BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors nonprofit publishers academic institutions researchlibraries and research funders in the common goal of maximizing access to critical research

Methodology for Sampling Questing Nymphs of Ixodes Ricinus (Acari Ixodidae)the Principal Vector of Lyme Disease in EuropeAuthor(s) Marie Vassallo Bruno Pichon Jacques Cabaret Claude Figureau and Claudine Peacuterez-EidSource Journal of Medical Entomology 37(3)335-339 2000Published By Entomological Society of AmericaDOI httpdxdoiorg1016030022-2585(2000)037[0335MFSQNO]20CO2URL httpwwwbiooneorgdoifull1016030022-2585282000290375B03353AMFSQNO5D20CO3B2

BioOne (wwwbiooneorg) is a nonprofit online aggregation of core research in the biological ecological andenvironmental sciences BioOne provides a sustainable online platform for over 170 journals and books publishedby nonprofit societies associations museums institutions and presses

Your use of this PDF the BioOne Web site and all posted and associated content indicates your acceptance ofBioOnersquos Terms of Use available at wwwbiooneorgpageterms_of_use

Usage of BioOne content is strictly limited to personal educational and non-commercial use Commercial inquiriesor rights and permissions requests should be directed to the individual publisher as copyright holder

ARTICLE

Methodology for Sampling Questing Nymphs of Ixodes Ricinus(Acari Ixodidae) the Principal Vector of Lyme Disease in Europe

MARIE VASSALLO BRUNO PICHON JACQUES CABARET1 CLAUDE FIGUREAU2

AND CLAUDINE PEREZ-EID

Unite dOtildeEcologie des Systemes Vectoriels Institut Pasteur 25 rue du Docteur Roux 75724 Paris Cedex 15 France

J Med Entomol 37(3) 335ETH339 (2000)

ABSTRACT To assess the Lyme borreliosis vector population density we set up a methodology forsampling the Ixodes ricinus L population host questing on the vegetation We focused on thecollection of the nymphal stage which is the principal stage of disease transmission to humans Thisstudy was carried out in Rambouillet forest (Yvelines France) where seven study areas weredemarcated These areas are maximally homogeneous for plant species using a THORNner scale than thephytosociological classiTHORNcation as deTHORNned by the method of landscape diagnostics Out of 23collections performed from March 1997 to May 1998 2906 I ricinus nymphs were collected Thesampling technique chosenwas the cloth lure technique The technical parameterswere studied andTHORNxed (cloth type cloth size sample size researcher position) It appeared that toweling was the bestcloth type to optimize the number of ticks collected the position of the researcher had no effecton tick samples To satisfy the criteria for correct sampling we studied representativity randomnessand nonselectivity of our methodology The spatial distribution of nymphs in a homogeneous areawas close to random and thus very few subsamples were needed to obtain a relative density whichwas representative No signiTHORNcant differences were found between random samples and followingtransect samples and nonselectivity was totally satisTHORNed because we only worked on questingnymphs We grouped the samples that presented no signiTHORNcant differences to attribute a densityindexwhichvaried from0 to 5Thismethodology appliedwith the sameparameters offers potentialfor producing comparable results from studies in different geographical areas and at different timesof the years

KEY WORDS Ixodes ricinustick sampling methodology

EPIDEMIOLOGICAL STUDY OF vector-borne diseases in-volves to a large extent study of the vector and itsecology particularly vector densities The principlevector of Lyme disease in Europe is Ixodes ricinus Lwhich transmits the spirochete Borrelia burgdorferisensu lato Johnson Hyde amp Rumpel among the wildvertebrate population and which can infect man

Ixodes ricinus is an exophilic species found in openor semiopen biotopes most often on the surface ofvegetation litter or herbaceous shrubs Although atleast three techniques are used to collect exophilicticks CO2 traps have proved to be effective for ac-tively questing tick species such as Amblyomma spp(Solberg et al 1992 Davidson et al 1994 Schulze andJordan 1996) Some authors have also used this tech-nique for sampling ofAmerican Ixodes spp (Falco andFish 1992 Schulze and Jordan 1996) For those tickspecies that adopt the sit-and-wait strategy such as Iricinus a THORNxed trap is less useful except when usingmultiple traps for extended periods (at least 7 d ac-cording to Gray [1985]) The technique known as the

ldquowalking surveyrdquo (Lane 1996 Schulze and Jordan1996) which consists of collecting ticks seen directlyon oneself after moving through the vegetation hasbeen used but there are many biases inherent in thismethod (eg differences between researchers vari-ability in the extent of contact with the vegetation)Third the technique previously used in 1932 byMacLeod and the one most widely used to collectquesting I ricinus consists of slowly dragging clothacross the vegetation as a lure The ticks which re-spond to mechanical stimuli attach themselves to thecloth during the sweep The lure can be moved lat-erally by the investigator (ldquoszligagrdquo technique) or pulledbehind (ldquodragrdquo technique)

The cloth-lure technique is currently used follow-ing a minimal methodology that consists of countingthe number of ticks caught by unit time (hours) or bysurface covered (hectares) and is generally done byextrapolation of results obtained fromaperiod shorterthan 1 h or a surface area smaller than 1 ha Further-more sampling parameters (eg vegetation typetechnical or otherwise) are not usually consideredproducing tick density data that are not comparablebetween studies To our knowledge only one studyhas addressed the deTHORNnition of an objective method-ology for I ricinus nymph sampling but is of limited

1 INRA Station de Pathologie Aviaire et de Parasitologie UnitedOtildeEcologie des Parasites 37380 Nouzilly France

2 Jardin Botanique de la Ville de Nantes 6 rue Stanislas Baudry44000 Nantes France

0022-2585000335ETH0339$02000 q 2000 Entomological Society of America

value requiring at least 50 samples per site (Daniel etal 1986) Therefore we consider it pertinent to de-velop a statistically rigorous objective methodologyhighlighting the key parameters

Materials and Methods

Study and Sampling Sites The study site was Ram-bouillet forest which is a recognized focus of Lymedisease (Zhioua et al 1996 Perez-Eid 1998) The for-est is 60 km southwest of Paris and covers an area of15000 ha The climate is oceanic with a weak conti-nental disposition beneTHORNting from a relatively lowannual precipitation of 650 mm The forest is dividedinto plots separated by roads and paths as well asbeing demarcated by landscapes of different vegeta-tion types

Apreliminary phytosociological study enabledus toestablish that the vast majority of the forest belongedto the vegetation group Quercion robori-petreae de-scribed as a deciduous Quercus robur L forest on acidsoil To take into account the variability in vegetationtypes in different plots of the forest plant speciescomposition density and surface area covered wehave adapted the method of landscape diagnostics tothe forest system (Boiret et al 1988) This methodallows us to demarcate and map relatively small areasthat are maximally homogeneous Within phytosocio-logical group this methodology enables us to identifythe different states of degradation characterized byspeciTHORNc plant species (eg stable or pioneers) Anarea deTHORNned as homogeneous has the same vegetationcharacteristics including plant species groups theirrelative species density and height

Seven areas (a-g)were thus determined each char-acteristic within the given phytosociological group 5different areas were deTHORNned within the Quercion ro-bori-petreaegroupand twowithin the coniferousPiceaexcelsa Lamarck group

