Distribution of ticks, tick-borne pathogens and the associated local environmental factors including small mammals and livestock, in two French agricultural sites: the OSCAR database

In Europe, ticks are major vectors of both human and livestock pathogens (e.g. Lyme disease, granulocytic anaplasmosis, bovine babesiosis). Agricultural landscapes, where animal breeding is a major activity, constitute a mosaic of habitat types of various quality for tick survival and are used at different frequencies by wild and domestic hosts across seasons. This habitat heterogeneity, in time and space, conditions the dynamics of these host-vector-pathogen systems and thus drives acarological risk (defined as the density of infected ticks). The principal objective of the OSCAR project (2011-2016) was to examine the links between this heterogeneity and acarological risk for humans and their domestic animals. Here, we present the data associated with this project.
 This paper reports a database on the distribution and densities of I. ricinus ticks - the most common tick species in French agricultural landscapes - and the prevalence of three tick-borne pathogens (Anaplasma phagocytophilum, Borrelia spp. and Babesia spp.) in two sites in north-western (“Zone Atelier Armorique”: ZA site) and south-western (“Vallées et Coteaux de Gascogne”: VG site) France. The distribution and density of ticks along a gradient of wooded habitats, as well as biotic variables, such as the presence and abundance of their principal domestic (livestock) and wild hosts (small mammals), were measured from forest cores and edges to more or less isolated hedges, all bordering meadows. Ticks, small mammals and information on local environmental conditions were collected along 90 transects in each of the two sites in spring and autumn 2012 and 2013 and in spring 2014, corresponding to the main periods of tick activity. Local environmental conditions were recorded along each tick and small mammal transect: habitat type, vegetation type and characteristics, slope and traces of livestock presence. Samples consisted of questing ticks collected on the vegetation (mainly I. ricinus nymphs), biopsies of captured small mammals and ticks fixed on small mammals. In the VG site, livestock occurrence and abundance were recorded each week along each tick transect.
 A total of 29004 questing ticks and 1230 small mammals were captured during the study across the two sites and over the five field campaigns. All questing nymphs (N = 12287) and questing adults (N = 646) were identified to species. Ticks from small mammals (N = 1359) were also identified to life stage. Questing nymphs (N = 4518 I. ricinus) and trapped small mammals (N = 908) were analysed for three pathogenic agents: A. phagocytophilum, Borrelia spp. and Babesia spp.
 In the VG site, the average prevalence in I. ricinus nymphs for A. phagocytophilum, Borrelia spp. and Babesia spp. were, respectively 1.9% [95% CI: 1.2-2.5], 2.5% [95% CI: 1.8-3.2] and 2.7% [95% CI: 2.0-3.4]. In small mammals, no A. phagocytophilum was detected, but the prevalence for Borrelia spp. was 4.2% [95% CI: 0.9-7.5]. On this site, there was no screening of small mammals for Babesia spp. In ZA site, the average prevalence in nymphs for A. phagocytophilum, Borrelia spp. and Babesia were, respectively 2.2% [95% CI: 1.6-2.7], 3.0% [95% CI: 2.3-3.6] and 3.1% [95% CI: 2.5-3.8]. In small mammals, the prevalence of A. phagocytophilum and Borrelia spp. were, respectively 6.9% [95% CI: 4.9-8.9] and 4.1% [95% CI: 2.7-5.9]. A single animal was found positive for Babesia microti at this site amongst the 597 tested.


