The Influence of Prescribed Fire, Habitat, and Weather on Amblyomma americanum (Ixodida: Ixodidae) in West-Central Illinois, USA

The distribution of Amblyomma americanum (L.) is changing and reports of tick-borne disease transmitted by A. americanum are increasing in the USA. We used flagging to collect ticks, surveyed vegetation and collected weather data in 2015 and 2016. A. americanum dominated collections in both years (97%). Ticks did not differ among burn treatments; however, tick abundance differed between years among total, adult, and larval ticks. Habitat variables showed a weak negative correlation to total ticks in respect to: Shannon diversity index, percent bare ground, perennial cover, and coarse woody debris. Nymphal ticks showed a weak negative correlation to percent bare ground and fewer adults were collected in areas with more leaf litter and coarse woody debris. Conversely, we found larvae more often in areas with more total cover, biennials, vines, shrubs, and leaf litter, suggesting habitat is important for this life stage. We compared weather variables to tick presence and found, in 2015, temperature, precipitation, humidity, and sample period influenced tick collection and were life stage specific. In 2016, temperature, precipitation, humidity, wind, and sample period influenced tick collection and were also life stage specific. These results indicate that spring burns in an oak woodland do not reduce ticks; other variables such as habitat and weather are more influential on tick abundance or presence at different life stages.

A. americanum are known to be aggressive, generalist feeders [14] and historically have been found in higher densities in the southern and eastern USA [15,16]. The geographic range of the lone star The study was conducted in an open oak woodland barren complex at Western Illinois University's Alice L. Kibbe Field Station located in Warsaw (40.3650 • N, 91.4075 • W), in Hancock County, IL, USA. The field station consists of~90 ha owned by Western Illinois University and is adjacent to~520 ha owned by the Illinois Department of Natural Resources. These areas are comprised of multiple habitat types including, oak-hickory woodlands, early successional woodlands, oak barrens, floodplain forests, restored tallgrass prairies and hill prairies. The Illinois Natural Areas Inventory (INAI) considers the study site to be a community high in floristic quality and fauna [42]. The entire study site was last burned in 2004 (B04) and additional burns took place in spring of 2014 (B14) and 2015 (B15). Burns were considered low intensity according to Whelan [43] because most flame heights were less than 1 m and plant mortality was limited to the understory vegetative community. Ten 40 m transects were established and georeferenced within each treatment (B04, B14 and B15) and varied among slope position and aspect (Table 1).

Tick Collection
Ticks were collected during two consecutive years (9 May 2015 until 30 October 2015 and 22 April 2016 until 4 November 2016), every two weeks, when vegetation was dry between 1200 and 1800 h. We used a flagging method to collect ticks that was similar to the method used by Rulison [44]. However, instead of using a wooden dowel, we used a bamboo stem that was not held in place with clamps; it was guided through a tightly stitched pocket. At the beginning of each transect, a sealed bag containing a 1 m 2 flannel cloth was opened and threaded onto a bamboo stem. Flagging began at 0 m along one side of each transect. At 40 m, the same flag was flipped over so the reverse side faced the ground and flagging continued along the opposite side, yielding a total flagged area of 80 m 2 per transect. When sampling at each transect was complete, the cloth was folded and placed in a labeled, sealable plastic bag. Bags containing flaggings were returned to the lab and frozen at −10 • C for a minimum of 3 days to ensure tick mortality [45]. Flags were cleaned of vegetation and ticks were removed, enumerated, and identified using taxonomic keys [46,47]. DNA was extracted from each tick collected in 2015 and 2016 and is available for testing veracity of species identifications. It is stored, along with all collection data, at the Vector Biology Initiative Laboratory, Department of Biological Sciences, Western Illinois University, IL, USA.

Weather Data
Prior to tick sampling, weather data from the National Weather Service (NWS) were recorded from the nearest weather station located at Keokuk Municipal Airport, Keokuk, IA (KEOK). These data included temperature, wind speed and relative humidity. Cloud cover was observed and noted by researchers at the site during 2016; cloud cover was not recorded in 2015. We recorded and averaged total precipitation for the two weeks prior to tick collection dates and included the day of sampling using data from the National Centers for Environmental Information (NOAA) [48].