Sampling Technique and Tick Stage Studied Thecloth-lure technique was used throughout The clothsurface area was THORNxed at 1 m2 (1 by 1 m) to allowunrestricted passage across all types of vegetation andto facilitate density calculations So that the entirecloth surfacewas incontactwith thevegetation a stripof wood was attached to the leading edge and pulledby string Each subsample was carried out along avegetation strip of 10 m After each subsample theticks were removed from the cloth placed into tubesand then counted and identiTHORNedApath length of 10mwas chosen empirically being equally useful duringperiods of high and low tick abundance During hightick abundance this length limits the detachment ofticks (because theyarephysically removedbyrubbingagainst the vegetation or dropping off after havingrecognized that the lure was not a host) during lowtick abundance this length provides a sufTHORNcient num-ber of ticks

From March 1997 to May 1998 samples composedof between 7 and 40 subsamples were made A par-ticular sample included all subsamples realized in agiven area date and period of the day The study

targeted only the nymphal stages that were questingIt is the stage most pertinent to disease epidemiologybecause it is themost abundant and it is anthropophilymore marked than the other stages (Gray 1991) Inaddition because of its small size the nymph oftenpasses unnoticed remaining attached to a humanlonger than the adult stage and thus having moreopportunity to transmit pathogens

Sampling Methodology To satisfy the criteria forcorrect sampling (Southwood 1966) we studied theldquorepresentativityrdquo randomness and nonselectivity ofthe technique

The Minimum Sampled Surface Area Required (Sam-pling Representativity) Randomness and Nonselectiv-ityAssessmentof theminimumrepresentative surfacearea in each of the seven deTHORNned study areas fordifferent times of the year was made to optimize themethodology with a view to sampling regularlyTherefore we aimed to THORNnd the smallest possible sur-face area that is the smallest number of samples percollection required to give a representative sample ofthe area First the spatial distribution of each collec-tion was assessed by the Poisson dispersal index I (I 5Vm where V 5 variance and M 5 mean of thesample we tested if I was signiTHORNcantly different from1 by examination of the auxiliary variable [([N 2 1]3V)m]wheren is the sample sizewhich follows thechi-square law at n 2 1 degrees of freedom at a 5005) Analysis of the distribution provides an indica-tion of the representativity of the collection and if inthe case of a highly signiTHORNcant difference fromPoisson(such as an aggregated distribution) for example fur-ther samples are required Second a decreasing num-ber of subsamples (15 10 or 5) were randomly se-lected from each sample (Total n subsamples)regardless of their distribution and the minimumnumber was the smallest that gave no signiTHORNcant dif-ference in tick density from the overall collection

Two other sampling methods were compared sam-pling by transect or randomly The two methods werecarried out on the same area and at the same time andwere repeated three times in threedifferent areasThenonselectivity of themethodwasnot assessedbecausefrom the outset the study was selective for the quest-ing nymphal stage

Cloth-LureTechnical ParametersThe effect of thecloth type was evaluated comparing four differenttypes of cloth cotton woollen szligannel ldquomolletonrdquo(soft thick cotton) and toweling (a towel spongecloth) To carry out the comparison under standard-ized conditions we chose to use all four cloth types atonce using strips of cloth 25 cmby 1m in parallel Therelative positions of the cloth strips were changedregularly at randomtoexcludeanypositional bias Thecomparison was carried out for seven samples allperformed at the same speed (50 cm z s21) and withthe investigator always stationed ahead pulling thelure To avoid losing information because of the vari-ability (eg tick density time of day) between sam-ples each sample was analyzed separately

Two parameters inherent to the investigator werealso assessed his positionwith respect to the cloth and

336 JOURNAL OF MEDICAL ENTOMOLOGY Vol 37 no 3

the speed of passage The effect of position (in frontof or beside the cloth) was assessed twice on twosamples (each composed of 30 subsamples) and per-formed at the same speed (50 cm z s21) during Apriland May 1998 Each of the two couplet samples wascarried out in the same area and on two consecutivedays that had the same climatic conditions The effectof the speedofpassage twowasalso assessedunder thesame conditions using speeds of one m z s21 and 50cm z s21

Statistical Analyses Statistical analyses were per-formed inSIMSTAT(NPeladeauProvalisResearch)using nonparametric tests MannETHWhitney (used tocompare the mean of two tick collections) andKruskallETHWallis (used to compare the mean of n (2)tick collections) Factorial analysis of variance(ANOVA) was carried out for those collections with30 subsamples to test the effect of the two factors(tissue type and collection)

Grouping of Collection for Estimation of IndexDensity To provide an estimate of index nymph den-sity we regrouped the samples that were not signiTHORN-cantly different This index density enables a straight-forward heuristic comparison between differentsamples and facilitates the acarologic index calcula-tions (probability of THORNnding an infected nymph for agiven area and period of time) and their comparisons

Results

A total of 2906 I ricinus nymphs was found in 43collections relating to the whole studies

Representativity and Randomness The determina-tion of the minimum surface area required for repre-sentative sampling of an area for Rambouillet forestwas carried out on the data from 23 samples (com-posed of 7ETH40 subsamples) performed at regular in-tervals during the study in all the described areas Thecalculations of the Poisson dispersion index (I) showsthat the spatial distribution of nymphs whatever thearea or date or tick abundance is either random (10samples 085 I 148 where I does not differsigniTHORNcantly from1)ormildly aggregated (13 samples154 I 898 Imean5 25) (Table 1)Theminimumrepresentative sample size (number of subsamples)could therefore be small In this case we chose 5 10or 15 random subsamples from each collection (Totaln subsamples) for comparison There was no signiTHORN-cant difference for any sample between the four sub-sample size groups (5 10 15 or n) (KruskallETHWallis052 P 098) Five subsamples carried out in ahomogeneous study area were therefore sufTHORNcient toobtain a mean number of nymphs per 10 m2 sub-samples which was representative of the samplingarea

The two sampling methods transect and randomwere not found to produce signiTHORNcantly differentnymph numbers (MannETHWhitney 051 P 057)

Cloth-Lure Technical Parameters The study of theeffect of cloth type on nymph numbers collected wascarried out for seven samples signiTHORNcant differencesbetween the four cloth types were found in three of

the seven samples (Table 2)Apairwise comparisonofthe different cloth types carried out on the threesigniTHORNcant samples taken independently (Table 2)showed that toweling optimizes the collection (MannETHWhitney P 001 whichever collection was ana-lyzed) This cloth difference was conTHORNrmed by anal-ysis of the four other samples by ANOVA (P 0001)which showed that there were signiTHORNcant differencesbetween the cloth types despite the nonsigniTHORNcantKruskallETHWallis test (Table 2) because of the greatersensitivity of ANOVA when analyzing low tick abun-dance

A slow speed of passage (50 cms21) was clearlybetter for nymph collection (Mann-Whitney P 0001) No difference however was found betweenthe two positions of the investigator with respect tothe cloth (Mann-Whitney 031 P 043)

Grouping of Samples for Estimation of Index Den-sity Sampleswere grouped to be able to assign to eacha level of relative tick density The comparison of 23

Table 1 Poisson law dispersion index established for 23 sam-ples performed in different areas of Rambouillet forest April1997ndashMay 1998

Sample I Date Area A n

1 176 Late April 1997 a 05 302 148a Late April 1997 b 05 303 343 Late April 1997 c 215 394 093a Late April 1997 d 05 305 423 Late April 1997 e 51 306 371 Mid-May 1997 b 174 207 51 Mid-May 1997 e 1161 238 228 Mid-May 1997 e 418 409 898 Late May 1997 e9 2225 16

