller-Anneling et al 2000) howeverthat study exclusively used serologic assaysand did not include either confirmatory mi-crobiological diagnostics or contemporaneoustick vector monitoring Collectively these twostudies confirm the effectiveness of this proto-typical surveillance system for monitoring thedistribution of this emerging human pathogen

Although several studies have examined theprevalence and distribution of E chaffeensis an-tibodies among many southeastern and southcentral WTD populations (Dawson et al 1994aLockhart et al 1996 Little et al 1997 Yabsleyet al 2002) the extensive additional testingfilled many geographic gaps and provide amore complete fine scale distribution of E chaf-feensis among WTD throughout the regions ofthe United States with the highest HME risk(Paddock and Childs 2003) The combined dataevaluated herein disclosed a wide distributionof seropositive WTD populations across all ofthe 18 states represented except for West Vir-

ginia however some distributional limits alsowere discerned including a western boundaryextending across Kansas Oklahoma andTexas a southern boundary across peninsularFlorida and a large cluster of many seronega-tive populations centered along the Ap-palachian Mountains (Fig 1) Other prior stud-ies of WTD have documented an apparentnorthern range limit for E chaffeensis in south-ern portions of Iowa Illinois Indiana and Ohio(Dawson et al 1994a Irving et al 2000 Mueller-Anneling et al 2000) Along the East coast sur-veys have indicated that A americanum and Echaffeensis are present as far north as Connecti-cut Maine and Rhode Island (Keirans and La-combe 1998 Ijdo et al 2000)

The accuracy of diagnostic assays utilized iscritical to any pathogen surveillance system In-fection with E chaffeensis was confirmed innearly half (46) of 122 seropositive WTD pop-ulations tested by PCR and was supplementedby isolation from others In contrast PCR evi-dence of infection was not detected in any of20 seronegative WTD populations Thus PCRand culture testing applied at the populationlevel were effective at confirming infection inmany seropositive populations and demon-strating a strong association between the sero-logic status of populations and infection withE chaffeensis

Because WTD in these regions are known to be infected with three other related rickettsiae(E ewingii A phagocytophilum and an Ana-plasma sp [WTD-agent]) in addition to E chaf-feensis (Dawson et al 1996b Little et al 1998Brandsma et al 1999 Magnarelli et al 1999Yabsley et al 2002) the potential for serologiccross-reaction is an important considerationAlthough antibodies reactive with E chaffeensisin WTD from Maryland have been confirmedby immunoblotting using the 28- to 29-kDa anti-gens of E chaffeensis (Walls et al 1998) the ex-tent of serologic cross-reactions among thesefour species is not fully understood (Dumlerand Walker 2001) However data from otherstudies suggests that infection with these otherrickettsiae may not commonly result in pro-duction of antibodies that cross-react with Echaffeensis Cross-reactions between E chaffeen-sis and E ewingii have been noted in infectedhumans and dogs however not all E ewingii

203EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

infected dogs or humans develop antibodies toE chaffeensis antigens (Murphy et al 1998 Pad-dock et al 2001) Similarly a WTD fawn ex-perimentally infected with E ewingii did not de-velop antibodies reactive with E chaffeensis(Yabsley et al 2002) Cross-reactions between Echaffeensis and A phagocytophilum have been re-ported in humans and may complicate diagno-sis (Comer et al 1999) Far fewer WTD in thesoutheastern states have antibodies reactive toA phagocytophilum (25) than to E chaffeensisantigens however although some deer reactedto both E chaffeensis and A phagocytophilum ator above the 1128 cut-off many also reacted toonly one of these antigens at titers $1128 (Lit-tle et al 1998 Walls et al 1998 VG Dugan andMJ Yabsley unpublished data) The high per-centage of populations where seropositive orseronegative status was confirmed by PCR orculture together with data from these otherstudies provide considerable evidence that theseroreactivity reported herein largely repre-sents specific seroconversion to E chaffeensis

The distribution of antibodies within WTDage and gender categories was investigated be-cause these variables are an important consid-eration for a WTD natural sentinel systemOnly a limited investigation of geometric meantiters among age categories has been reportedfor WTD (Lockhart et al 1995) and any signif-icant influence of age or gender on the occur-rence of antibodies would require samplingstratified according to these host attributes Im-portantly differences in antibody prevalencewere not noted among age or gender categoriesindicating that all WTD particularly animals $6 mo of age are suitable for use in a WTD sur-veillance system In contrast to the high stableprevalence of antibodies among age classes(Fig 2A) PCR evidence of rickettsemia de-clined with age (Fig 2B) This pattern conformsto proposed infection dynamics in naturally in-fected WTD whereby animals from approxi-mately 6 months to 15 years of age are morelikely to be rickettsemic than older adults (Pad-dock and Childs 2003 Dawson et al 1994Lockhart et al 1997a Davidson et al 2001) Inthe present study the prevalence of PCR pos-itivity dropped dramatically from over half(53) of WTD younger than 075 years to lessthan 8 of WTD older than 15 years The ma-

jority of deer tested from some seropositivepopulations were 15 years and only deer15 years from other populations were avail-able for PCR testing which may explain whynot all seropositive populations were con-firmed by PCR Because the probability of aWTD being rickettsemic declines with ageserology represents a better surveillance toolthan PCR and under natural conditions of re-exposure to ticks titers likely do not declineover time as often occurred in single exposureexperimental infections of WTD (Dawson et al1994b Ewing et al 1995 Davidson et al 2001)

Previous serologic surveys of E chaffeensis inWTD (Dawson et al 1994a Lockhart et al 1996Little et al 1997 Yabsley et al 2002) have uti-lized relatively small sample sizes per popula-tion but except for one (Mueller-Anneling etal 2000) have not addressed the adequacy ofthese sample sizes to correctly classify infectedand uninfected populations Data from the cur-rent and prior studies (Dawson et al 1994aLockhart et al 1996 Little et al 1997) haveshown that infected WTD populations in thesoutheastern and southcentral United Statestypically have high (70) seroprevalencesCalculations (Thrusfield 1995) based on a mean73 seroprevalence indicated that use of a sam-ple size of five WTD per population should re-liably detect nearly all positive populations Al-though it was beyond the scope of this studyto test enough deer to detect very low popula-tion prevalences larger sample sizes and re-peated sampling of seronegative populationsdid not reveal seropositive populations thatwould have been misclassified as seronegativebased on smaller sample sizes A chance of mis-classifying newly infected populations of WTDexists because seroprevalence may be low ini-tially however field evidence shows that anti-body prevalence can increase rapidly once apopulation becomes infected with E chaffeensis(Lockhart et al 1995) Utilizing per county sam-ple sizes equivalent to this study statewideserotesting of WTD from Iowa demonstrated aclear north-south gradient of seroprevalencethat corresponded to key epidemiologic factors(eg distributions of ticks deer and habitats)(Mueller-Anneling et al 2000)

Because the distributions of pathogens oftenare reflected by some combination of both sta-

YABSLEY ET AL204

tic and changing geographic distributions animportant attribute of an effective surveillancesystem is the ability to identify both persis-tently enzootic locales and spread to new loca-tions Testing of WTD from LST infested pop-ulations over the span of 20 years showed thatonce this biologically important vector is pres-ent and WTD populations are infected with Echaffeensis there is a high probability that thepopulations will remain infected and that WTDsentinels reflect these enzootic conditions Con-versely populations with no history of LST in-festation remained consistently seronegativewhen tested multiple times These results arecompatible with a strong site-specific associa-tion between E chaffeensis antibodies and thepresence of LST (Lockhart et al 1996) Of in-terest each of two populations (Choctaw CoAL and Bolivar Co MS) that were earlier re-ported as seropositive but LST negative (Lock-hart et al 1996) were shown in this study to beboth LST positive and seropositive for E chaf-feensis The rare failure to find LST in seropos-itive WTD populations may have occurred be-cause many collections were made during thefall hunting season when LST infestations areundergoing rapid seasonal declines (Allan2001)

Four instances of E chaffeensis spread to newlocations were detected by monitoring WTDfor multiple years (Table 2) At each of these lo-cations the change in population serologic sta-tus from negative to positive coincided withdocumented appearance of LST These findingsare very similar to those from a 12-year-longmonitoring of a WTD population that demon-strated the introduction of E chaffeensis was at-tributable to the establishment of LST (Lock-hart et al 1995) Because all of these changes toseropositive status were linked to the principalvector they provide further evidence that thissurveillance system reflects important epi-demiologic factors and that it should discernchanging local HME risks