Vegetation Sampling
During late August and September of 2015 and 2016, we used the modified Daubenmire cover scale (1 = 0-5%, 2 = 5-25%, 3 = 25-50%, 4 = 50-75%, 5 = 75-95%, and 6 = 95-100%) to assign cover classes to individual ground flora species, and to measure percent bare ground, leaf litter and coarse woody debris [49]. Tallest vegetation height was recorded every 5 m and averaged for each transect. A GRS TM densitometer (Arcarta, CA, USA) was used to view canopy coverage at 16 points along the midline of each transect. During 2016, tree species found within the 4 by 40 m transect with a diameter at breast height (DBH) >5 cm were considered trees and identified to species [50].

Statistical Analyses
Prior to analyses we calculated mean midpoint cover values for vegetation variables by growth form (annual, biennial, perennial, grass, sedge, shrub, and vine) on each transect. We additionally calculated mean garlic mustard [Alliaria petiolata (M. Bieb.) Cavara and Grande] cover, Shannon diversity, Pielou's evenness, total cover, canopy cover, vegetation height, percent leaf litter, bare ground, coarse woody debris, and total basal area as continuous predictor variables. Categorical predictor variables included treatment (B04, B14, and B15), aspect, transect position (high or low slope) and year. We only conducted analyses on A. americanum because it was the most frequently collected tick and sample sizes of other species were small. For analyses of habitat variables, we summed ticks collected on each transect per year for adult and nymphal life stages, and total ticks (all life stages). Larval ticks were considered present (1) or absent (0) because larval ticks are often found in large clusters that may skew the data [31]. Weather analyses were conducted on presence/absence on each tick life stage, per transect and sample period. Quantile plots and histograms were used to assess distribution of response variables.
We fit mixed models with a negative binomial distribution using package "glmmADMB" [51] to total, adult, and nymphal ticks. Logistic models were used for larval ticks, with predictor as a fixed effect and transect as a random effect to test how treatment and habitat variables affected tick abundance. For categorical predictor variables we also evaluated a predictor-by-year interaction to assess differences among categorical variables between years. We then conducted analysis of variance (ANOVA) on each model using package "car" [52]. We used package "lsmeans" [53] to conduct post hoc tests of least-square means with Tukey adjustment for categorical treatment variables that were significant. Spearman's rank correlation was used to determine if there were linear correlations Insects 2018, 9, 36 5 of 15 between response and predictor variables. For weather analyses we used function "glm" to conduct logistic regression on total ticks and each life stage, with a binomial logit link. Weather analyses were conducted for 2015 and 2016 separately. We used Wald chi-square analysis to test for differences of response and continuous predictor variables in logistic models. T-tests were used to assess continuous weather variables between years. All statistical analyses were conducted in program R ver. 3.4.1 [54], and all graphics were produced using SigmaPlot ver. 10.0 [55].
Total abundance. For categorical predictor variables we also evaluated a predictor-by-year interaction to assess differences among categorical variables between years. We then conducted analysis of variance (ANOVA) on each model using package "car" [52]. We used package "lsmeans" [53] to conduct post hoc tests of least-square means with Tukey adjustment for categorical treatment variables that were significant. Spearman's rank correlation was used to determine if there were linear correlations between response and predictor variables. For weather analyses we used function "glm" to conduct logistic regression on total ticks and each life stage, with a binomial logit link. Weather analyses were conducted for 2015 and 2016 separately. We used Wald chi-square analysis to test for differences of response and continuous predictor variables in logistic models. T-tests were used to assess continuous weather variables between years. All statistical analyses were conducted in program R ver. 3.4.1 [54], and all graphics were produced using SigmaPlot ver. 10.0 [55].

Habitat
Total tick abundance differed with respect to Shannon-diversity index for vegetation, perennial cover, percent bare ground and coarse woody debris (Table 2). Adult tick abundance differed with respect to total vegetation cover and leaf litter cover ( differed with respect to annual cover; however, there was no linear trend. Nymphal tick abundance also differed with respect to percent bare ground (Table 2). Although not significant in ANOVA tests, Spearman rank correlation indicated significant negative linear relationships between total ticks and biennial cover, while adult ticks were negatively associated with biennial and annual cover and percent coarse woody debris (Table 2).