10 211 Late May 1997 c 123 3011 345 Late May 1997 b 574 4012 227 Late May 1997 b9 1771 713 245 Late June 1997 d 26 2014 143a Late June 1997 g 076 2115 369 Late June 1997 e 576 2116 112

aLate June 1997 c 06 20

17 13a Early March 1998 b 02 2118 085a Early March 1998 e 018 2319 129a Late March 1998 b 204 2020 095a Late March 1998 g 32 2021 154 Mid-May 1998 b 44 2022 088a Mid-May 1998 f 275 1823 126a Late May 1998 e 1433 16

I Poisson law dispersion index a to g Rambouillet forest areas Athe average number of nymphs collected per sub sample n the totalnumber of subsamples per sample

a Samples where I does not signiTHORNcantly differ from 1

Table 2 KruskallndashWallis test P values for the comparison ofmean nymph number (per 10 m2 subsample) collected using the 4cloth types

Sample P Mean 6 SE n

1 037 102 6 009 402 0094 2425 6 106 403 0059 277 6 123 304 019 009 6 0004 405 0002a 1471 6 609 306 0042a 527 6 214 277 0003a 769 6 346 25

n total number of subsamples per collectiona SigniTHORNcant differences found between the cloth types

May 2000 VASSALLO ET AL SAMPLING METHODOLOGY FOR I ricinus 337

samples (on samples of n subsamples on the one handand THORNve subsamples on the other) suggest the exis-tence of signiTHORNcant differences in nymphal densitybetween samples (KruskallETHWallis P 00001 in bothcases) To avoid testing samples pairwise we directlycompared the 90 conTHORNdence intervals (CI) for allsamples The conTHORNdence intervalswere established bybootstrap on the estimated mean (resampling withreplacement 500 times) (Fig 1)

Most samples divide cleanly into groups those bor-derline collections were tested against their adjacentgroups to assess whether they qualiTHORNed as a separategroup or belonged to a neighboring group Five dis-tinct groups were classiTHORNed according to their densityindex 0 90 CI 116 group 1 1455 90 CI 441 group 2 45 90 CI 738 group 3 913 90CI 1756 group 4 1452 90 CI 2806 group 5

Discussion

For studies of vector densities and from that theprevalence of infection an objective sampling meth-odology is essential The THORNrst step is to demarcatehomogeneous botanical areas because the vegetationreszligects an ensemble of factors of the biotope thatdetermine to a large extent the density of exophilicticks such as I ricinus ClassiTHORNcation of the vegetationonly by large scale grouping (phytosociology) is in-sufTHORNcient and a THORNner scale classiTHORNcation not only byspecies present but also by their dimensions (breadthand height) is necessary We used the method of

landscape diagnostics (Boiret et al 1988) Howeverthis method does not take into account the inszliguenceof tickhostswhich the investigatormustdoandhencedetermine and demarcate the ldquomicro-areasrdquo inside thestudy area such as favored animal resting sites andanimal paths (For example the areas b9 and e9 Table1) These micro-areas can have questing nymph den-sities three or four times greater than the rest of thearea which could explain the signiTHORNcant variationfound in previous studies

The second step is to establish a minimum numberof samples that are required to give a representativesample of the study area In our case having eluci-dated that the spatial distribution of nymphs was closeto random we were able to reduce the number ofsubsamples required even during times of low abun-dance There was no signiTHORNcant difference betweensample sizes of 5 10 15 or n (total subsample size ina sample) in thedensityofnymphscollectedwhateverthe area time of year or tick abundance The methodis therefore simple and generally applicable differingsubstantially from the methodology of Daniel et al(1986) However it is advisable to survey the mini-mum surface area required regularly because verte-brate hosts may change their habits and therefore thedistribution of ticks

No signiTHORNcant difference in tick number was foundbetween a sample taken at random or transect Thiswas expected having found that the spatial distribu-tion of nymphs was random and conTHORNrms that ourmethodology conforms to the criterion of sampling at

Fig 1 ConTHORNdence intervals of the mean number of nymphs collected (range 0ETH30) per 10 m2 subsample for 23 samplesperformed from April 1997 to May 1998 in the Rambouillet forest

338 JOURNAL OF MEDICAL ENTOMOLOGY Vol 37 no 3

random Grouping samples according to their conTHORN-dence intervals enables rapid visualization of the dataand determination of variation in tick densities be-tween sites Parallel to these fundamental aspects ofmethodology we have been able to establish the op-timal nature of the cloth-lure as well as the speed ofpassage and the position of the investigator althoughthe position does not seem to make a difference

This methodology conforms to the three indispens-able criteria for correct sampling representativityrandomness and nonselectivity (because we are onlyworking with questing nymphs) Such a methodologyprovides a valid estimate of relative tick density and sowill enable the comparison of results collected inother sites

Acknowledgments

We thank J P Widmer (OfTHORNce National des Forets) forgrantingus free access to theRambouillet forest andRichardPaul for reading the manuscript This work was supported bya grant from the Fondation Merieux (Lyon France) to MV

References Cited

Boiret F J B Bouzille J M Gehu and M Godeau 1988Phytosociologie paysagere du systeme pelouses-landesdes falaises des otildeles ouest et sud armoricaines ColloquesPhytosociologiques Versailles

Daniel M V Cerny and EIA Korenberg 1986 Contri-bution to the methods of estimating absolute tick num-bers Folia Parasitol 33 371ETH379

Davidson W R D A Siefken and L H Creekmore 1994Inszliguence of annual abundance of Amblyomma america-num in central Georgia J Med Vet 31 67ETH71

Falco R C and D A Fish 1992 Comparison of methodsfor sampling the deer tick Ixodes dammini in a Lymedisease endemic area Exp Appl Acarol 14 165ETH173

Gray J S 1985 A carbon dioxide trap for prolonged sam-pling of Ixodes ricinus L populations Exp Appl Acarol1 35ETH44

Gray J S 1991 The development and seasonal activity ofthe tick Ixodes ricinus a vector of Lyme borreliosis RevMed Vet Entomol 79 323ETH333

Lane R S 1996 Risk of human exposure to vector ticks inheavily used recreational area in Northern CaliforniaAm J Trop Med Hyg 55 165ETH173

MacLeod J 1932 The bionomics of Ixodes ricinus thesheep tick of Scotland Parasitology 24 382ETH400

Perez-Eid C B Pichon E Zhioua N Tremel R VilleretD Deruaz L Mousson M Vassallo and E Ferquel1998 Lymeborreliosis emergent disease linkedwith theenvironment Bull Acad Natl Med 182 267ETH83

Schulze T L and R A Jordan 1996 Seasonal and longterm variations in abundance of adult Ixodes scapularis indifferent coastal plainhabitats ofNew-Jersey JMedVet33 963ETH970

Solberg V B K Neidhardt M R Sardelis C HildebrandtF J Hoffmann and L R Boobar 1992 Quantitativeevaluation of sampling methods for Ixodes dammini andAmblyomma americanum (Acari Ixodidae) J Med Vet29 451ETH456

Southwood TRE 1966 Ecological methods with particu-lar reference to the study of insect populationsMethuenLondon

Zhioua E D Postic F Rodhain and C Perez-Eid 1996Infection of Ixodes ricinus (Acari Ixodidae) by Borreliaburgdorferi in Ile de France J Med Entomol 33 694ETH697

Received for publication 17 August 1999 accepted 3 Decem-ber 1999

May 2000 VASSALLO ET AL SAMPLING METHODOLOGY FOR I ricinus 339

value requiring at least 50 samples per site (Daniel etal 1986) Therefore we consider it pertinent to de-velop a statistically rigorous objective methodologyhighlighting the key parameters