In summary the biological reasons a sentinelWTD surveillance system functions effectivelyare that the force of transmission of E chaffeen-sis is focused on WTD and that aspects of WTDbiology allow attribution of infection to a spe-cific geographic location From an operationalperspective the system has the desirable at-

tributes of applicability on a wide geographicscale and logistic feasibility by using ldquofreerdquo(hunter-harvested) samples The present workwas done in a research context however it ispossible to implement sentinel WTD serologicsurveillance for E chaffeensis in an operationalpublic health context as clearly demonstratedpreviously in Iowa (Mueller-Anneling et al2000)

ACKNOWLEDGMENTS

We are indebted to over 100 individuals rep-resenting state wildlife agency personnelSCWDS staff and white-tailed deer researchersfor help in field collection of samples but es-pecially the following individuals for coordi-nating specific collections Chris Cook (AL)John Morgan (FL) Lloyd Fox (KS) Jon Gassettand David Yancy (KY) Jeff Beringer (MO)Douglas Roscoe (NJ) Evin Stanford (NC) AAlan Kocan (OK) Charles Ruth Jr (SC) BenLayton (TN) Matt Knox (VA) and Jim Crum(WV) Sarah Cross Page Luttrell and DarrellKavanaugh from SCWDS provided excellentfield and laboratory help Work was supportedprimarily by the National Institutes of Allergyand Infectious Diseases (5 R01 AI044235-02)Further support was provided by the FederalAid to Wildlife Restoration Act (50 Stat 917)and through sponsorship from fish and wildlifeagencies in Alabama Arkansas Florida Geor-gia Kansas Kentucky Louisiana MarylandMississippi Missouri North Carolina Okla-homa Puerto Rico South Carolina TennesseeVirginia and West Virginia

REFERENCES

Allan SA Ticks (Class Arachnida Order Acarina) InSamuel WM Pybus MJ and Kocan AA ed ParasiticDiseases of Wild Mammals Ames IA Iowa State Uni-versity Press 2001 72ndash106

Anderson BE Sumner JW Dawson JE et al Detection ofthe etiologic agent of human ehrlichiosis by polymerasechain reaction J Clin Microbiol 1992 30775ndash780

Anderson BE Sims KG Olson JG et al Amblyommaamericanum a potential vector of human ehrlichiosisAm J Trop Med Hyg 1993 49239ndash244

Bloemer SR Zimmerman RH Fairbanks K Abundanceattachment sites and density estimators of lone star

205EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

ticks (Acari Ixodidae) infesting white-tailed deer JMed Entomol 1988 25295ndash300

Boromisa RD Grimstand PR Seroconversion rates toJamestown Canyon virus among six populations ofwhite-tailed deer (Odocoileus virginianus) in Indiana JWildl Dis 1987 2723ndash33

Brandsma AR Little SE Lockhart JM et al NovelEhrlichia organism (Rickettsiales Ehrlichieae) in white-tailed deer associated with lone star tick (Acari Ixodi-dae) parasitism J Med Entomol 1999 36190ndash194

Childs JE Paddock CD The ascendancy of Amblyommaamericanum as a vector of pathogens affecting humansin the United States Annu Rev Entomol 2003 48307ndash337

Comer JA Nicholson WL Olson JG et al Serologic test-ing for human granulocytic ehrlichiosis at a national re-ferral center J Clin Microbiol 1999 37558ndash564

Dawson JE Childs JE Biggie KL et al White-tailed deeras a potential reservoir of Ehrlichia spp J Wildl Dis1994a 30162ndash168

Dawson JE Stallknecht D Howerth EW et al Suscep-tibility of white-tailed deer (Odocoileus virginianus) toinfection with Ehrlichia chaffeensis the etiologic agent ofhuman ehrlichiosis J Clin Microbiol 1994b 322725ndash2728

Dawson JE Anderson BE Fishbein DB et al Poly-merase chain reaction evidence of Ehrlichia chaffeensisetiologic agent of human ehrlichiosis in dogs fromsoutheast Virginia Am J Vet Res 1996a 571175ndash1179

Dawson JE Warner CK Baker V et al Ehrlichia-like 16SrDNA sequence from wild white-tailed deer (Odocoileusvirginianus) J Parasitol 1996b 8252ndash58

Davidson WR Lockhart JM Stallknecht DE et al Per-sistent Ehrlichia chaffeensis infection in white-tailed deerJ Wildl Dis 2001 37538ndash546

Dumler JS Walker DH Tick-borne ehrlichiosis LancetInfect Dis 2001 021ndash28

Ewing SA Dawson JE Kocan AA et al Experimentaltransmission of Ehrlichia chaffeensis (RickettsialesEhrlichieae) among white-tailed deer by Amblyommaamericanum (Acari Ixodidae) J Med Entomol 199532368ndash374

Fletcher WO Stallknecht DE Kearney MT et al Anti-bodies to vesicular stomatitis New Jersey type virus inwhite-tailed deer on Ossabaw Island Georgia 1985ndash1989 J Wildl Dis 1991 27675ndash680

Gill JS McLean RG Neitzil DF et al Serologic analysisof white-tailed deer sera for antibodies to Borreliaburgdorferi by enzyme-linked immunosorbent assayand western blotting J Clin Microbiol 1993 31318ndash322

Ijdo JW Wu C Magnarelli LA et al Detection ofEhrlichia chaffeensis DNA in Amblyomma americanumticks in Connecticut and Rhode Island J Clin Microbiol2000 384655ndash4656

Irving RP Pinger RR Vann CN et al Distribution ofEhrlichia chaffeensis (Rickettsiales Rickettsiaeceae) inAmblyomma americanum in southern Indiana and preva-lence of E chaffeensis-reactive antibodies in white-taileddeer in Indiana and Ohio in 1998 J Med Entomol 200037595ndash600

Keirans JE Lacombe EH First records of Amblyommaamericanum Ixodes (Ixodes) dentatus and Ixodes (Cer-atixodes) uriae (Acari Ixodidae) from Maine J Parasitol1998 84629ndash631

Little SE Dawson JE Lockhart JM et al Developmentand use of specific polymerase chain reaction for thedetection of an Ehrlichia-like organism in white-taileddeer J Wildl Dis 1997 33246ndash253

Little SE Stallknecht DE Lockhart JM et al Naturalcoinfection of a white-tailed deer (Odocoileus virgini-anus) population with three Ehrlichia spp J Parasitol1998 84897ndash901

Lockhart JM Davidson WR Dawson JE et al Tempo-ral association of Amblyomma americanumwith the pres-ence of Ehrlichia chaffeensis reactive antibodies in white-tailed deer J Wildl Dis 1995 31119ndash124

Lockhart JM Davidson WR Stallknecht DE et al Site-specific geographic association between Amblyommaamericanum (Acari Ixodidae) infestations and Ehrlichiachaffeensis-reactive (Rickettsiales Ehrlichieae) antibod-ies in white-tailed deer J Med Entomol 1996 33153ndash158

Lockhart JM Davidson WR Stallknecht DE et al Iso-lation of Ehrlichia chaffeensis from wild white-tailed deer(Odocoileus virginianus) confirms their role as naturalreservoir hosts J Clin Microbiol 1997a 351681ndash1686

Lockhart JM Davidson WR Stallknecht DE et al Nat-ural history of Ehrlichia chaffeensis (RickettsialesEhrlichieae) in the Piedmont physiographic province ofGeorgia J Parasitol 1997b 83887ndash894

Magnarelli LA Anderson JF Stafford KC et al Anti-bodies to multiple tick-borne pathogens of babesiosisehrlichiosis and Lyme borreliosis in white-footed miceJ Wildl Dis 1997 33466ndash473

Mueller-Anneling L Gilchrist MJ Thorne PS Ehrlichiachaffeensis antibodies in white-tailed deer Iowa 1994and 1996 Emerg Infect Dis 2000 6397ndash400

Murphy GL Ewing SA Whitworth LC et al A molec-ular and serologic survey of Ehrlichia canis E chaffeen-sis and E ewingii in dogs and ticks from OklahomaVet Parasitol 1998 79325ndash339

Paddock CD Childs JE Ehrlichia chaffeensis a prototyp-ical emerging pathogen Clin Microbiol Rev 2003 1637ndash64

Paddock CD Folk SM Shore GM et al Infections withEhrlichia chaffeensisand Ehrlichia ewingii in persons coin-fected with human immunodeficiency virus Clin InfectDis 2001 31586ndash1594

Roland WE Everett ED Cyr TL et al Ehrlichia chaffeensisin Missouri ticks Am J Trop Med Hyg 1998 59641ndash643

Smart DL and Trainer DO Serologic evidence ofVenezuelan equine encephalitis in some wild and do-mestic populations of southern Texas J Wildl Dis 197511195ndash200

Stallknecht DE Blue JL Rollor EA et al Precipitatingantibodies to bluetongue and epizootic hemorrhagicdisease viruses in white-tailed deer in the southeasternUnited States J Wildl Dis 1991 27238ndash247

