Ixodes ricinus were collected from four geographically distinct locations: northeast Scotland, North Wales, South England, and central France (Table 1). Sites were chosen to represent a cline of thermal climates (Fig. 2); we used a latitudinal gradient to achieve this (Fig. 3). Within the latitude remit, sites were chosen primarily on the basis that we had prior knowledge that they all had high tick densities that enabled the collection of hundreds of ticks within a few hours. This was essential to minimize the potentially confounding factor of different tick storage periods both between and within sites. However, this meant that some of the sites differed in habitat. Different habitats can create different micro-climates, for example, tall and dense vegetation that creates a canopy can modify the climate slightly, by buffering ticks from the extremes of heat and wind; in contrast, open habitats with little ground vegetation, such as short cropped pastures are more exposed to the wind and direct sunshine. While some of our sites differed in habitat (Table 1), all had a type of vegetation that similarly created good canopy cover (trees, bracken, or heather). We are therefore confident that the ticks from each site were subject to the intended thermal climate gradient, from the coldest in northeast Scotland to the warmest climate in central France.
Table 1. Details of the four geographic locations from which ticks were collected.
|Region||Area||Latitude, longitude||Altitude (m)||Estimated yearly mean max. temp. (°C)||Vegetation type||Main large tick hosts present|
|Northeast Scotland||Aberdeenshire||56°54′N, −2°31′E||300||9.9||Calluna-dominated heath||Red deer Cervus elaphus|
|North Wales||Denbighshire||53°5′N, −3°15′E||230||12.4||Bracken-dominated rough pasture||Sheep Ovis aries|
|South England||Hampshire||50°51′N, −1°38′E||40||13.8||Mixed deciduous woodland: Quercus, Fagus, Pinus||Roe deer Capreolus capreolus, fallow deer Dama dama|
|Central France||Auvergne||45°47′N, 03°27′E||380||16.0||Mixed deciduous woodland: Quercus, Fagus, Acer||Roe deer Capreolus capreolus|
Figure 2. Map of western Europe showing the locations of each site from which ticks were collected for the experiments. From top to bottom (North to South): Aberdeenshire, northeast Scotland; Denbighshire, North Wales; Hampshire, South England; Auvergne, central France.
Download figure to PowerPoint
Figure 3. Estimated mean monthly temperatures (1971–2000) at the four geographic locations (northeast Scotland, North Wales, South England, and central France) from which ticks were sampled. Data were obtained from the UK Meteorological Office (Meteorological Office 2014) and from Meteo France via the World Weather Information Service (2014) for the nearest weather station to each location and then estimated for each tick sampling location by adjust for altitude.
Download figure to PowerPoint
To control for interstadial differences in temperature thresholds (Clark 1995; Randolph 2004) only nymphs were used in our experiments. As a tick's fat reserves diminish through the questing season, potentially affecting activity levels (Crooks and Randolph 2006; Herrmann and Gern 2012), nymphs from all sites were collected within the last 2 weeks of May 2012. However, ticks from warmer climates are likely to start questing earlier in the season (and therefore have used up slightly more fat) than cooler climate ticks. To mitigate this to some extent, ticks from central France were collected first, followed 10 days later by the English, then Welsh, and lastly the Scottish ticks (which were collected 2 weeks after the French ticks). Nymphs were collected by dragging a 1 m × 1 m piece of blanket material along the ground. Conditions during tick collection over all sites ranged from 17 to 22°C and 60–70% relative humidity (recorded using Tinytag digital data loggers placed on the ground vegetation). Nymphs were stored in plastic boxes (5 cm × 10 cm), containing damp tissue to maintain high humidity, and refrigerated at 4–6°C for 1–4 days until the experiments began.
Tick questing in response to increased temperature
Nymphs from each of the four populations were put in tubes made from nylon mesh (250 μm diameter) that were 3 cm diameter, 13 cm high and glued at the base to a 3 cm diameter Petri dish. Each tube was filled with 1.5 cm of wet sand, then 1 cm of damp moss, to provide a moist environment where ticks could rehydrate if necessary as they would in nature. Immediately after adding the ticks, the tops of the tubes were sealed. Twelve tubes, each with 30 nymphs (i.e., 360 nymphs in total per site), were used for each population, except for the North Wales population that had 10 tubes (i.e., 300 nymphs in total) due to fewer ticks collected from that site.