Materials and Methods

Study and Sampling Sites The study site was Ram-bouillet forest which is a recognized focus of Lymedisease (Zhioua et al 1996 Perez-Eid 1998) The for-est is 60 km southwest of Paris and covers an area of15000 ha The climate is oceanic with a weak conti-nental disposition beneTHORNting from a relatively lowannual precipitation of 650 mm The forest is dividedinto plots separated by roads and paths as well asbeing demarcated by landscapes of different vegeta-tion types

Apreliminary phytosociological study enabledus toestablish that the vast majority of the forest belongedto the vegetation group Quercion robori-petreae de-scribed as a deciduous Quercus robur L forest on acidsoil To take into account the variability in vegetationtypes in different plots of the forest plant speciescomposition density and surface area covered wehave adapted the method of landscape diagnostics tothe forest system (Boiret et al 1988) This methodallows us to demarcate and map relatively small areasthat are maximally homogeneous Within phytosocio-logical group this methodology enables us to identifythe different states of degradation characterized byspeciTHORNc plant species (eg stable or pioneers) Anarea deTHORNned as homogeneous has the same vegetationcharacteristics including plant species groups theirrelative species density and height

Seven areas (a-g)were thus determined each char-acteristic within the given phytosociological group 5different areas were deTHORNned within the Quercion ro-bori-petreaegroupand twowithin the coniferousPiceaexcelsa Lamarck group

Sampling Technique and Tick Stage Studied Thecloth-lure technique was used throughout The clothsurface area was THORNxed at 1 m2 (1 by 1 m) to allowunrestricted passage across all types of vegetation andto facilitate density calculations So that the entirecloth surfacewas incontactwith thevegetation a stripof wood was attached to the leading edge and pulledby string Each subsample was carried out along avegetation strip of 10 m After each subsample theticks were removed from the cloth placed into tubesand then counted and identiTHORNedApath length of 10mwas chosen empirically being equally useful duringperiods of high and low tick abundance During hightick abundance this length limits the detachment ofticks (because theyarephysically removedbyrubbingagainst the vegetation or dropping off after havingrecognized that the lure was not a host) during lowtick abundance this length provides a sufTHORNcient num-ber of ticks

From March 1997 to May 1998 samples composedof between 7 and 40 subsamples were made A par-ticular sample included all subsamples realized in agiven area date and period of the day The study

targeted only the nymphal stages that were questingIt is the stage most pertinent to disease epidemiologybecause it is themost abundant and it is anthropophilymore marked than the other stages (Gray 1991) Inaddition because of its small size the nymph oftenpasses unnoticed remaining attached to a humanlonger than the adult stage and thus having moreopportunity to transmit pathogens

Sampling Methodology To satisfy the criteria forcorrect sampling (Southwood 1966) we studied theldquorepresentativityrdquo randomness and nonselectivity ofthe technique

The Minimum Sampled Surface Area Required (Sam-pling Representativity) Randomness and Nonselectiv-ityAssessmentof theminimumrepresentative surfacearea in each of the seven deTHORNned study areas fordifferent times of the year was made to optimize themethodology with a view to sampling regularlyTherefore we aimed to THORNnd the smallest possible sur-face area that is the smallest number of samples percollection required to give a representative sample ofthe area First the spatial distribution of each collec-tion was assessed by the Poisson dispersal index I (I 5Vm where V 5 variance and M 5 mean of thesample we tested if I was signiTHORNcantly different from1 by examination of the auxiliary variable [([N 2 1]3V)m]wheren is the sample sizewhich follows thechi-square law at n 2 1 degrees of freedom at a 5005) Analysis of the distribution provides an indica-tion of the representativity of the collection and if inthe case of a highly signiTHORNcant difference fromPoisson(such as an aggregated distribution) for example fur-ther samples are required Second a decreasing num-ber of subsamples (15 10 or 5) were randomly se-lected from each sample (Total n subsamples)regardless of their distribution and the minimumnumber was the smallest that gave no signiTHORNcant dif-ference in tick density from the overall collection

Two other sampling methods were compared sam-pling by transect or randomly The two methods werecarried out on the same area and at the same time andwere repeated three times in threedifferent areasThenonselectivity of themethodwasnot assessedbecausefrom the outset the study was selective for the quest-ing nymphal stage

Cloth-LureTechnical ParametersThe effect of thecloth type was evaluated comparing four differenttypes of cloth cotton woollen szligannel ldquomolletonrdquo(soft thick cotton) and toweling (a towel spongecloth) To carry out the comparison under standard-ized conditions we chose to use all four cloth types atonce using strips of cloth 25 cmby 1m in parallel Therelative positions of the cloth strips were changedregularly at randomtoexcludeanypositional bias Thecomparison was carried out for seven samples allperformed at the same speed (50 cm z s21) and withthe investigator always stationed ahead pulling thelure To avoid losing information because of the vari-ability (eg tick density time of day) between sam-ples each sample was analyzed separately

Two parameters inherent to the investigator werealso assessed his positionwith respect to the cloth and

336 JOURNAL OF MEDICAL ENTOMOLOGY Vol 37 no 3

the speed of passage The effect of position (in frontof or beside the cloth) was assessed twice on twosamples (each composed of 30 subsamples) and per-formed at the same speed (50 cm z s21) during Apriland May 1998 Each of the two couplet samples wascarried out in the same area and on two consecutivedays that had the same climatic conditions The effectof the speedofpassage twowasalso assessedunder thesame conditions using speeds of one m z s21 and 50cm z s21

Statistical Analyses Statistical analyses were per-formed inSIMSTAT(NPeladeauProvalisResearch)using nonparametric tests MannETHWhitney (used tocompare the mean of two tick collections) andKruskallETHWallis (used to compare the mean of n (2)tick collections) Factorial analysis of variance(ANOVA) was carried out for those collections with30 subsamples to test the effect of the two factors(tissue type and collection)

Grouping of Collection for Estimation of IndexDensity To provide an estimate of index nymph den-sity we regrouped the samples that were not signiTHORN-cantly different This index density enables a straight-forward heuristic comparison between differentsamples and facilitates the acarologic index calcula-tions (probability of THORNnding an infected nymph for agiven area and period of time) and their comparisons

Results

A total of 2906 I ricinus nymphs was found in 43collections relating to the whole studies

Representativity and Randomness The determina-tion of the minimum surface area required for repre-sentative sampling of an area for Rambouillet forestwas carried out on the data from 23 samples (com-posed of 7ETH40 subsamples) performed at regular in-tervals during the study in all the described areas Thecalculations of the Poisson dispersion index (I) showsthat the spatial distribution of nymphs whatever thearea or date or tick abundance is either random (10samples 085 I 148 where I does not differsigniTHORNcantly from1)ormildly aggregated (13 samples154 I 898 Imean5 25) (Table 1)Theminimumrepresentative sample size (number of subsamples)could therefore be small In this case we chose 5 10or 15 random subsamples from each collection (Totaln subsamples) for comparison There was no signiTHORN-cant difference for any sample between the four sub-sample size groups (5 10 15 or n) (KruskallETHWallis052 P 098) Five subsamples carried out in ahomogeneous study area were therefore sufTHORNcient toobtain a mean number of nymphs per 10 m2 sub-samples which was representative of the samplingarea

The two sampling methods transect and randomwere not found to produce signiTHORNcantly differentnymph numbers (MannETHWhitney 051 P 057)