Steiert JG Gilfoy F Infection rates of Amblyomma ameri-canum and Dermacentor variabilis by Ehrlichia chaffeensis

YABSLEY ET AL206

compared to populations only recently infested(mean 5 538 x2 5 155 p 0001) One re-cently infested population (Anson Co NC)was confirmed by isolation of E chaffeensisfrom a single WTD in 2001

DISCUSSION

The overarching goal of this study was toimplement and evaluate a prototype surveil-lance system using WTD as natural sentinels todetermine the geographic distribution of Echaffeensis across most of the suspected HMEendemic region of the United States To thisend the data obtained demonstrate the fol-lowing critical attributes (1) serologic findingsreflect E chaffeensis infection (2) effective sur-veillance can be achieved with small samplesizes (3) any deer 6 month old is suitable forsurveillance purposes (4) enzootic and consis-tently negative locales were identified repeat-edly across time (5) surveillance of deer hasthe ability to detect spread to new locales and(6) serologic molecular and culture diagnosticfindings among deer could be related to pres-ence of the principal tick vector A americanumThese findings compliment and expand on arecent WTD serologic surveillance effort withinIowa (Mueller-Anneling et al 2000) howeverthat study exclusively used serologic assaysand did not include either confirmatory mi-crobiological diagnostics or contemporaneoustick vector monitoring Collectively these twostudies confirm the effectiveness of this proto-typical surveillance system for monitoring thedistribution of this emerging human pathogen

Although several studies have examined theprevalence and distribution of E chaffeensis an-tibodies among many southeastern and southcentral WTD populations (Dawson et al 1994aLockhart et al 1996 Little et al 1997 Yabsleyet al 2002) the extensive additional testingfilled many geographic gaps and provide amore complete fine scale distribution of E chaf-feensis among WTD throughout the regions ofthe United States with the highest HME risk(Paddock and Childs 2003) The combined dataevaluated herein disclosed a wide distributionof seropositive WTD populations across all ofthe 18 states represented except for West Vir-

ginia however some distributional limits alsowere discerned including a western boundaryextending across Kansas Oklahoma andTexas a southern boundary across peninsularFlorida and a large cluster of many seronega-tive populations centered along the Ap-palachian Mountains (Fig 1) Other prior stud-ies of WTD have documented an apparentnorthern range limit for E chaffeensis in south-ern portions of Iowa Illinois Indiana and Ohio(Dawson et al 1994a Irving et al 2000 Mueller-Anneling et al 2000) Along the East coast sur-veys have indicated that A americanum and Echaffeensis are present as far north as Connecti-cut Maine and Rhode Island (Keirans and La-combe 1998 Ijdo et al 2000)

The accuracy of diagnostic assays utilized iscritical to any pathogen surveillance system In-fection with E chaffeensis was confirmed innearly half (46) of 122 seropositive WTD pop-ulations tested by PCR and was supplementedby isolation from others In contrast PCR evi-dence of infection was not detected in any of20 seronegative WTD populations Thus PCRand culture testing applied at the populationlevel were effective at confirming infection inmany seropositive populations and demon-strating a strong association between the sero-logic status of populations and infection withE chaffeensis

Because WTD in these regions are known to be infected with three other related rickettsiae(E ewingii A phagocytophilum and an Ana-plasma sp [WTD-agent]) in addition to E chaf-feensis (Dawson et al 1996b Little et al 1998Brandsma et al 1999 Magnarelli et al 1999Yabsley et al 2002) the potential for serologiccross-reaction is an important considerationAlthough antibodies reactive with E chaffeensisin WTD from Maryland have been confirmedby immunoblotting using the 28- to 29-kDa anti-gens of E chaffeensis (Walls et al 1998) the ex-tent of serologic cross-reactions among thesefour species is not fully understood (Dumlerand Walker 2001) However data from otherstudies suggests that infection with these otherrickettsiae may not commonly result in pro-duction of antibodies that cross-react with Echaffeensis Cross-reactions between E chaffeen-sis and E ewingii have been noted in infectedhumans and dogs however not all E ewingii

203EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

infected dogs or humans develop antibodies toE chaffeensis antigens (Murphy et al 1998 Pad-dock et al 2001) Similarly a WTD fawn ex-perimentally infected with E ewingii did not de-velop antibodies reactive with E chaffeensis(Yabsley et al 2002) Cross-reactions between Echaffeensis and A phagocytophilum have been re-ported in humans and may complicate diagno-sis (Comer et al 1999) Far fewer WTD in thesoutheastern states have antibodies reactive toA phagocytophilum (25) than to E chaffeensisantigens however although some deer reactedto both E chaffeensis and A phagocytophilum ator above the 1128 cut-off many also reacted toonly one of these antigens at titers $1128 (Lit-tle et al 1998 Walls et al 1998 VG Dugan andMJ Yabsley unpublished data) The high per-centage of populations where seropositive orseronegative status was confirmed by PCR orculture together with data from these otherstudies provide considerable evidence that theseroreactivity reported herein largely repre-sents specific seroconversion to E chaffeensis

The distribution of antibodies within WTDage and gender categories was investigated be-cause these variables are an important consid-eration for a WTD natural sentinel systemOnly a limited investigation of geometric meantiters among age categories has been reportedfor WTD (Lockhart et al 1995) and any signif-icant influence of age or gender on the occur-rence of antibodies would require samplingstratified according to these host attributes Im-portantly differences in antibody prevalencewere not noted among age or gender categoriesindicating that all WTD particularly animals $6 mo of age are suitable for use in a WTD sur-veillance system In contrast to the high stableprevalence of antibodies among age classes(Fig 2A) PCR evidence of rickettsemia de-clined with age (Fig 2B) This pattern conformsto proposed infection dynamics in naturally in-fected WTD whereby animals from approxi-mately 6 months to 15 years of age are morelikely to be rickettsemic than older adults (Pad-dock and Childs 2003 Dawson et al 1994Lockhart et al 1997a Davidson et al 2001) Inthe present study the prevalence of PCR pos-itivity dropped dramatically from over half(53) of WTD younger than 075 years to lessthan 8 of WTD older than 15 years The ma-

jority of deer tested from some seropositivepopulations were 15 years and only deer15 years from other populations were avail-able for PCR testing which may explain whynot all seropositive populations were con-firmed by PCR Because the probability of aWTD being rickettsemic declines with ageserology represents a better surveillance toolthan PCR and under natural conditions of re-exposure to ticks titers likely do not declineover time as often occurred in single exposureexperimental infections of WTD (Dawson et al1994b Ewing et al 1995 Davidson et al 2001)

Previous serologic surveys of E chaffeensis inWTD (Dawson et al 1994a Lockhart et al 1996Little et al 1997 Yabsley et al 2002) have uti-lized relatively small sample sizes per popula-tion but except for one (Mueller-Anneling etal 2000) have not addressed the adequacy ofthese sample sizes to correctly classify infectedand uninfected populations Data from the cur-rent and prior studies (Dawson et al 1994aLockhart et al 1996 Little et al 1997) haveshown that infected WTD populations in thesoutheastern and southcentral United Statestypically have high (70) seroprevalencesCalculations (Thrusfield 1995) based on a mean73 seroprevalence indicated that use of a sam-ple size of five WTD per population should re-liably detect nearly all positive populations Al-though it was beyond the scope of this studyto test enough deer to detect very low popula-tion prevalences larger sample sizes and re-peated sampling of seronegative populationsdid not reveal seropositive populations thatwould have been misclassified as seronegativebased on smaller sample sizes A chance of mis-classifying newly infected populations of WTDexists because seroprevalence may be low ini-tially however field evidence shows that anti-body prevalence can increase rapidly once apopulation becomes infected with E chaffeensis(Lockhart et al 1995) Utilizing per county sam-ple sizes equivalent to this study statewideserotesting of WTD from Iowa demonstrated aclear north-south gradient of seroprevalencethat corresponded to key epidemiologic factors(eg distributions of ticks deer and habitats)(Mueller-Anneling et al 2000)

Because the distributions of pathogens oftenare reflected by some combination of both sta-

YABSLEY ET AL204

tic and changing geographic distributions animportant attribute of an effective surveillancesystem is the ability to identify both persis-tently enzootic locales and spread to new loca-tions Testing of WTD from LST infested pop-ulations over the span of 20 years showed thatonce this biologically important vector is pres-ent and WTD populations are infected with Echaffeensis there is a high probability that thepopulations will remain infected and that WTDsentinels reflect these enzootic conditions Con-versely populations with no history of LST in-festation remained consistently seronegativewhen tested multiple times These results arecompatible with a strong site-specific associa-tion between E chaffeensis antibodies and thepresence of LST (Lockhart et al 1996) Of in-terest each of two populations (Choctaw CoAL and Bolivar Co MS) that were earlier re-ported as seropositive but LST negative (Lock-hart et al 1996) were shown in this study to beboth LST positive and seropositive for E chaf-feensis The rare failure to find LST in seropos-itive WTD populations may have occurred be-cause many collections were made during thefall hunting season when LST infestations areundergoing rapid seasonal declines (Allan2001)