Introduction
In agricultural landscapes, where livestock production occupies a large proportion of the surface area, pastures often adjoin different semi-natural ecosystems (forests, woods, hedges). This type of landscape mosaic implies that areas exploited by livestock are also frequently used by a diverse range of wild fauna. Many parasites and pathogens are shared amongst these animal species, even in the absence of direct contact and some may be transmitted between agricultural and semi-natural systems via common arthropod vectors. In France, ticks are major vectors for both human (e.g. Borrelia burgdorferi s.l., the agent of Lyme disease) and livestock pathogens (e.g. Anaplasma phagocytophilum, inducing granulocytic anaplasmosis or Babesia divergens, causing bovine babesiosis), with Ixodes ricinus being the most commonly-involved vector.
I. ricinus is a three-stage tick that feeds on a wide variety of vertebrate hosts (Sonenshine andRoe 2014, Bonnet et al. 2016). While larvae and nymphs may feed on a range of different-sized hosts, adult ticks require a bloodmeal from a larger host, like roe deer Capreolus capreolus or domestic animals (Ruiz-Fons et al. 2012). Host species are differently exploited by ticks and display variable susceptibilities to infection by different tick-borne infectious agents, exhibiting different levels of reservoir competence (Ostfeld et al. 2014). The abundance and diversity of different hosts thus influence the density of infected ticks (i.e. the "acarological risk") and hence the probability of contact with humans and livestock (LoGiudice et al. 2003, Boyard et al. 2007, Takumi et al. 2019.
Agricultural landscapes constitute a mosaic of habitat types that vary in quality for tick survival and host use. The habitat composition of a given plot and its connection with other habitats will determine its use by wild vertebrates and will thus shape local tick-host interactions (Estrada-Peña 2002, Li et al. 2012, Werden et al. 2014, Heylen et al. 2019. Breeding practices and particularly, the management of animal grazing in different types of pastures, will also influence exposure risk of livestock to ticks and the pathogens they carry (Richter and Matuschka 2006, Boyard et al. 2007, Gassner et al. 2008, Agoulon et al. 2012, Ruiz-Fons et al. 2012. However, agricultural mosaics are not temporally fixed and can vary both seasonally and yearly. We are also currently witnessing rapid landscape modifications due to the influence of global changes and particularly those associated with land-use (i.e. relative proportions of breeding/crop surfaces, forest or hedge fragmentation) and climate change (i.e. tick population dynamics are tightly linked to temperature and humidity regimes) (Medlock et al. 2013. The main goal of the OSCAR project (Outil de Simulation Cartographique à l'échelle du paysage Agricole du Risque acarologique / Simulation Tool for Mapping Acarological Risk in Agricultural Landscapes) was to explore the relationships between landscape structure and acarological risk. The study was carried out in two agricultural sites that are part of the International Long-Term Ecological Research (ILTER) network (Zones Ateliers network in France http://www.za-inee.org/en/node/804) and encompass the intrinsic diversity of agricultural landscape features: one LTER site -the "Zone Atelier Armorique" ("ZA site" hereafter) -in north-western France and the second in south-western France in the region of "Vallées et Coteaux de Gascogne" ("VG site" hereafter, belonging to the recently labelled "Zone Atelier PyGar"). Before conducting analyses, the initial task of the OSCAR project consisted of mapping the distribution of ticks, pathogens and the principal domestic (cattle) and wild (small mammals and roe deer) hosts, along a gradient of landscape fragmentation, from forest cores and edges to more or less isolated hedges, all bordering meadows. This paper describes the collected datasets ( Fig. 1) (1) on questing tick and small mammal densities, (2) on local environmental conditions (habitat, vegetation and livestock densities) of sampled transects and (3) on pathogen prevalence in ticks and small mammals. Due to time and manpower constraints, we restricted our assessment of tick host species to small mammals, livestock and roe deer, the principal reservoir hosts implicated in disease for production animals. Additional datasets used in some analyses, such as roe deer presence, were not collected in the framework of this study ( Fig. 1), but are available elsewhere as outlined in the text.  Type of collected data used to study the relationships between landscape structure and acarological risk (i.e. density of infected ticks). Dataset origins: in bold, datasets presented in the datapaper; (*) collected in the field or analysed in the laboratory; (+) calculated from field data; (o) obtained from independent databases.  (Fig. 2). The area is hilly (altitude 200-400 m above sea level) and dissected by north-south valleys with a mild oceanic climate and summer droughts. Woodland covers 24% of the area with two main forest patches of about 500 and 700 ha, many woods smaller than 50 ha and hedges dominated by Quercus spp. Areas dedicated to cultivated crops cover 32% of the main study site. Meadows cover another 40%, amongst which half are grazed by domestic animals (mostly cattle, horses, sheep, but sometimes goats and pigs), either individually or in mixed groups. The roe deer density has been estimated at around 6 roe deer/km in open areas and more than 30 roe deer/km in one of the forest areas (Hewison et al. 2007). Map of the two studied sites in France: the "Vallées et Coteaux de Gascogne" LTER site (VG) and the "Zone Atelier Armorique" LTER site (ZA). Landscape types: LH, Agricultural landscapes with a Low Hedgerow network density; HH, Agricultural landscapes with a High H edgerow network density; FE, Forest Edge; FC, Forest Core. A single label per landscape type was drawn on the map (LH, HH, FE, FC), but corresponds to several sampling points in the field. For example, for the FE label, 20 sampling points were designated around the forest (see Fig. 3 for the number of points).