Cloth-Lure Technical Parameters The study of theeffect of cloth type on nymph numbers collected wascarried out for seven samples signiTHORNcant differencesbetween the four cloth types were found in three of

the seven samples (Table 2)Apairwise comparisonofthe different cloth types carried out on the threesigniTHORNcant samples taken independently (Table 2)showed that toweling optimizes the collection (MannETHWhitney P 001 whichever collection was ana-lyzed) This cloth difference was conTHORNrmed by anal-ysis of the four other samples by ANOVA (P 0001)which showed that there were signiTHORNcant differencesbetween the cloth types despite the nonsigniTHORNcantKruskallETHWallis test (Table 2) because of the greatersensitivity of ANOVA when analyzing low tick abun-dance

A slow speed of passage (50 cms21) was clearlybetter for nymph collection (Mann-Whitney P 0001) No difference however was found betweenthe two positions of the investigator with respect tothe cloth (Mann-Whitney 031 P 043)

Grouping of Samples for Estimation of Index Den-sity Sampleswere grouped to be able to assign to eacha level of relative tick density The comparison of 23

Table 1 Poisson law dispersion index established for 23 sam-ples performed in different areas of Rambouillet forest April1997ndashMay 1998

Sample I Date Area A n

1 176 Late April 1997 a 05 302 148a Late April 1997 b 05 303 343 Late April 1997 c 215 394 093a Late April 1997 d 05 305 423 Late April 1997 e 51 306 371 Mid-May 1997 b 174 207 51 Mid-May 1997 e 1161 238 228 Mid-May 1997 e 418 409 898 Late May 1997 e9 2225 16

10 211 Late May 1997 c 123 3011 345 Late May 1997 b 574 4012 227 Late May 1997 b9 1771 713 245 Late June 1997 d 26 2014 143a Late June 1997 g 076 2115 369 Late June 1997 e 576 2116 112

aLate June 1997 c 06 20

17 13a Early March 1998 b 02 2118 085a Early March 1998 e 018 2319 129a Late March 1998 b 204 2020 095a Late March 1998 g 32 2021 154 Mid-May 1998 b 44 2022 088a Mid-May 1998 f 275 1823 126a Late May 1998 e 1433 16

I Poisson law dispersion index a to g Rambouillet forest areas Athe average number of nymphs collected per sub sample n the totalnumber of subsamples per sample

a Samples where I does not signiTHORNcantly differ from 1

Table 2 KruskallndashWallis test P values for the comparison ofmean nymph number (per 10 m2 subsample) collected using the 4cloth types

Sample P Mean 6 SE n

1 037 102 6 009 402 0094 2425 6 106 403 0059 277 6 123 304 019 009 6 0004 405 0002a 1471 6 609 306 0042a 527 6 214 277 0003a 769 6 346 25

n total number of subsamples per collectiona SigniTHORNcant differences found between the cloth types

May 2000 VASSALLO ET AL SAMPLING METHODOLOGY FOR I ricinus 337

samples (on samples of n subsamples on the one handand THORNve subsamples on the other) suggest the exis-tence of signiTHORNcant differences in nymphal densitybetween samples (KruskallETHWallis P 00001 in bothcases) To avoid testing samples pairwise we directlycompared the 90 conTHORNdence intervals (CI) for allsamples The conTHORNdence intervalswere established bybootstrap on the estimated mean (resampling withreplacement 500 times) (Fig 1)

Most samples divide cleanly into groups those bor-derline collections were tested against their adjacentgroups to assess whether they qualiTHORNed as a separategroup or belonged to a neighboring group Five dis-tinct groups were classiTHORNed according to their densityindex 0 90 CI 116 group 1 1455 90 CI 441 group 2 45 90 CI 738 group 3 913 90CI 1756 group 4 1452 90 CI 2806 group 5

Discussion

For studies of vector densities and from that theprevalence of infection an objective sampling meth-odology is essential The THORNrst step is to demarcatehomogeneous botanical areas because the vegetationreszligects an ensemble of factors of the biotope thatdetermine to a large extent the density of exophilicticks such as I ricinus ClassiTHORNcation of the vegetationonly by large scale grouping (phytosociology) is in-sufTHORNcient and a THORNner scale classiTHORNcation not only byspecies present but also by their dimensions (breadthand height) is necessary We used the method of

landscape diagnostics (Boiret et al 1988) Howeverthis method does not take into account the inszliguenceof tickhostswhich the investigatormustdoandhencedetermine and demarcate the ldquomicro-areasrdquo inside thestudy area such as favored animal resting sites andanimal paths (For example the areas b9 and e9 Table1) These micro-areas can have questing nymph den-sities three or four times greater than the rest of thearea which could explain the signiTHORNcant variationfound in previous studies

The second step is to establish a minimum numberof samples that are required to give a representativesample of the study area In our case having eluci-dated that the spatial distribution of nymphs was closeto random we were able to reduce the number ofsubsamples required even during times of low abun-dance There was no signiTHORNcant difference betweensample sizes of 5 10 15 or n (total subsample size ina sample) in thedensityofnymphscollectedwhateverthe area time of year or tick abundance The methodis therefore simple and generally applicable differingsubstantially from the methodology of Daniel et al(1986) However it is advisable to survey the mini-mum surface area required regularly because verte-brate hosts may change their habits and therefore thedistribution of ticks

No signiTHORNcant difference in tick number was foundbetween a sample taken at random or transect Thiswas expected having found that the spatial distribu-tion of nymphs was random and conTHORNrms that ourmethodology conforms to the criterion of sampling at

Fig 1 ConTHORNdence intervals of the mean number of nymphs collected (range 0ETH30) per 10 m2 subsample for 23 samplesperformed from April 1997 to May 1998 in the Rambouillet forest

338 JOURNAL OF MEDICAL ENTOMOLOGY Vol 37 no 3

random Grouping samples according to their conTHORN-dence intervals enables rapid visualization of the dataand determination of variation in tick densities be-tween sites Parallel to these fundamental aspects ofmethodology we have been able to establish the op-timal nature of the cloth-lure as well as the speed ofpassage and the position of the investigator althoughthe position does not seem to make a difference

This methodology conforms to the three indispens-able criteria for correct sampling representativityrandomness and nonselectivity (because we are onlyworking with questing nymphs) Such a methodologyprovides a valid estimate of relative tick density and sowill enable the comparison of results collected inother sites

Acknowledgments

We thank J P Widmer (OfTHORNce National des Forets) forgrantingus free access to theRambouillet forest andRichardPaul for reading the manuscript This work was supported bya grant from the Fondation Merieux (Lyon France) to MV

References Cited

Boiret F J B Bouzille J M Gehu and M Godeau 1988Phytosociologie paysagere du systeme pelouses-landesdes falaises des otildeles ouest et sud armoricaines ColloquesPhytosociologiques Versailles

Daniel M V Cerny and EIA Korenberg 1986 Contri-bution to the methods of estimating absolute tick num-bers Folia Parasitol 33 371ETH379

Davidson W R D A Siefken and L H Creekmore 1994Inszliguence of annual abundance of Amblyomma america-num in central Georgia J Med Vet 31 67ETH71

Falco R C and D A Fish 1992 Comparison of methodsfor sampling the deer tick Ixodes dammini in a Lymedisease endemic area Exp Appl Acarol 14 165ETH173

Gray J S 1985 A carbon dioxide trap for prolonged sam-pling of Ixodes ricinus L populations Exp Appl Acarol1 35ETH44