Habitat
Total tick abundance differed with respect to Shannon-diversity index for vegetation, perennial cover, percent bare ground and coarse woody debris ( Table 2). Adult tick abundance differed with respect to total vegetation cover and leaf litter cover (Table 2). Nymphal tick abundance differed with respect to annual cover; however, there was no linear trend. Nymphal tick abundance also differed with respect to percent bare ground (Table 2). Although not significant in ANOVA tests, Spearman rank correlation indicated significant negative linear relationships between total ticks and biennial cover, while adult ticks were negatively associated with biennial and annual cover and percent coarse woody debris (Table 2).

Treatment and Landscape
Of the three species of ticks collected, A. americanum made up the majority (97%), which supports previous research documenting their aggressive questing nature and preference for secondary growth woodland habitat [11,14]. Total A. americanum and all life stages collected did not differ among burn treatments, which suggests that the low intensity burning conducted at the site had a limited effect on the number of ticks collected. Additionally, it is possible that hosts harboring ticks could have immigrated to recently burned areas in search of mast and young vegetation for food, and consequently, ticks could have dropped off in these areas and re-established. Previous research describing the effects of fire on ticks are conflicting; some studies show an initial decrease in ticks with populations rebounding in subsequent years [27,31,[56][57][58][59]. Our study and several others have suggested that tick presence may not be significantly affected by burning [39,40,60]. In the future, seasonality of burns should be investigated to compare the effects of warm season (fall) and cool season (spring) burns on tick presence.
Interestingly, when total ticks, adults, nymphs, and larvae were compared in relation to landscape position (slope and aspect) ticks were not distributed differently, which implies that hosts had no preference regarding slope or aspect. Lack of effect with respect to slope could have been caused by low variation between transects due to spatial proximity and minimal elevation differences at the site. Steep slopes are probably less attractive to hosts and consequently may have fewer ticks [61]. We also expected some variation in tick presence among aspects, yet none were found. South-facing aspects are warmer and drier, whereas north facing aspects are cooler with more moisture. Variations in moisture and temperature also contribute to differences in vegetation communities.
When each life stage was compared between sample years, nymphs collected were similar between years, but all other life stages were significantly different between years (Figure 1). A model developed by Haile and Mount [62] for simulating environmental variables has been used to investigate why ticks vary among years. Using this model, it was determined that annual variation of ticks could be due to host-finding rate (based on an interaction of tick life stage, day length, and host density) and weather (temperature and relative humidity) [27]. With regards to our study, we examined how tick life stages varied between two years and found that life stages varied seasonally and annually. Similarly, we found that the effect of weather variables was life stage-specific and affected tick collection. Nymphs collected may have been similar in both years simply by chance.

Habitat
In a previous study, plant diversity was found to be an insignificant predictor of A. americanum adult and nymphal abundance [63]. Our results indicate that plant diversity (based on Shannon diversity index) had a weak negative correlation to total A. americanum abundance. Unmanaged areas typically have more invasive vegetation that decreases diversity, provides food and cover for hosts, and may provide a favorable microhabitat with a more constant relative humidity [20,64]. Habitats with high abundances of exotic, invasive species such as bush honeysuckle [Lonicera macckii (Rupr.) Herder], Japanese barberry [Berberis thunbergii de Candolle] and multiflora rose [Rosa multiflora Thunb.] have been shown to support more ticks [20,65,66]. Additionally, total ticks showed a weak negative correlation to perennial cover, which likely provided questing habitat and contributed to the overall diversity at the site (Table 2). Thus, it would be ecologically plausible if ticks were negatively associated with areas with high diversity.
Total A. americanum and nymphs collected were negatively associated with percent bare ground. These results suggest that more bare ground reduces tick abundance and creates an unfavorable environment for ticks. Conversely, adult abundance differed in respect to leaf litter cover with tick collection decreasing slightly as litter cover increased. Schulze [67] found that nymphal A. americanum were more tolerant than I. scapularis to dry conditions. Additionally, since adults have a smaller surface area to volume ratio than larvae, they may be able to tolerate less leaf litter [19,68]. Other studies that have examined litter cover, grass cover and litter mass have showed positive associations with tick abundance, which suggests that ground cover is important to the life cycle of A. americanum [31,39]. Total cover differed in respect to adult A. americanum; however, there was no significant linear relationship.
Larval A. americanum presence was associated with several vegetation variables which included: biennial cover, vine cover, shrub cover, leaf litter cover and total cover. As these vegetation variables increased, larval tick presence increased, suggesting that larval ticks may be affected by differences in vegetation to a greater degree than other life stages, perhaps due to their surface area to volume ratio, or that hosts prefer using areas with greater vegetation cover. Biennial plants invest their energy in basal leaves during the first year of growth and provide ground cover that could protect ticks from desiccation [69]. Leaf litter cover and total vegetation cover may be even more important to larval ticks due to their strong moisture requirements [33]. Shrub cover was marginally insignificant but likely provided cover for O. virginianus and shade that contributes to a mesic environment [20,70]. Vines such as Virginia creeper [Parthenocissus quinquefolia (L.) Planch], Vitis spp., poison ivy [Toxicodendron radicans (L.) Kuntze], common periwinkle, Vinca minor (L.), and Smilax spp. are ubiquitous at the study site and have been found in gut diet analyses of white-tailed deer [70]. Documenting species browsed and camera-trap surveys along transects should be included in future studies to further examine the relationship between host dynamics and tick presence.
Previously, tick sampling biases have been found between species and life stages [67,71]. Presence of coarse woody debris had a weak negative association with total ticks collected, which could have been due to the sampling method used. We used a flagging method to collect ticks and previous studies have shown that CO 2 traps collect more A. americanum [71]. It is possible that flagging over debris could be less efficient and result in the loss of ticks.