Four instances of E chaffeensis spread to newlocations were detected by monitoring WTDfor multiple years (Table 2) At each of these lo-cations the change in population serologic sta-tus from negative to positive coincided withdocumented appearance of LST These findingsare very similar to those from a 12-year-longmonitoring of a WTD population that demon-strated the introduction of E chaffeensis was at-tributable to the establishment of LST (Lock-hart et al 1995) Because all of these changes toseropositive status were linked to the principalvector they provide further evidence that thissurveillance system reflects important epi-demiologic factors and that it should discernchanging local HME risks

In summary the biological reasons a sentinelWTD surveillance system functions effectivelyare that the force of transmission of E chaffeen-sis is focused on WTD and that aspects of WTDbiology allow attribution of infection to a spe-cific geographic location From an operationalperspective the system has the desirable at-

tributes of applicability on a wide geographicscale and logistic feasibility by using ldquofreerdquo(hunter-harvested) samples The present workwas done in a research context however it ispossible to implement sentinel WTD serologicsurveillance for E chaffeensis in an operationalpublic health context as clearly demonstratedpreviously in Iowa (Mueller-Anneling et al2000)

ACKNOWLEDGMENTS

We are indebted to over 100 individuals rep-resenting state wildlife agency personnelSCWDS staff and white-tailed deer researchersfor help in field collection of samples but es-pecially the following individuals for coordi-nating specific collections Chris Cook (AL)John Morgan (FL) Lloyd Fox (KS) Jon Gassettand David Yancy (KY) Jeff Beringer (MO)Douglas Roscoe (NJ) Evin Stanford (NC) AAlan Kocan (OK) Charles Ruth Jr (SC) BenLayton (TN) Matt Knox (VA) and Jim Crum(WV) Sarah Cross Page Luttrell and DarrellKavanaugh from SCWDS provided excellentfield and laboratory help Work was supportedprimarily by the National Institutes of Allergyand Infectious Diseases (5 R01 AI044235-02)Further support was provided by the FederalAid to Wildlife Restoration Act (50 Stat 917)and through sponsorship from fish and wildlifeagencies in Alabama Arkansas Florida Geor-gia Kansas Kentucky Louisiana MarylandMississippi Missouri North Carolina Okla-homa Puerto Rico South Carolina TennesseeVirginia and West Virginia

REFERENCES

Allan SA Ticks (Class Arachnida Order Acarina) InSamuel WM Pybus MJ and Kocan AA ed ParasiticDiseases of Wild Mammals Ames IA Iowa State Uni-versity Press 2001 72ndash106

Anderson BE Sumner JW Dawson JE et al Detection ofthe etiologic agent of human ehrlichiosis by polymerasechain reaction J Clin Microbiol 1992 30775ndash780

Anderson BE Sims KG Olson JG et al Amblyommaamericanum a potential vector of human ehrlichiosisAm J Trop Med Hyg 1993 49239ndash244

Bloemer SR Zimmerman RH Fairbanks K Abundanceattachment sites and density estimators of lone star

205EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

ticks (Acari Ixodidae) infesting white-tailed deer JMed Entomol 1988 25295ndash300

Boromisa RD Grimstand PR Seroconversion rates toJamestown Canyon virus among six populations ofwhite-tailed deer (Odocoileus virginianus) in Indiana JWildl Dis 1987 2723ndash33

Brandsma AR Little SE Lockhart JM et al NovelEhrlichia organism (Rickettsiales Ehrlichieae) in white-tailed deer associated with lone star tick (Acari Ixodi-dae) parasitism J Med Entomol 1999 36190ndash194

Childs JE Paddock CD The ascendancy of Amblyommaamericanum as a vector of pathogens affecting humansin the United States Annu Rev Entomol 2003 48307ndash337

Comer JA Nicholson WL Olson JG et al Serologic test-ing for human granulocytic ehrlichiosis at a national re-ferral center J Clin Microbiol 1999 37558ndash564

Dawson JE Childs JE Biggie KL et al White-tailed deeras a potential reservoir of Ehrlichia spp J Wildl Dis1994a 30162ndash168

Dawson JE Stallknecht D Howerth EW et al Suscep-tibility of white-tailed deer (Odocoileus virginianus) toinfection with Ehrlichia chaffeensis the etiologic agent ofhuman ehrlichiosis J Clin Microbiol 1994b 322725ndash2728

Dawson JE Anderson BE Fishbein DB et al Poly-merase chain reaction evidence of Ehrlichia chaffeensisetiologic agent of human ehrlichiosis in dogs fromsoutheast Virginia Am J Vet Res 1996a 571175ndash1179

Dawson JE Warner CK Baker V et al Ehrlichia-like 16SrDNA sequence from wild white-tailed deer (Odocoileusvirginianus) J Parasitol 1996b 8252ndash58

Davidson WR Lockhart JM Stallknecht DE et al Per-sistent Ehrlichia chaffeensis infection in white-tailed deerJ Wildl Dis 2001 37538ndash546

Dumler JS Walker DH Tick-borne ehrlichiosis LancetInfect Dis 2001 021ndash28

Ewing SA Dawson JE Kocan AA et al Experimentaltransmission of Ehrlichia chaffeensis (RickettsialesEhrlichieae) among white-tailed deer by Amblyommaamericanum (Acari Ixodidae) J Med Entomol 199532368ndash374

Fletcher WO Stallknecht DE Kearney MT et al Anti-bodies to vesicular stomatitis New Jersey type virus inwhite-tailed deer on Ossabaw Island Georgia 1985ndash1989 J Wildl Dis 1991 27675ndash680

Gill JS McLean RG Neitzil DF et al Serologic analysisof white-tailed deer sera for antibodies to Borreliaburgdorferi by enzyme-linked immunosorbent assayand western blotting J Clin Microbiol 1993 31318ndash322

Ijdo JW Wu C Magnarelli LA et al Detection ofEhrlichia chaffeensis DNA in Amblyomma americanumticks in Connecticut and Rhode Island J Clin Microbiol2000 384655ndash4656

Irving RP Pinger RR Vann CN et al Distribution ofEhrlichia chaffeensis (Rickettsiales Rickettsiaeceae) inAmblyomma americanum in southern Indiana and preva-lence of E chaffeensis-reactive antibodies in white-taileddeer in Indiana and Ohio in 1998 J Med Entomol 200037595ndash600

Keirans JE Lacombe EH First records of Amblyommaamericanum Ixodes (Ixodes) dentatus and Ixodes (Cer-atixodes) uriae (Acari Ixodidae) from Maine J Parasitol1998 84629ndash631

Little SE Dawson JE Lockhart JM et al Developmentand use of specific polymerase chain reaction for thedetection of an Ehrlichia-like organism in white-taileddeer J Wildl Dis 1997 33246ndash253

Little SE Stallknecht DE Lockhart JM et al Naturalcoinfection of a white-tailed deer (Odocoileus virgini-anus) population with three Ehrlichia spp J Parasitol1998 84897ndash901

Lockhart JM Davidson WR Dawson JE et al Tempo-ral association of Amblyomma americanumwith the pres-ence of Ehrlichia chaffeensis reactive antibodies in white-tailed deer J Wildl Dis 1995 31119ndash124

Lockhart JM Davidson WR Stallknecht DE et al Site-specific geographic association between Amblyommaamericanum (Acari Ixodidae) infestations and Ehrlichiachaffeensis-reactive (Rickettsiales Ehrlichieae) antibod-ies in white-tailed deer J Med Entomol 1996 33153ndash158

Lockhart JM Davidson WR Stallknecht DE et al Iso-lation of Ehrlichia chaffeensis from wild white-tailed deer(Odocoileus virginianus) confirms their role as naturalreservoir hosts J Clin Microbiol 1997a 351681ndash1686

Lockhart JM Davidson WR Stallknecht DE et al Nat-ural history of Ehrlichia chaffeensis (RickettsialesEhrlichieae) in the Piedmont physiographic province ofGeorgia J Parasitol 1997b 83887ndash894

Magnarelli LA Anderson JF Stafford KC et al Anti-bodies to multiple tick-borne pathogens of babesiosisehrlichiosis and Lyme borreliosis in white-footed miceJ Wildl Dis 1997 33466ndash473