Study extent:
The study was performed in the two LTER sites (ZA and VG) from 2012 to 2014. Questing ticks and small mammals were sampled during five field campaigns: spring and autumn 2012, spring and autumn 2013 and spring 2014. The sampling design is presented in Fig. 3.
The sampling zones (n = 60) were located in 4 landscape types: Agricultural landscapes with a Low Hedgerow network density (LH); Agricultural landscapes with a High Hedgerow network density (HH); Forest Edge (FE); and Forest Core (FC) (Fig. 3). Small mammals were sampled in 24 zones (amongst the 60 sampling zones), trap-lines being systematically paired with one or two questing tick transect-lines (Fig. 3). Small mammal trap-lines were distributed amongst the four landscape types as follows: six in LH, six in HH, six in FE and six in FC. For each trap-line, 34 traps were spaced 3 m apart along the 100 m line.
Questing ticks were sampled in all 60 zones (including the 24 zones for small mammal sampling). In each zone, one or two transect-lines were defined: 1) a single transect-line was sampled when found along hedgerows and in FC; 2) two transect-lines were run when situated at wood and forest edges (i.e. on either side of the ecotone: one in the meadow and one in the forest) (Fig. 3). This resulted in a total of 90 questing tick transect-lines which were distributed as follows: 30 in LH, 30 in HH, 20 in FE and 10 in FC. For each transect-line, ticks were collected along lines of 300 m, divided into 10 sub-transects of 10 m each (10 m length x 1 m width), with a space of 20 m between sub-transects (Fig. 3).
The design was fully applied (60 sampling zones) in four campaigns (spring and autumn 2012, spring 2013 and 2014), but only 36 transect lines from the 24 zones used to quantify small mammal presence were sampled during autumn 2013, corresponding to an optimisation of the sampling effort during a less favourable period of tick activity.

Sampling description: Recording local environmental conditions
Georeferencing of sampling locations of ticks (Table 1) and small mammals (Table 2) was obtained in the field using a Trimble GNSS GeoExplorer XT 6000 receiver. A differential correction in post-processing made it possible to obtain decimetric precision. The points obtained were exported in a shape (shp) format and inserted into Geographic Information System (ArcGIS) software. Drawings of the sampling lines were performed on maps by the operators during sampling and were corrected with the GIS database with the help of orthophotos (BD ORTHO®, resolution 50 cm x 50 cm, IGN). During sampling, local environmental conditions were recorded for the questing tick transect-lines, the tick subtransects and the small mammal trap-lines. The following variables were recorded in the  Field description for tick sub-transect locations. c., characters. Table 2.
Field description for small mammal trap-line locations. c., characters. Table 3.
Field description of the dataset including the characteristics of the tick transect lines. c., characters.
Distribution of ticks, tick-borne pathogens and the associated local environmental ...  Table 3 Text ( Table 4.

Field
Field description of the dataset including characteristics of tick sampling in each tick sub-transect. c., characters.