Gray J S 1991 The development and seasonal activity ofthe tick Ixodes ricinus a vector of Lyme borreliosis RevMed Vet Entomol 79 323ETH333

Lane R S 1996 Risk of human exposure to vector ticks inheavily used recreational area in Northern CaliforniaAm J Trop Med Hyg 55 165ETH173

MacLeod J 1932 The bionomics of Ixodes ricinus thesheep tick of Scotland Parasitology 24 382ETH400

Perez-Eid C B Pichon E Zhioua N Tremel R VilleretD Deruaz L Mousson M Vassallo and E Ferquel1998 Lymeborreliosis emergent disease linkedwith theenvironment Bull Acad Natl Med 182 267ETH83

Schulze T L and R A Jordan 1996 Seasonal and longterm variations in abundance of adult Ixodes scapularis indifferent coastal plainhabitats ofNew-Jersey JMedVet33 963ETH970

Solberg V B K Neidhardt M R Sardelis C HildebrandtF J Hoffmann and L R Boobar 1992 Quantitativeevaluation of sampling methods for Ixodes dammini andAmblyomma americanum (Acari Ixodidae) J Med Vet29 451ETH456

Southwood TRE 1966 Ecological methods with particu-lar reference to the study of insect populationsMethuenLondon

Zhioua E D Postic F Rodhain and C Perez-Eid 1996Infection of Ixodes ricinus (Acari Ixodidae) by Borreliaburgdorferi in Ile de France J Med Entomol 33 694ETH697

Received for publication 17 August 1999 accepted 3 Decem-ber 1999

May 2000 VASSALLO ET AL SAMPLING METHODOLOGY FOR I ricinus 339

the speed of passage The effect of position (in frontof or beside the cloth) was assessed twice on twosamples (each composed of 30 subsamples) and per-formed at the same speed (50 cm z s21) during Apriland May 1998 Each of the two couplet samples wascarried out in the same area and on two consecutivedays that had the same climatic conditions The effectof the speedofpassage twowasalso assessedunder thesame conditions using speeds of one m z s21 and 50cm z s21

Statistical Analyses Statistical analyses were per-formed inSIMSTAT(NPeladeauProvalisResearch)using nonparametric tests MannETHWhitney (used tocompare the mean of two tick collections) andKruskallETHWallis (used to compare the mean of n (2)tick collections) Factorial analysis of variance(ANOVA) was carried out for those collections with30 subsamples to test the effect of the two factors(tissue type and collection)

Grouping of Collection for Estimation of IndexDensity To provide an estimate of index nymph den-sity we regrouped the samples that were not signiTHORN-cantly different This index density enables a straight-forward heuristic comparison between differentsamples and facilitates the acarologic index calcula-tions (probability of THORNnding an infected nymph for agiven area and period of time) and their comparisons

Results

A total of 2906 I ricinus nymphs was found in 43collections relating to the whole studies

Representativity and Randomness The determina-tion of the minimum surface area required for repre-sentative sampling of an area for Rambouillet forestwas carried out on the data from 23 samples (com-posed of 7ETH40 subsamples) performed at regular in-tervals during the study in all the described areas Thecalculations of the Poisson dispersion index (I) showsthat the spatial distribution of nymphs whatever thearea or date or tick abundance is either random (10samples 085 I 148 where I does not differsigniTHORNcantly from1)ormildly aggregated (13 samples154 I 898 Imean5 25) (Table 1)Theminimumrepresentative sample size (number of subsamples)could therefore be small In this case we chose 5 10or 15 random subsamples from each collection (Totaln subsamples) for comparison There was no signiTHORN-cant difference for any sample between the four sub-sample size groups (5 10 15 or n) (KruskallETHWallis052 P 098) Five subsamples carried out in ahomogeneous study area were therefore sufTHORNcient toobtain a mean number of nymphs per 10 m2 sub-samples which was representative of the samplingarea

The two sampling methods transect and randomwere not found to produce signiTHORNcantly differentnymph numbers (MannETHWhitney 051 P 057)

Cloth-Lure Technical Parameters The study of theeffect of cloth type on nymph numbers collected wascarried out for seven samples signiTHORNcant differencesbetween the four cloth types were found in three of

the seven samples (Table 2)Apairwise comparisonofthe different cloth types carried out on the threesigniTHORNcant samples taken independently (Table 2)showed that toweling optimizes the collection (MannETHWhitney P 001 whichever collection was ana-lyzed) This cloth difference was conTHORNrmed by anal-ysis of the four other samples by ANOVA (P 0001)which showed that there were signiTHORNcant differencesbetween the cloth types despite the nonsigniTHORNcantKruskallETHWallis test (Table 2) because of the greatersensitivity of ANOVA when analyzing low tick abun-dance

A slow speed of passage (50 cms21) was clearlybetter for nymph collection (Mann-Whitney P 0001) No difference however was found betweenthe two positions of the investigator with respect tothe cloth (Mann-Whitney 031 P 043)

Grouping of Samples for Estimation of Index Den-sity Sampleswere grouped to be able to assign to eacha level of relative tick density The comparison of 23

Table 1 Poisson law dispersion index established for 23 sam-ples performed in different areas of Rambouillet forest April1997ndashMay 1998

Sample I Date Area A n

1 176 Late April 1997 a 05 302 148a Late April 1997 b 05 303 343 Late April 1997 c 215 394 093a Late April 1997 d 05 305 423 Late April 1997 e 51 306 371 Mid-May 1997 b 174 207 51 Mid-May 1997 e 1161 238 228 Mid-May 1997 e 418 409 898 Late May 1997 e9 2225 16

10 211 Late May 1997 c 123 3011 345 Late May 1997 b 574 4012 227 Late May 1997 b9 1771 713 245 Late June 1997 d 26 2014 143a Late June 1997 g 076 2115 369 Late June 1997 e 576 2116 112

aLate June 1997 c 06 20

17 13a Early March 1998 b 02 2118 085a Early March 1998 e 018 2319 129a Late March 1998 b 204 2020 095a Late March 1998 g 32 2021 154 Mid-May 1998 b 44 2022 088a Mid-May 1998 f 275 1823 126a Late May 1998 e 1433 16

I Poisson law dispersion index a to g Rambouillet forest areas Athe average number of nymphs collected per sub sample n the totalnumber of subsamples per sample

a Samples where I does not signiTHORNcantly differ from 1

Table 2 KruskallndashWallis test P values for the comparison ofmean nymph number (per 10 m2 subsample) collected using the 4cloth types

Sample P Mean 6 SE n

1 037 102 6 009 402 0094 2425 6 106 403 0059 277 6 123 304 019 009 6 0004 405 0002a 1471 6 609 306 0042a 527 6 214 277 0003a 769 6 346 25

n total number of subsamples per collectiona SigniTHORNcant differences found between the cloth types

May 2000 VASSALLO ET AL SAMPLING METHODOLOGY FOR I ricinus 337

samples (on samples of n subsamples on the one handand THORNve subsamples on the other) suggest the exis-tence of signiTHORNcant differences in nymphal densitybetween samples (KruskallETHWallis P 00001 in bothcases) To avoid testing samples pairwise we directlycompared the 90 conTHORNdence intervals (CI) for allsamples The conTHORNdence intervalswere established bybootstrap on the estimated mean (resampling withreplacement 500 times) (Fig 1)

Most samples divide cleanly into groups those bor-derline collections were tested against their adjacentgroups to assess whether they qualiTHORNed as a separategroup or belonged to a neighboring group Five dis-tinct groups were classiTHORNed according to their densityindex 0 90 CI 116 group 1 1455 90 CI 441 group 2 45 90 CI 738 group 3 913 90CI 1756 group 4 1452 90 CI 2806 group 5