Mueller-Anneling L Gilchrist MJ Thorne PS Ehrlichiachaffeensis antibodies in white-tailed deer Iowa 1994and 1996 Emerg Infect Dis 2000 6397ndash400

Murphy GL Ewing SA Whitworth LC et al A molec-ular and serologic survey of Ehrlichia canis E chaffeen-sis and E ewingii in dogs and ticks from OklahomaVet Parasitol 1998 79325ndash339

Paddock CD Childs JE Ehrlichia chaffeensis a prototyp-ical emerging pathogen Clin Microbiol Rev 2003 1637ndash64

Paddock CD Folk SM Shore GM et al Infections withEhrlichia chaffeensisand Ehrlichia ewingii in persons coin-fected with human immunodeficiency virus Clin InfectDis 2001 31586ndash1594

Roland WE Everett ED Cyr TL et al Ehrlichia chaffeensisin Missouri ticks Am J Trop Med Hyg 1998 59641ndash643

Smart DL and Trainer DO Serologic evidence ofVenezuelan equine encephalitis in some wild and do-mestic populations of southern Texas J Wildl Dis 197511195ndash200

Stallknecht DE Blue JL Rollor EA et al Precipitatingantibodies to bluetongue and epizootic hemorrhagicdisease viruses in white-tailed deer in the southeasternUnited States J Wildl Dis 1991 27238ndash247

Steiert JG Gilfoy F Infection rates of Amblyomma ameri-canum and Dermacentor variabilis by Ehrlichia chaffeensis

YABSLEY ET AL206

infected dogs or humans develop antibodies toE chaffeensis antigens (Murphy et al 1998 Pad-dock et al 2001) Similarly a WTD fawn ex-perimentally infected with E ewingii did not de-velop antibodies reactive with E chaffeensis(Yabsley et al 2002) Cross-reactions between Echaffeensis and A phagocytophilum have been re-ported in humans and may complicate diagno-sis (Comer et al 1999) Far fewer WTD in thesoutheastern states have antibodies reactive toA phagocytophilum (25) than to E chaffeensisantigens however although some deer reactedto both E chaffeensis and A phagocytophilum ator above the 1128 cut-off many also reacted toonly one of these antigens at titers $1128 (Lit-tle et al 1998 Walls et al 1998 VG Dugan andMJ Yabsley unpublished data) The high per-centage of populations where seropositive orseronegative status was confirmed by PCR orculture together with data from these otherstudies provide considerable evidence that theseroreactivity reported herein largely repre-sents specific seroconversion to E chaffeensis

The distribution of antibodies within WTDage and gender categories was investigated be-cause these variables are an important consid-eration for a WTD natural sentinel systemOnly a limited investigation of geometric meantiters among age categories has been reportedfor WTD (Lockhart et al 1995) and any signif-icant influence of age or gender on the occur-rence of antibodies would require samplingstratified according to these host attributes Im-portantly differences in antibody prevalencewere not noted among age or gender categoriesindicating that all WTD particularly animals $6 mo of age are suitable for use in a WTD sur-veillance system In contrast to the high stableprevalence of antibodies among age classes(Fig 2A) PCR evidence of rickettsemia de-clined with age (Fig 2B) This pattern conformsto proposed infection dynamics in naturally in-fected WTD whereby animals from approxi-mately 6 months to 15 years of age are morelikely to be rickettsemic than older adults (Pad-dock and Childs 2003 Dawson et al 1994Lockhart et al 1997a Davidson et al 2001) Inthe present study the prevalence of PCR pos-itivity dropped dramatically from over half(53) of WTD younger than 075 years to lessthan 8 of WTD older than 15 years The ma-

jority of deer tested from some seropositivepopulations were 15 years and only deer15 years from other populations were avail-able for PCR testing which may explain whynot all seropositive populations were con-firmed by PCR Because the probability of aWTD being rickettsemic declines with ageserology represents a better surveillance toolthan PCR and under natural conditions of re-exposure to ticks titers likely do not declineover time as often occurred in single exposureexperimental infections of WTD (Dawson et al1994b Ewing et al 1995 Davidson et al 2001)

Previous serologic surveys of E chaffeensis inWTD (Dawson et al 1994a Lockhart et al 1996Little et al 1997 Yabsley et al 2002) have uti-lized relatively small sample sizes per popula-tion but except for one (Mueller-Anneling etal 2000) have not addressed the adequacy ofthese sample sizes to correctly classify infectedand uninfected populations Data from the cur-rent and prior studies (Dawson et al 1994aLockhart et al 1996 Little et al 1997) haveshown that infected WTD populations in thesoutheastern and southcentral United Statestypically have high (70) seroprevalencesCalculations (Thrusfield 1995) based on a mean73 seroprevalence indicated that use of a sam-ple size of five WTD per population should re-liably detect nearly all positive populations Al-though it was beyond the scope of this studyto test enough deer to detect very low popula-tion prevalences larger sample sizes and re-peated sampling of seronegative populationsdid not reveal seropositive populations thatwould have been misclassified as seronegativebased on smaller sample sizes A chance of mis-classifying newly infected populations of WTDexists because seroprevalence may be low ini-tially however field evidence shows that anti-body prevalence can increase rapidly once apopulation becomes infected with E chaffeensis(Lockhart et al 1995) Utilizing per county sam-ple sizes equivalent to this study statewideserotesting of WTD from Iowa demonstrated aclear north-south gradient of seroprevalencethat corresponded to key epidemiologic factors(eg distributions of ticks deer and habitats)(Mueller-Anneling et al 2000)

Because the distributions of pathogens oftenare reflected by some combination of both sta-

YABSLEY ET AL204

tic and changing geographic distributions animportant attribute of an effective surveillancesystem is the ability to identify both persis-tently enzootic locales and spread to new loca-tions Testing of WTD from LST infested pop-ulations over the span of 20 years showed thatonce this biologically important vector is pres-ent and WTD populations are infected with Echaffeensis there is a high probability that thepopulations will remain infected and that WTDsentinels reflect these enzootic conditions Con-versely populations with no history of LST in-festation remained consistently seronegativewhen tested multiple times These results arecompatible with a strong site-specific associa-tion between E chaffeensis antibodies and thepresence of LST (Lockhart et al 1996) Of in-terest each of two populations (Choctaw CoAL and Bolivar Co MS) that were earlier re-ported as seropositive but LST negative (Lock-hart et al 1996) were shown in this study to beboth LST positive and seropositive for E chaf-feensis The rare failure to find LST in seropos-itive WTD populations may have occurred be-cause many collections were made during thefall hunting season when LST infestations areundergoing rapid seasonal declines (Allan2001)

Four instances of E chaffeensis spread to newlocations were detected by monitoring WTDfor multiple years (Table 2) At each of these lo-cations the change in population serologic sta-tus from negative to positive coincided withdocumented appearance of LST These findingsare very similar to those from a 12-year-longmonitoring of a WTD population that demon-strated the introduction of E chaffeensis was at-tributable to the establishment of LST (Lock-hart et al 1995) Because all of these changes toseropositive status were linked to the principalvector they provide further evidence that thissurveillance system reflects important epi-demiologic factors and that it should discernchanging local HME risks

In summary the biological reasons a sentinelWTD surveillance system functions effectivelyare that the force of transmission of E chaffeen-sis is focused on WTD and that aspects of WTDbiology allow attribution of infection to a spe-cific geographic location From an operationalperspective the system has the desirable at-

tributes of applicability on a wide geographicscale and logistic feasibility by using ldquofreerdquo(hunter-harvested) samples The present workwas done in a research context however it ispossible to implement sentinel WTD serologicsurveillance for E chaffeensis in an operationalpublic health context as clearly demonstratedpreviously in Iowa (Mueller-Anneling et al2000)

ACKNOWLEDGMENTS

We are indebted to over 100 individuals rep-resenting state wildlife agency personnelSCWDS staff and white-tailed deer researchersfor help in field collection of samples but es-pecially the following individuals for coordi-nating specific collections Chris Cook (AL)John Morgan (FL) Lloyd Fox (KS) Jon Gassettand David Yancy (KY) Jeff Beringer (MO)Douglas Roscoe (NJ) Evin Stanford (NC) AAlan Kocan (OK) Charles Ruth Jr (SC) BenLayton (TN) Matt Knox (VA) and Jim Crum(WV) Sarah Cross Page Luttrell and DarrellKavanaugh from SCWDS provided excellentfield and laboratory help Work was supportedprimarily by the National Institutes of Allergyand Infectious Diseases (5 R01 AI044235-02)Further support was provided by the FederalAid to Wildlife Restoration Act (50 Stat 917)and through sponsorship from fish and wildlifeagencies in Alabama Arkansas Florida Geor-gia Kansas Kentucky Louisiana MarylandMississippi Missouri North Carolina Okla-homa Puerto Rico South Carolina TennesseeVirginia and West Virginia