Description of the sub-transect Type
ECHT_IND_VEG Vegetation index (hedge or wood): 1 = no hedge, 2 = discontinuous hedge, 3 = continuous hedge not deeper than 2 m, 4 = deeper hedge, between 2 and 5 m, 5 = hedge deeper than 5 m or wood Boolean ECHT_PARASOL Misaligned parasol above sampling: A = no branches (no parasol), F = dense branches over less than 2/3 of the sub-transect, D = dense branches over more than 2/3 of the sub-transect

Sampling of questing ticks
Questing ticks (Fig. 3) were sampled by flagging (Boyard et al. 2007). In each sub-transect, a 1x1 m white flannel cloth (or 'flag') was slowly dragged (0.5 m/s) along 9 m (explored surface of 10 m ) across the lower vegetation and leaf-litter (Agoulon et al. 2012). Ticks were counted, collected from the flag and stored in 70% ethanol for later identification (life stage and species) and detection of infectious agents using molecular analyses (Fig. 6, Table 7). Tick identifications were performed using a binocular microscope, according to Pérez-Eid (2007). 2

Sampling of small mammals
The 100 m trap-line contained 34 INRAE live-traps, fitted with dormitory boxes and baited with a mixture of seeds and fresh apple. After placement, the traps were checked in the morning 24-and 48-hours after setup (Figs 3, 7). Captured small mammals were identified to species, sexed and weighed to 0.5 g in a field laboratory (Table 6). They were euthanised by authorised experimenters in accordance with French law and dissected. A blood sample and ear and spleen biopsies were performed for the detection and characterisation of infectious agents during the first four field campaigns. Blood sampling was performed on trapped animals using the retro-orbital method (Hoff 2000). Blood pellets were separated from serum by centrifugation. Serum samples were stored at −20°C and are available for supplementary analysis upon request. Ticks from small mammals were counted immediately after being euthanised in VG, but in ZA, due to the high number of captured mammals, dead animals were frozen and ticks were collected later during dissections. All collected ticks were stored in 70% ethanol for later identification and use for molecular analyses. The animals captured in spring 2014 were not euthanised, but were released at least 500 m away from the capture site to avoid recapture and ticks were quickly collected on these individuals.   Table 7.
Field description of the dataset concerning the analyses of tick DNA for infectious agents. c.: characters Figure 6.
Molecular analyses of ticks; +ve, positive sample.

Molecular analyses
• In tick (Fig. 6) ( Table 7): Amongst the 12287 nymphs collected during the five campaigns, 4518 I. ricinus nymphs were selected at random from the two major periods of tick activity, i.e. spring campaigns of 2012 and 2013. For each tick, DNA was extracted using the ammonia-based protocol described in Schouls et al. (1999). Borrelia detection was performed using the quantitative PCR (SYBRGreen) protocol outlined in Jacquot et al. (2016). To identify the infecting Borrelia species, positive samples were re-amplified using nested PCR protocols for the

Livestock survey in VG site
Livestock abundance was measured in the VG site on the pasture adjoining each tick transect-line in 2012 and 2013 (
Field description for livestock dataset. c., characters. Heads.day refers to the number of individual animals that were counted in a pasture on a given day.

Figure 8.
Relational model for ticks: relationships between tables concerning tick sampling and analyses. Similar colour corresponds to similar data present in two tables. Key is primary key. ECHLT_*, Identifier code for tick transect-line; ECHT_*, Identifier code for tick sub-transect line.
The data presented in this dataset are detailed by campaign and by site in Table 10.  Relational model for small mammals: relationships between tables concerning small mammal sampling and analyses. Similar colour corresponds to similar data present in two tables. Key is primary key. ECHLM_*, Identifier code for small mammal trap-line; MAM_*, Identifier code for captured small mammal. Table 10.