Discussion

For studies of vector densities and from that theprevalence of infection an objective sampling meth-odology is essential The THORNrst step is to demarcatehomogeneous botanical areas because the vegetationreszligects an ensemble of factors of the biotope thatdetermine to a large extent the density of exophilicticks such as I ricinus ClassiTHORNcation of the vegetationonly by large scale grouping (phytosociology) is in-sufTHORNcient and a THORNner scale classiTHORNcation not only byspecies present but also by their dimensions (breadthand height) is necessary We used the method of

landscape diagnostics (Boiret et al 1988) Howeverthis method does not take into account the inszliguenceof tickhostswhich the investigatormustdoandhencedetermine and demarcate the ldquomicro-areasrdquo inside thestudy area such as favored animal resting sites andanimal paths (For example the areas b9 and e9 Table1) These micro-areas can have questing nymph den-sities three or four times greater than the rest of thearea which could explain the signiTHORNcant variationfound in previous studies

The second step is to establish a minimum numberof samples that are required to give a representativesample of the study area In our case having eluci-dated that the spatial distribution of nymphs was closeto random we were able to reduce the number ofsubsamples required even during times of low abun-dance There was no signiTHORNcant difference betweensample sizes of 5 10 15 or n (total subsample size ina sample) in thedensityofnymphscollectedwhateverthe area time of year or tick abundance The methodis therefore simple and generally applicable differingsubstantially from the methodology of Daniel et al(1986) However it is advisable to survey the mini-mum surface area required regularly because verte-brate hosts may change their habits and therefore thedistribution of ticks

No signiTHORNcant difference in tick number was foundbetween a sample taken at random or transect Thiswas expected having found that the spatial distribu-tion of nymphs was random and conTHORNrms that ourmethodology conforms to the criterion of sampling at

Fig 1 ConTHORNdence intervals of the mean number of nymphs collected (range 0ETH30) per 10 m2 subsample for 23 samplesperformed from April 1997 to May 1998 in the Rambouillet forest

338 JOURNAL OF MEDICAL ENTOMOLOGY Vol 37 no 3

random Grouping samples according to their conTHORN-dence intervals enables rapid visualization of the dataand determination of variation in tick densities be-tween sites Parallel to these fundamental aspects ofmethodology we have been able to establish the op-timal nature of the cloth-lure as well as the speed ofpassage and the position of the investigator althoughthe position does not seem to make a difference

This methodology conforms to the three indispens-able criteria for correct sampling representativityrandomness and nonselectivity (because we are onlyworking with questing nymphs) Such a methodologyprovides a valid estimate of relative tick density and sowill enable the comparison of results collected inother sites

Acknowledgments

We thank J P Widmer (OfTHORNce National des Forets) forgrantingus free access to theRambouillet forest andRichardPaul for reading the manuscript This work was supported bya grant from the Fondation Merieux (Lyon France) to MV

References Cited

Boiret F J B Bouzille J M Gehu and M Godeau 1988Phytosociologie paysagere du systeme pelouses-landesdes falaises des otildeles ouest et sud armoricaines ColloquesPhytosociologiques Versailles

Daniel M V Cerny and EIA Korenberg 1986 Contri-bution to the methods of estimating absolute tick num-bers Folia Parasitol 33 371ETH379

Davidson W R D A Siefken and L H Creekmore 1994Inszliguence of annual abundance of Amblyomma america-num in central Georgia J Med Vet 31 67ETH71

Falco R C and D A Fish 1992 Comparison of methodsfor sampling the deer tick Ixodes dammini in a Lymedisease endemic area Exp Appl Acarol 14 165ETH173

Gray J S 1985 A carbon dioxide trap for prolonged sam-pling of Ixodes ricinus L populations Exp Appl Acarol1 35ETH44

Gray J S 1991 The development and seasonal activity ofthe tick Ixodes ricinus a vector of Lyme borreliosis RevMed Vet Entomol 79 323ETH333

Lane R S 1996 Risk of human exposure to vector ticks inheavily used recreational area in Northern CaliforniaAm J Trop Med Hyg 55 165ETH173

MacLeod J 1932 The bionomics of Ixodes ricinus thesheep tick of Scotland Parasitology 24 382ETH400

Perez-Eid C B Pichon E Zhioua N Tremel R VilleretD Deruaz L Mousson M Vassallo and E Ferquel1998 Lymeborreliosis emergent disease linkedwith theenvironment Bull Acad Natl Med 182 267ETH83

Schulze T L and R A Jordan 1996 Seasonal and longterm variations in abundance of adult Ixodes scapularis indifferent coastal plainhabitats ofNew-Jersey JMedVet33 963ETH970

Solberg V B K Neidhardt M R Sardelis C HildebrandtF J Hoffmann and L R Boobar 1992 Quantitativeevaluation of sampling methods for Ixodes dammini andAmblyomma americanum (Acari Ixodidae) J Med Vet29 451ETH456

Southwood TRE 1966 Ecological methods with particu-lar reference to the study of insect populationsMethuenLondon

Zhioua E D Postic F Rodhain and C Perez-Eid 1996Infection of Ixodes ricinus (Acari Ixodidae) by Borreliaburgdorferi in Ile de France J Med Entomol 33 694ETH697

Received for publication 17 August 1999 accepted 3 Decem-ber 1999

May 2000 VASSALLO ET AL SAMPLING METHODOLOGY FOR I ricinus 339

samples (on samples of n subsamples on the one handand THORNve subsamples on the other) suggest the exis-tence of signiTHORNcant differences in nymphal densitybetween samples (KruskallETHWallis P 00001 in bothcases) To avoid testing samples pairwise we directlycompared the 90 conTHORNdence intervals (CI) for allsamples The conTHORNdence intervalswere established bybootstrap on the estimated mean (resampling withreplacement 500 times) (Fig 1)

Most samples divide cleanly into groups those bor-derline collections were tested against their adjacentgroups to assess whether they qualiTHORNed as a separategroup or belonged to a neighboring group Five dis-tinct groups were classiTHORNed according to their densityindex 0 90 CI 116 group 1 1455 90 CI 441 group 2 45 90 CI 738 group 3 913 90CI 1756 group 4 1452 90 CI 2806 group 5

Discussion

For studies of vector densities and from that theprevalence of infection an objective sampling meth-odology is essential The THORNrst step is to demarcatehomogeneous botanical areas because the vegetationreszligects an ensemble of factors of the biotope thatdetermine to a large extent the density of exophilicticks such as I ricinus ClassiTHORNcation of the vegetationonly by large scale grouping (phytosociology) is in-sufTHORNcient and a THORNner scale classiTHORNcation not only byspecies present but also by their dimensions (breadthand height) is necessary We used the method of

landscape diagnostics (Boiret et al 1988) Howeverthis method does not take into account the inszliguenceof tickhostswhich the investigatormustdoandhencedetermine and demarcate the ldquomicro-areasrdquo inside thestudy area such as favored animal resting sites andanimal paths (For example the areas b9 and e9 Table1) These micro-areas can have questing nymph den-sities three or four times greater than the rest of thearea which could explain the signiTHORNcant variationfound in previous studies