REFERENCES

Allan SA Ticks (Class Arachnida Order Acarina) InSamuel WM Pybus MJ and Kocan AA ed ParasiticDiseases of Wild Mammals Ames IA Iowa State Uni-versity Press 2001 72ndash106

Anderson BE Sumner JW Dawson JE et al Detection ofthe etiologic agent of human ehrlichiosis by polymerasechain reaction J Clin Microbiol 1992 30775ndash780

Anderson BE Sims KG Olson JG et al Amblyommaamericanum a potential vector of human ehrlichiosisAm J Trop Med Hyg 1993 49239ndash244

Bloemer SR Zimmerman RH Fairbanks K Abundanceattachment sites and density estimators of lone star

205EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

ticks (Acari Ixodidae) infesting white-tailed deer JMed Entomol 1988 25295ndash300

Boromisa RD Grimstand PR Seroconversion rates toJamestown Canyon virus among six populations ofwhite-tailed deer (Odocoileus virginianus) in Indiana JWildl Dis 1987 2723ndash33

Brandsma AR Little SE Lockhart JM et al NovelEhrlichia organism (Rickettsiales Ehrlichieae) in white-tailed deer associated with lone star tick (Acari Ixodi-dae) parasitism J Med Entomol 1999 36190ndash194

Childs JE Paddock CD The ascendancy of Amblyommaamericanum as a vector of pathogens affecting humansin the United States Annu Rev Entomol 2003 48307ndash337

Comer JA Nicholson WL Olson JG et al Serologic test-ing for human granulocytic ehrlichiosis at a national re-ferral center J Clin Microbiol 1999 37558ndash564

Dawson JE Childs JE Biggie KL et al White-tailed deeras a potential reservoir of Ehrlichia spp J Wildl Dis1994a 30162ndash168

Dawson JE Stallknecht D Howerth EW et al Suscep-tibility of white-tailed deer (Odocoileus virginianus) toinfection with Ehrlichia chaffeensis the etiologic agent ofhuman ehrlichiosis J Clin Microbiol 1994b 322725ndash2728

Dawson JE Anderson BE Fishbein DB et al Poly-merase chain reaction evidence of Ehrlichia chaffeensisetiologic agent of human ehrlichiosis in dogs fromsoutheast Virginia Am J Vet Res 1996a 571175ndash1179

Dawson JE Warner CK Baker V et al Ehrlichia-like 16SrDNA sequence from wild white-tailed deer (Odocoileusvirginianus) J Parasitol 1996b 8252ndash58

Davidson WR Lockhart JM Stallknecht DE et al Per-sistent Ehrlichia chaffeensis infection in white-tailed deerJ Wildl Dis 2001 37538ndash546

Dumler JS Walker DH Tick-borne ehrlichiosis LancetInfect Dis 2001 021ndash28

Ewing SA Dawson JE Kocan AA et al Experimentaltransmission of Ehrlichia chaffeensis (RickettsialesEhrlichieae) among white-tailed deer by Amblyommaamericanum (Acari Ixodidae) J Med Entomol 199532368ndash374

Fletcher WO Stallknecht DE Kearney MT et al Anti-bodies to vesicular stomatitis New Jersey type virus inwhite-tailed deer on Ossabaw Island Georgia 1985ndash1989 J Wildl Dis 1991 27675ndash680

Gill JS McLean RG Neitzil DF et al Serologic analysisof white-tailed deer sera for antibodies to Borreliaburgdorferi by enzyme-linked immunosorbent assayand western blotting J Clin Microbiol 1993 31318ndash322

Ijdo JW Wu C Magnarelli LA et al Detection ofEhrlichia chaffeensis DNA in Amblyomma americanumticks in Connecticut and Rhode Island J Clin Microbiol2000 384655ndash4656

Irving RP Pinger RR Vann CN et al Distribution ofEhrlichia chaffeensis (Rickettsiales Rickettsiaeceae) inAmblyomma americanum in southern Indiana and preva-lence of E chaffeensis-reactive antibodies in white-taileddeer in Indiana and Ohio in 1998 J Med Entomol 200037595ndash600

Keirans JE Lacombe EH First records of Amblyommaamericanum Ixodes (Ixodes) dentatus and Ixodes (Cer-atixodes) uriae (Acari Ixodidae) from Maine J Parasitol1998 84629ndash631

Little SE Dawson JE Lockhart JM et al Developmentand use of specific polymerase chain reaction for thedetection of an Ehrlichia-like organism in white-taileddeer J Wildl Dis 1997 33246ndash253

Little SE Stallknecht DE Lockhart JM et al Naturalcoinfection of a white-tailed deer (Odocoileus virgini-anus) population with three Ehrlichia spp J Parasitol1998 84897ndash901

Lockhart JM Davidson WR Dawson JE et al Tempo-ral association of Amblyomma americanumwith the pres-ence of Ehrlichia chaffeensis reactive antibodies in white-tailed deer J Wildl Dis 1995 31119ndash124

Lockhart JM Davidson WR Stallknecht DE et al Site-specific geographic association between Amblyommaamericanum (Acari Ixodidae) infestations and Ehrlichiachaffeensis-reactive (Rickettsiales Ehrlichieae) antibod-ies in white-tailed deer J Med Entomol 1996 33153ndash158

Lockhart JM Davidson WR Stallknecht DE et al Iso-lation of Ehrlichia chaffeensis from wild white-tailed deer(Odocoileus virginianus) confirms their role as naturalreservoir hosts J Clin Microbiol 1997a 351681ndash1686

Lockhart JM Davidson WR Stallknecht DE et al Nat-ural history of Ehrlichia chaffeensis (RickettsialesEhrlichieae) in the Piedmont physiographic province ofGeorgia J Parasitol 1997b 83887ndash894

Magnarelli LA Anderson JF Stafford KC et al Anti-bodies to multiple tick-borne pathogens of babesiosisehrlichiosis and Lyme borreliosis in white-footed miceJ Wildl Dis 1997 33466ndash473

Mueller-Anneling L Gilchrist MJ Thorne PS Ehrlichiachaffeensis antibodies in white-tailed deer Iowa 1994and 1996 Emerg Infect Dis 2000 6397ndash400

Murphy GL Ewing SA Whitworth LC et al A molec-ular and serologic survey of Ehrlichia canis E chaffeen-sis and E ewingii in dogs and ticks from OklahomaVet Parasitol 1998 79325ndash339

Paddock CD Childs JE Ehrlichia chaffeensis a prototyp-ical emerging pathogen Clin Microbiol Rev 2003 1637ndash64

Paddock CD Folk SM Shore GM et al Infections withEhrlichia chaffeensisand Ehrlichia ewingii in persons coin-fected with human immunodeficiency virus Clin InfectDis 2001 31586ndash1594

Roland WE Everett ED Cyr TL et al Ehrlichia chaffeensisin Missouri ticks Am J Trop Med Hyg 1998 59641ndash643

Smart DL and Trainer DO Serologic evidence ofVenezuelan equine encephalitis in some wild and do-mestic populations of southern Texas J Wildl Dis 197511195ndash200

Stallknecht DE Blue JL Rollor EA et al Precipitatingantibodies to bluetongue and epizootic hemorrhagicdisease viruses in white-tailed deer in the southeasternUnited States J Wildl Dis 1991 27238ndash247

Steiert JG Gilfoy F Infection rates of Amblyomma ameri-canum and Dermacentor variabilis by Ehrlichia chaffeensis

YABSLEY ET AL206

tic and changing geographic distributions animportant attribute of an effective surveillancesystem is the ability to identify both persis-tently enzootic locales and spread to new loca-tions Testing of WTD from LST infested pop-ulations over the span of 20 years showed thatonce this biologically important vector is pres-ent and WTD populations are infected with Echaffeensis there is a high probability that thepopulations will remain infected and that WTDsentinels reflect these enzootic conditions Con-versely populations with no history of LST in-festation remained consistently seronegativewhen tested multiple times These results arecompatible with a strong site-specific associa-tion between E chaffeensis antibodies and thepresence of LST (Lockhart et al 1996) Of in-terest each of two populations (Choctaw CoAL and Bolivar Co MS) that were earlier re-ported as seropositive but LST negative (Lock-hart et al 1996) were shown in this study to beboth LST positive and seropositive for E chaf-feensis The rare failure to find LST in seropos-itive WTD populations may have occurred be-cause many collections were made during thefall hunting season when LST infestations areundergoing rapid seasonal declines (Allan2001)