Variables not included in the datapaper
Information on the infection rate and movement of roe deer in some of the studied habitat types were recorded at the VG site (see, for exemple, Martin et al. 2018). They are available on http://eurodeer.org/ or upon request to the CEFS.
Weather data were obtained from Météo-France weather stations close to ZA (Broualan, Rennes-St Jacques, Pontorson) and VG (Boussan, Fabas, Palaminy) sites. Additional weather data were measured near the VG site at the meteorological weather station (INRAE in SAMAN), located at the UMR DYNAFOR (INRAE-INPT) in Saint-André (F-31420) or near the ZA site at the COSTEL meteorological weather station (CNRS in COSTEL), located in the LEGT RENNES. According to the location, the weather stations were equipped with sensors to measure air and ground temperatures, air humidity, pluviometry, wind speed and direction, relative humidity, atmospheric pressure and light intensity. The data (2011)(2012)(2013)(2014) are available upon request to the corresponding author.
Additional variables were calculated to measure landscape heterogeneity around the sampling locations. These data and their production (ecotone length between wooded habitat and meadows, proportion of woodland cover, grassland cover and crops, mean distance between wooded patches, perimeter-area ratio of wooded patches, connectivity of wooded habitat patch) are presented in  and Perez et al. (2020).    The data concerning small mammal sampling are presented in the 3 following tables.

Additional information
We provide a quick description of the results in the following section. A total of 29004 questing ticks and 1230 small mammals were collected during the study at the two sites and over the five campaigns. All questing nymphal (N = 12311) and adult ticks (646) were identified to species. Ticks from small mammals (N = 1359) were also identified to the stage.

Sampled small mammals
Over the study, 335 small mammals were trapped in the VG site (Table 12) and 895 in the ZA site (Table 13). Seven different species were found in VG against five in ZA. In both sites, wood mice (Apodemus sylvaticus) were the dominant species, accounting for 75% of the captured individuals. Bank vole (Myodes glareolus) was the second most frequentlyencountered species in both sites (VG: 11% and ZA: 24%).

Local environmental conditions
In the VG site, the forest type was mainly deciduous (N = 41) with one mixed forest (including coniferous trees). In the ZA site, collections were performed in 33 deciduous forest type and eight mixed forests. Table 14 presents some results of local environmental variables collected during tick sampling. Table 12.
Small mammal species in the VG site over the 5 field campaigns Table 13.
Small mammal species in the ZA site over the 5 field campaigns.  Table 14.
Summary values of local environmental conditions for transects and sub-transects in VG and ZA sites for the 5 field campaigns (1 to 5). Description of the fields are given in Tables 3, 4. NC: Not concerned (The field makes no sense for the landscape type in question. For example, there cannot be information in a field concerning meadows when the sub-transect line is in the forest); ND: Not documented (missing data). The livestock survey was performed in the VG site: livestock occurred on 28 of the 90 questing tick transect-lines, cattle being the main species present in pastures (Table 15). Median heads.day values at pasture was 112 for the 3 seasons (min = 0, max = 1848). Caprine were present along two transect-lines, equines along three transect-lines and ovine along three transect-lines. One meadow along a transect-line (VG-BD-L002) was occupied by the four livestock species.   Table 16.

Transects and sub-transects Site
Results of A. phagocytophilum, Borrelia spp. and Babesia spp. in nymphs from field campaigns 1 to 3 and in small mammals from field campaigns 1 to 4. No Babesia-positive small mammals were found. n/N, number of positive samples/number of analysed samples; Prev, prevalence in %; 95% CI, in [], 95% Confidence Interval for prevalence. Table 17.

Identification of Borrelia species in infected nymphs.
Pathogen results in small mammals (Table 16). A. phagocytophilum was not found in VG, but showed a prevalence of 6.9% in small mammals of ZA (Chastagner et al. 2016). Small mammals were infected only by B. afzelii with respective prevalences of 4.2% and 4.1% in VG and ZA. Amongst the six small mammals infected by Borrelia in the VG site, five were A. sylvaticus and one was M. glareolus. In the ZA site, amongst the 26 infected small mammals, 14 were A. sylvaticus, 11 were M. glareolus and one Microtus subterraneus ). In the VG site, small mammals were not screened for Babesia spp. In the ZA site, one small mammal (M. glareolus, 2-ZA-CF-LM092-M3) amongst 597 tested was positive for Babesia (Jouglin et al. 2017).