Weather
A. americanum have distinct seasonal trends in the midwestern USA. Adult ticks in the Midwest are most active from May through July, nymphs from May through August, and larvae July through September [22]. Similar seasonal tick activity patterns were observed during our study. As the sample period progressed in 2015, total, adult and nymphal collection decreased. In 2016, nymphal collection only decreased and conversely, larval tick collection increased as the sampling period progressed. Mean monthly temperatures were higher in 2016 during months larval ticks were collected, which would have caused soil temperatures to stay warmer. Increases in soil temperature have been shown to influence the timing of larval A. americanum hatch [72].
Studies have shown that A. americanum prefer a relative humidity of 85% or greater for reproduction and molting success [73,74]. A. americanum are more dependent on humidity than D. variabilis [75]. We expected ticks to quest when humidity was high because moisture from the atmosphere would be readily available. As relative humidity increased during 2015, more total, adult and nymphal ticks were collected; conversely during 2016, fewer adult and larval ticks were collected as relative humidity increased. These results were inconsistent between years and could have been influenced by temporal differences in precipitation. Precipitation and moisture have previously been shown to highly influence tick activity [61,76]. All adult and nymphal ticks were collected in May, June and early July of both years and total precipitation during these months in 2015 was over double that during these same months in 2016 (50.75 cm and 18.75 cm, respectively), based on data from the National Centers for Environmental Information (NOAA) [48]. Collections during 2015 yielded more adult and nymphs than in 2016 and higher precipitation was a likely contributor. In future studies, it would be beneficial to have weather data collected at each transect, due to the high sensitivity of ticks to microclimate.
Other weather variables such as wind and cloud cover were significant for nymphal and larval stages only during 2016-however, confidence intervals overlapped zero and trends were not consistent between years, making them less reliable predictors of tick collection. Of these other significant variables, wind appeared to be the only reliable predictor for larval tick collection. Results showed that an increase in wind decreased the likelihood of collecting larval ticks, which is not surprising considering their small size and tendency to lose moisture quickly. Larval ticks are more likely to be negatively affected by increased wind speed due to their greater susceptibility of desiccation [32].

Conclusions
With tick-borne diseases on the rise in Illinois and worldwide, our results are important because we were able to document additional records for vector ticks in Illinois. These data serve as a baseline for the effects of low intensity burns, habitat, and weather patterns on tick presence in a midwestern USA oak woodland community. In areas where vector tick collection is higher, disease risk for those who are active in outdoor environments is likely greater. Although significant habitat variables were found, most of these were weakly correlated to A. americanum presence or abundance and several were inconsistent between years. Conversely, the odds of encountering larvae with respect to significant habitat variables were stronger. Thus, it is vital to continue long-term monitoring of factors that may contribute to tick presence and abundance which include both spatial and temporal trends, particularly in light of future effects of climate change on these factors.