Four instances of E chaffeensis spread to newlocations were detected by monitoring WTDfor multiple years (Table 2) At each of these lo-cations the change in population serologic sta-tus from negative to positive coincided withdocumented appearance of LST These findingsare very similar to those from a 12-year-longmonitoring of a WTD population that demon-strated the introduction of E chaffeensis was at-tributable to the establishment of LST (Lock-hart et al 1995) Because all of these changes toseropositive status were linked to the principalvector they provide further evidence that thissurveillance system reflects important epi-demiologic factors and that it should discernchanging local HME risks

In summary the biological reasons a sentinelWTD surveillance system functions effectivelyare that the force of transmission of E chaffeen-sis is focused on WTD and that aspects of WTDbiology allow attribution of infection to a spe-cific geographic location From an operationalperspective the system has the desirable at-

tributes of applicability on a wide geographicscale and logistic feasibility by using ldquofreerdquo(hunter-harvested) samples The present workwas done in a research context however it ispossible to implement sentinel WTD serologicsurveillance for E chaffeensis in an operationalpublic health context as clearly demonstratedpreviously in Iowa (Mueller-Anneling et al2000)

ACKNOWLEDGMENTS

We are indebted to over 100 individuals rep-resenting state wildlife agency personnelSCWDS staff and white-tailed deer researchersfor help in field collection of samples but es-pecially the following individuals for coordi-nating specific collections Chris Cook (AL)John Morgan (FL) Lloyd Fox (KS) Jon Gassettand David Yancy (KY) Jeff Beringer (MO)Douglas Roscoe (NJ) Evin Stanford (NC) AAlan Kocan (OK) Charles Ruth Jr (SC) BenLayton (TN) Matt Knox (VA) and Jim Crum(WV) Sarah Cross Page Luttrell and DarrellKavanaugh from SCWDS provided excellentfield and laboratory help Work was supportedprimarily by the National Institutes of Allergyand Infectious Diseases (5 R01 AI044235-02)Further support was provided by the FederalAid to Wildlife Restoration Act (50 Stat 917)and through sponsorship from fish and wildlifeagencies in Alabama Arkansas Florida Geor-gia Kansas Kentucky Louisiana MarylandMississippi Missouri North Carolina Okla-homa Puerto Rico South Carolina TennesseeVirginia and West Virginia

REFERENCES

Allan SA Ticks (Class Arachnida Order Acarina) InSamuel WM Pybus MJ and Kocan AA ed ParasiticDiseases of Wild Mammals Ames IA Iowa State Uni-versity Press 2001 72ndash106

Anderson BE Sumner JW Dawson JE et al Detection ofthe etiologic agent of human ehrlichiosis by polymerasechain reaction J Clin Microbiol 1992 30775ndash780

Anderson BE Sims KG Olson JG et al Amblyommaamericanum a potential vector of human ehrlichiosisAm J Trop Med Hyg 1993 49239ndash244

Bloemer SR Zimmerman RH Fairbanks K Abundanceattachment sites and density estimators of lone star

205EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER

ticks (Acari Ixodidae) infesting white-tailed deer JMed Entomol 1988 25295ndash300

Boromisa RD Grimstand PR Seroconversion rates toJamestown Canyon virus among six populations ofwhite-tailed deer (Odocoileus virginianus) in Indiana JWildl Dis 1987 2723ndash33

Brandsma AR Little SE Lockhart JM et al NovelEhrlichia organism (Rickettsiales Ehrlichieae) in white-tailed deer associated with lone star tick (Acari Ixodi-dae) parasitism J Med Entomol 1999 36190ndash194

Childs JE Paddock CD The ascendancy of Amblyommaamericanum as a vector of pathogens affecting humansin the United States Annu Rev Entomol 2003 48307ndash337

Comer JA Nicholson WL Olson JG et al Serologic test-ing for human granulocytic ehrlichiosis at a national re-ferral center J Clin Microbiol 1999 37558ndash564

Dawson JE Childs JE Biggie KL et al White-tailed deeras a potential reservoir of Ehrlichia spp J Wildl Dis1994a 30162ndash168

Dawson JE Stallknecht D Howerth EW et al Suscep-tibility of white-tailed deer (Odocoileus virginianus) toinfection with Ehrlichia chaffeensis the etiologic agent ofhuman ehrlichiosis J Clin Microbiol 1994b 322725ndash2728

Dawson JE Anderson BE Fishbein DB et al Poly-merase chain reaction evidence of Ehrlichia chaffeensisetiologic agent of human ehrlichiosis in dogs fromsoutheast Virginia Am J Vet Res 1996a 571175ndash1179

Dawson JE Warner CK Baker V et al Ehrlichia-like 16SrDNA sequence from wild white-tailed deer (Odocoileusvirginianus) J Parasitol 1996b 8252ndash58

Davidson WR Lockhart JM Stallknecht DE et al Per-sistent Ehrlichia chaffeensis infection in white-tailed deerJ Wildl Dis 2001 37538ndash546

Dumler JS Walker DH Tick-borne ehrlichiosis LancetInfect Dis 2001 021ndash28

Ewing SA Dawson JE Kocan AA et al Experimentaltransmission of Ehrlichia chaffeensis (RickettsialesEhrlichieae) among white-tailed deer by Amblyommaamericanum (Acari Ixodidae) J Med Entomol 199532368ndash374

Fletcher WO Stallknecht DE Kearney MT et al Anti-bodies to vesicular stomatitis New Jersey type virus inwhite-tailed deer on Ossabaw Island Georgia 1985ndash1989 J Wildl Dis 1991 27675ndash680

Gill JS McLean RG Neitzil DF et al Serologic analysisof white-tailed deer sera for antibodies to Borreliaburgdorferi by enzyme-linked immunosorbent assayand western blotting J Clin Microbiol 1993 31318ndash322

Ijdo JW Wu C Magnarelli LA et al Detection ofEhrlichia chaffeensis DNA in Amblyomma americanumticks in Connecticut and Rhode Island J Clin Microbiol2000 384655ndash4656

Irving RP Pinger RR Vann CN et al Distribution ofEhrlichia chaffeensis (Rickettsiales Rickettsiaeceae) inAmblyomma americanum in southern Indiana and preva-lence of E chaffeensis-reactive antibodies in white-taileddeer in Indiana and Ohio in 1998 J Med Entomol 200037595ndash600

Keirans JE Lacombe EH First records of Amblyommaamericanum Ixodes (Ixodes) dentatus and Ixodes (Cer-atixodes) uriae (Acari Ixodidae) from Maine J Parasitol1998 84629ndash631

Little SE Dawson JE Lockhart JM et al Developmentand use of specific polymerase chain reaction for thedetection of an Ehrlichia-like organism in white-taileddeer J Wildl Dis 1997 33246ndash253

Little SE Stallknecht DE Lockhart JM et al Naturalcoinfection of a white-tailed deer (Odocoileus virgini-anus) population with three Ehrlichia spp J Parasitol1998 84897ndash901

Lockhart JM Davidson WR Dawson JE et al Tempo-ral association of Amblyomma americanumwith the pres-ence of Ehrlichia chaffeensis reactive antibodies in white-tailed deer J Wildl Dis 1995 31119ndash124

Lockhart JM Davidson WR Stallknecht DE et al Site-specific geographic association between Amblyommaamericanum (Acari Ixodidae) infestations and Ehrlichiachaffeensis-reactive (Rickettsiales Ehrlichieae) antibod-ies in white-tailed deer J Med Entomol 1996 33153ndash158

Lockhart JM Davidson WR Stallknecht DE et al Iso-lation of Ehrlichia chaffeensis from wild white-tailed deer(Odocoileus virginianus) confirms their role as naturalreservoir hosts J Clin Microbiol 1997a 351681ndash1686

Lockhart JM Davidson WR Stallknecht DE et al Nat-ural history of Ehrlichia chaffeensis (RickettsialesEhrlichieae) in the Piedmont physiographic province ofGeorgia J Parasitol 1997b 83887ndash894

Magnarelli LA Anderson JF Stafford KC et al Anti-bodies to multiple tick-borne pathogens of babesiosisehrlichiosis and Lyme borreliosis in white-footed miceJ Wildl Dis 1997 33466ndash473

Mueller-Anneling L Gilchrist MJ Thorne PS Ehrlichiachaffeensis antibodies in white-tailed deer Iowa 1994and 1996 Emerg Infect Dis 2000 6397ndash400

Murphy GL Ewing SA Whitworth LC et al A molec-ular and serologic survey of Ehrlichia canis E chaffeen-sis and E ewingii in dogs and ticks from OklahomaVet Parasitol 1998 79325ndash339

Paddock CD Childs JE Ehrlichia chaffeensis a prototyp-ical emerging pathogen Clin Microbiol Rev 2003 1637ndash64

Paddock CD Folk SM Shore GM et al Infections withEhrlichia chaffeensisand Ehrlichia ewingii in persons coin-fected with human immunodeficiency virus Clin InfectDis 2001 31586ndash1594