The second step is to establish a minimum numberof samples that are required to give a representativesample of the study area In our case having eluci-dated that the spatial distribution of nymphs was closeto random we were able to reduce the number ofsubsamples required even during times of low abun-dance There was no signiTHORNcant difference betweensample sizes of 5 10 15 or n (total subsample size ina sample) in thedensityofnymphscollectedwhateverthe area time of year or tick abundance The methodis therefore simple and generally applicable differingsubstantially from the methodology of Daniel et al(1986) However it is advisable to survey the mini-mum surface area required regularly because verte-brate hosts may change their habits and therefore thedistribution of ticks

No signiTHORNcant difference in tick number was foundbetween a sample taken at random or transect Thiswas expected having found that the spatial distribu-tion of nymphs was random and conTHORNrms that ourmethodology conforms to the criterion of sampling at

Fig 1 ConTHORNdence intervals of the mean number of nymphs collected (range 0ETH30) per 10 m2 subsample for 23 samplesperformed from April 1997 to May 1998 in the Rambouillet forest

338 JOURNAL OF MEDICAL ENTOMOLOGY Vol 37 no 3

random Grouping samples according to their conTHORN-dence intervals enables rapid visualization of the dataand determination of variation in tick densities be-tween sites Parallel to these fundamental aspects ofmethodology we have been able to establish the op-timal nature of the cloth-lure as well as the speed ofpassage and the position of the investigator althoughthe position does not seem to make a difference

This methodology conforms to the three indispens-able criteria for correct sampling representativityrandomness and nonselectivity (because we are onlyworking with questing nymphs) Such a methodologyprovides a valid estimate of relative tick density and sowill enable the comparison of results collected inother sites

Acknowledgments

We thank J P Widmer (OfTHORNce National des Forets) forgrantingus free access to theRambouillet forest andRichardPaul for reading the manuscript This work was supported bya grant from the Fondation Merieux (Lyon France) to MV

References Cited

Boiret F J B Bouzille J M Gehu and M Godeau 1988Phytosociologie paysagere du systeme pelouses-landesdes falaises des otildeles ouest et sud armoricaines ColloquesPhytosociologiques Versailles

Daniel M V Cerny and EIA Korenberg 1986 Contri-bution to the methods of estimating absolute tick num-bers Folia Parasitol 33 371ETH379

Davidson W R D A Siefken and L H Creekmore 1994Inszliguence of annual abundance of Amblyomma america-num in central Georgia J Med Vet 31 67ETH71

Falco R C and D A Fish 1992 Comparison of methodsfor sampling the deer tick Ixodes dammini in a Lymedisease endemic area Exp Appl Acarol 14 165ETH173

Gray J S 1985 A carbon dioxide trap for prolonged sam-pling of Ixodes ricinus L populations Exp Appl Acarol1 35ETH44

Gray J S 1991 The development and seasonal activity ofthe tick Ixodes ricinus a vector of Lyme borreliosis RevMed Vet Entomol 79 323ETH333

Lane R S 1996 Risk of human exposure to vector ticks inheavily used recreational area in Northern CaliforniaAm J Trop Med Hyg 55 165ETH173

MacLeod J 1932 The bionomics of Ixodes ricinus thesheep tick of Scotland Parasitology 24 382ETH400

Perez-Eid C B Pichon E Zhioua N Tremel R VilleretD Deruaz L Mousson M Vassallo and E Ferquel1998 Lymeborreliosis emergent disease linkedwith theenvironment Bull Acad Natl Med 182 267ETH83

Schulze T L and R A Jordan 1996 Seasonal and longterm variations in abundance of adult Ixodes scapularis indifferent coastal plainhabitats ofNew-Jersey JMedVet33 963ETH970

Solberg V B K Neidhardt M R Sardelis C HildebrandtF J Hoffmann and L R Boobar 1992 Quantitativeevaluation of sampling methods for Ixodes dammini andAmblyomma americanum (Acari Ixodidae) J Med Vet29 451ETH456

Southwood TRE 1966 Ecological methods with particu-lar reference to the study of insect populationsMethuenLondon

Zhioua E D Postic F Rodhain and C Perez-Eid 1996Infection of Ixodes ricinus (Acari Ixodidae) by Borreliaburgdorferi in Ile de France J Med Entomol 33 694ETH697

Received for publication 17 August 1999 accepted 3 Decem-ber 1999

May 2000 VASSALLO ET AL SAMPLING METHODOLOGY FOR I ricinus 339

random Grouping samples according to their conTHORN-dence intervals enables rapid visualization of the dataand determination of variation in tick densities be-tween sites Parallel to these fundamental aspects ofmethodology we have been able to establish the op-timal nature of the cloth-lure as well as the speed ofpassage and the position of the investigator althoughthe position does not seem to make a difference

This methodology conforms to the three indispens-able criteria for correct sampling representativityrandomness and nonselectivity (because we are onlyworking with questing nymphs) Such a methodologyprovides a valid estimate of relative tick density and sowill enable the comparison of results collected inother sites

Acknowledgments

We thank J P Widmer (OfTHORNce National des Forets) forgrantingus free access to theRambouillet forest andRichardPaul for reading the manuscript This work was supported bya grant from the Fondation Merieux (Lyon France) to MV

References Cited

Boiret F J B Bouzille J M Gehu and M Godeau 1988Phytosociologie paysagere du systeme pelouses-landesdes falaises des otildeles ouest et sud armoricaines ColloquesPhytosociologiques Versailles

Daniel M V Cerny and EIA Korenberg 1986 Contri-bution to the methods of estimating absolute tick num-bers Folia Parasitol 33 371ETH379

Davidson W R D A Siefken and L H Creekmore 1994Inszliguence of annual abundance of Amblyomma america-num in central Georgia J Med Vet 31 67ETH71

Falco R C and D A Fish 1992 Comparison of methodsfor sampling the deer tick Ixodes dammini in a Lymedisease endemic area Exp Appl Acarol 14 165ETH173

Gray J S 1985 A carbon dioxide trap for prolonged sam-pling of Ixodes ricinus L populations Exp Appl Acarol1 35ETH44

Gray J S 1991 The development and seasonal activity ofthe tick Ixodes ricinus a vector of Lyme borreliosis RevMed Vet Entomol 79 323ETH333

Lane R S 1996 Risk of human exposure to vector ticks inheavily used recreational area in Northern CaliforniaAm J Trop Med Hyg 55 165ETH173

MacLeod J 1932 The bionomics of Ixodes ricinus thesheep tick of Scotland Parasitology 24 382ETH400

Perez-Eid C B Pichon E Zhioua N Tremel R VilleretD Deruaz L Mousson M Vassallo and E Ferquel1998 Lymeborreliosis emergent disease linkedwith theenvironment Bull Acad Natl Med 182 267ETH83

Schulze T L and R A Jordan 1996 Seasonal and longterm variations in abundance of adult Ixodes scapularis indifferent coastal plainhabitats ofNew-Jersey JMedVet33 963ETH970

Solberg V B K Neidhardt M R Sardelis C HildebrandtF J Hoffmann and L R Boobar 1992 Quantitativeevaluation of sampling methods for Ixodes dammini andAmblyomma americanum (Acari Ixodidae) J Med Vet29 451ETH456

Southwood TRE 1966 Ecological methods with particu-lar reference to the study of insect populationsMethuenLondon

Zhioua E D Postic F Rodhain and C Perez-Eid 1996Infection of Ixodes ricinus (Acari Ixodidae) by Borreliaburgdorferi in Ile de France J Med Entomol 33 694ETH697

Received for publication 17 August 1999 accepted 3 Decem-ber 1999

May 2000 VASSALLO ET AL SAMPLING METHODOLOGY FOR I ricinus 339