Roland WE Everett ED Cyr TL et al Ehrlichia chaffeensisin Missouri ticks Am J Trop Med Hyg 1998 59641ndash643

Smart DL and Trainer DO Serologic evidence ofVenezuelan equine encephalitis in some wild and do-mestic populations of southern Texas J Wildl Dis 197511195ndash200

Stallknecht DE Blue JL Rollor EA et al Precipitatingantibodies to bluetongue and epizootic hemorrhagicdisease viruses in white-tailed deer in the southeasternUnited States J Wildl Dis 1991 27238ndash247

Steiert JG Gilfoy F Infection rates of Amblyomma ameri-canum and Dermacentor variabilis by Ehrlichia chaffeensis

YABSLEY ET AL206

ticks (Acari Ixodidae) infesting white-tailed deer JMed Entomol 1988 25295ndash300

Boromisa RD Grimstand PR Seroconversion rates toJamestown Canyon virus among six populations ofwhite-tailed deer (Odocoileus virginianus) in Indiana JWildl Dis 1987 2723ndash33

Brandsma AR Little SE Lockhart JM et al NovelEhrlichia organism (Rickettsiales Ehrlichieae) in white-tailed deer associated with lone star tick (Acari Ixodi-dae) parasitism J Med Entomol 1999 36190ndash194

Childs JE Paddock CD The ascendancy of Amblyommaamericanum as a vector of pathogens affecting humansin the United States Annu Rev Entomol 2003 48307ndash337

Comer JA Nicholson WL Olson JG et al Serologic test-ing for human granulocytic ehrlichiosis at a national re-ferral center J Clin Microbiol 1999 37558ndash564

Dawson JE Childs JE Biggie KL et al White-tailed deeras a potential reservoir of Ehrlichia spp J Wildl Dis1994a 30162ndash168

Dawson JE Stallknecht D Howerth EW et al Suscep-tibility of white-tailed deer (Odocoileus virginianus) toinfection with Ehrlichia chaffeensis the etiologic agent ofhuman ehrlichiosis J Clin Microbiol 1994b 322725ndash2728

Dawson JE Anderson BE Fishbein DB et al Poly-merase chain reaction evidence of Ehrlichia chaffeensisetiologic agent of human ehrlichiosis in dogs fromsoutheast Virginia Am J Vet Res 1996a 571175ndash1179

Dawson JE Warner CK Baker V et al Ehrlichia-like 16SrDNA sequence from wild white-tailed deer (Odocoileusvirginianus) J Parasitol 1996b 8252ndash58

Davidson WR Lockhart JM Stallknecht DE et al Per-sistent Ehrlichia chaffeensis infection in white-tailed deerJ Wildl Dis 2001 37538ndash546

Dumler JS Walker DH Tick-borne ehrlichiosis LancetInfect Dis 2001 021ndash28

Ewing SA Dawson JE Kocan AA et al Experimentaltransmission of Ehrlichia chaffeensis (RickettsialesEhrlichieae) among white-tailed deer by Amblyommaamericanum (Acari Ixodidae) J Med Entomol 199532368ndash374

Fletcher WO Stallknecht DE Kearney MT et al Anti-bodies to vesicular stomatitis New Jersey type virus inwhite-tailed deer on Ossabaw Island Georgia 1985ndash1989 J Wildl Dis 1991 27675ndash680

Gill JS McLean RG Neitzil DF et al Serologic analysisof white-tailed deer sera for antibodies to Borreliaburgdorferi by enzyme-linked immunosorbent assayand western blotting J Clin Microbiol 1993 31318ndash322

Ijdo JW Wu C Magnarelli LA et al Detection ofEhrlichia chaffeensis DNA in Amblyomma americanumticks in Connecticut and Rhode Island J Clin Microbiol2000 384655ndash4656

Irving RP Pinger RR Vann CN et al Distribution ofEhrlichia chaffeensis (Rickettsiales Rickettsiaeceae) inAmblyomma americanum in southern Indiana and preva-lence of E chaffeensis-reactive antibodies in white-taileddeer in Indiana and Ohio in 1998 J Med Entomol 200037595ndash600

Keirans JE Lacombe EH First records of Amblyommaamericanum Ixodes (Ixodes) dentatus and Ixodes (Cer-atixodes) uriae (Acari Ixodidae) from Maine J Parasitol1998 84629ndash631

Little SE Dawson JE Lockhart JM et al Developmentand use of specific polymerase chain reaction for thedetection of an Ehrlichia-like organism in white-taileddeer J Wildl Dis 1997 33246ndash253

Little SE Stallknecht DE Lockhart JM et al Naturalcoinfection of a white-tailed deer (Odocoileus virgini-anus) population with three Ehrlichia spp J Parasitol1998 84897ndash901

Lockhart JM Davidson WR Dawson JE et al Tempo-ral association of Amblyomma americanumwith the pres-ence of Ehrlichia chaffeensis reactive antibodies in white-tailed deer J Wildl Dis 1995 31119ndash124

Lockhart JM Davidson WR Stallknecht DE et al Site-specific geographic association between Amblyommaamericanum (Acari Ixodidae) infestations and Ehrlichiachaffeensis-reactive (Rickettsiales Ehrlichieae) antibod-ies in white-tailed deer J Med Entomol 1996 33153ndash158

Lockhart JM Davidson WR Stallknecht DE et al Iso-lation of Ehrlichia chaffeensis from wild white-tailed deer(Odocoileus virginianus) confirms their role as naturalreservoir hosts J Clin Microbiol 1997a 351681ndash1686

Lockhart JM Davidson WR Stallknecht DE et al Nat-ural history of Ehrlichia chaffeensis (RickettsialesEhrlichieae) in the Piedmont physiographic province ofGeorgia J Parasitol 1997b 83887ndash894

Magnarelli LA Anderson JF Stafford KC et al Anti-bodies to multiple tick-borne pathogens of babesiosisehrlichiosis and Lyme borreliosis in white-footed miceJ Wildl Dis 1997 33466ndash473

Mueller-Anneling L Gilchrist MJ Thorne PS Ehrlichiachaffeensis antibodies in white-tailed deer Iowa 1994and 1996 Emerg Infect Dis 2000 6397ndash400

Murphy GL Ewing SA Whitworth LC et al A molec-ular and serologic survey of Ehrlichia canis E chaffeen-sis and E ewingii in dogs and ticks from OklahomaVet Parasitol 1998 79325ndash339

Paddock CD Childs JE Ehrlichia chaffeensis a prototyp-ical emerging pathogen Clin Microbiol Rev 2003 1637ndash64

Paddock CD Folk SM Shore GM et al Infections withEhrlichia chaffeensisand Ehrlichia ewingii in persons coin-fected with human immunodeficiency virus Clin InfectDis 2001 31586ndash1594

Roland WE Everett ED Cyr TL et al Ehrlichia chaffeensisin Missouri ticks Am J Trop Med Hyg 1998 59641ndash643

Smart DL and Trainer DO Serologic evidence ofVenezuelan equine encephalitis in some wild and do-mestic populations of southern Texas J Wildl Dis 197511195ndash200

Stallknecht DE Blue JL Rollor EA et al Precipitatingantibodies to bluetongue and epizootic hemorrhagicdisease viruses in white-tailed deer in the southeasternUnited States J Wildl Dis 1991 27238ndash247

Steiert JG Gilfoy F Infection rates of Amblyomma ameri-canum and Dermacentor variabilis by Ehrlichia chaffeensis

YABSLEY ET AL206

and Ehrlichia ewingii in southwest Missouri Vector-Borne Zoonotic Dis 2002 253ndash60

Sumner JW Childs JE Paddock CD Molecular cloningand characterization of the Ehrlichia chaffeensis variable-length PCR target an antigen-expressing gene that exhibits interstrain variation J Clin Microbiol 1999 371447ndash1453

Thrusfield M Veterinary Epidemiology Oxford EnglandBlackwell Science 1995 479 pp

Walls JJ Asanovich KM Bakken JS et al Serologic ev-idence of a natural infection of white-tailed deer withthe agent of human granulocytic ehrlichiosis in Wis-consin and Maryland Clin Diagn Lab Immunol 19985762ndash765

Yabsley MJ Varela AS Tate CM et al Ehrlichia ewingiiinfection in white-tailed deer (Odocoileus virginianus)Emerg Infect Dis 2002 8668ndash671

Address reprint requests toDr Michael J Yabsley

Wildlife Health BuildingSoutheastern Cooperative Wildlife Disease Study

College of Veterinary MedicineUniversity of GeorgiaAthens GA 30602

E-mail myabsleyvetugaedu

EHRLICHIA CHAFFEENSIS SURVEILLANCE IN DEER 207