The mesh tubes containing ticks from the UK sites were allocated equally between two portable incubators (Memmert IPP 200; Memmert, Schwabach, Germany). In order to control for any potential variation in conditions between the two incubators, an equal proportion of tubes from each population were allocated to each incubator, and the temperature and relative humidity in each incubator was calibrated by Tinytag digital data loggers. In addition, tubes from each population were arranged alternately in sequence to control for any potential effects of position within the incubator. For the French ticks, in order to ensure identical experimental conditions to the UK ticks, the same individual incubators were used, the temperature and relative humidity were calibrated with the same Tinytag digital data loggers and the same human observers were used for all populations. Experiments on the French ticks took place in France 2 weeks prior to the UK experiments. This was necessary, first, to ensure that ticks from all populations were subject to similar conditions for a similar length of time between collection and conducting the experiment (otherwise, the French ticks would have been in storage for a week longer than the UK ticks) and, second, to avoid the biosecurity risk associated with transporting live specimens across country borders.
In nature, a tick's response to increasing temperature may be confounded by variations in saturation deficit depending on the relative humidity at the time (which depends on the location, vegetation, and recent precipitation). Therefore, to ensure that we were testing the effect of temperature per se on questing we maintained the incubators at >90% humidity (as confirmed by Tinytag digital data loggers) by placing damp tissue around the interior. This also promoted maximum tick survival (MacLeod 1935) to minimize any error in our counts due to mortality over the experimental period.
To control for possible effects of photoperiod on questing behavior (Lees and Milne 1951; Randolph 2004), a “summer-time” regime of 16-h light: 8-h dark was maintained. Incubators were initially set to 5°C for 2 h, to ensure that no ticks were questing before the start of the experiment. The temperature of the incubator was then raised to 6°C for 24 h, marking the first experimental temperature and then further increased by 1°C every 24 h. At each temperature, we counted the number of ticks questing in each tube five times, at 06:30, 09:00, 13:30, 18:00, and 22:30 h, and immediately after the count at 22:30, we increased the temperature of the incubator by 1°C. Thus, 8 h had lapsed between setting the new temperature and the next count at 06:30. By allowing 24 h and five counts for each temperature, we could gauge whether or not the proportion of ticks questing stabilized at each temperature (see Supporting Information). Experiments on the UK populations ran for 10 days, covering a temperature range of 6–15°C. Due to time restrictions, experiments on the French ticks ran for 8 days, covering 7–14°C. Therefore, there were a total of 50 observations for each tube of UK ticks over the 10-day experimental period and 40 observations for each tube of French ticks over the 8-day experimental period. We had to ensure minimum tick mortality over the experiment by keeping the experimental period as short as possible, so there was a trade-off between the period at each temperature and the number of temperatures tested. For example, if we had allowed more than 24 h for each temperature, we would either have needed a longer experimental period (and therefore higher tick mortality), or not covered the critical range of temperatures. Previous trials (unpublished data) enabled us to conclude that the best compromise was an experimental period of no more than 10 days with no less than 24 h for each temperature. A temperature range of 6–15°C was chosen to encompass as much of the natural range at the sites as possible and, importantly, to approximately reflect the temperature range likely to be experienced by I. ricinus in spring as questing increases (this study may be less relevant to the end of the questing season because behavioral diapause in late summer or autumn is thought to be driven by reduction in day length as well as temperature; Randolph et al. 2002). Figure 3 shows that our chosen temperature range of 6–15°C achieves this by encompassing the mean monthly air temperatures estimated from the end of April to the end of October for the site in northeast Scotland (the entire tick activity season for this area; Ruiz-Fons and Gilbert 2010); the end of March to mid-November for North Wales, and the start of March to the end of November, but not including July, August or the first half of September for South England (encompassing the period when tick questing increases for North Wales and South England; Randolph 2004). For central France, our chosen experimental temperature range of 7–14°C encompasses early March to the start of December, but not including June, July, August or September. Our results confirmed that our chosen temperature range covered most of the questing range of I. ricinus nymphs over all from the four populations (from 5% to 91% ticks quested between 5°C and 15°C).
In order to minimize stimulation of questing in response to the observer (e.g., CO2, vibrations and shadows), incubators were located in a quiet laboratory and their internal glass doors were kept closed during observations. Observers moved slowly and quietly during counts.
The Tinytag digital data loggers were used not only to calibrate the temperature settings on each incubator but also subsequently to log the temperature and relative humidity inside the incubators every 5 min. These data showed that humidity was maintained at >90% and the incubators took approximately 10 min to stabilize from one temperature to the next. They also confirmed that all tubes were subject to the same temperatures and relative humidity (including, crucially, the French ticks that were trialed in the same incubator but a week earlier).
At the end of each experiment, all tubes were dismantled and the numbers of live and dead ticks counted.
To estimate the local thermal climatic conditions typically experienced by the four tick populations in their natural environments, 1971–2000 long-term average climate data were obtained from the UK Meteorological Office (Meteorological Office 2014) and from Meteo France, via the World Weather Information Service (2014). Data came from the nearest weather station to each location as follows: Craibstone, 102-m above sea level (m asl), 35-km NNE of the northeast Scotland collection site; Shawbury, 72 m asl, 50 km southeast of the North Wales collection site; Everton, 16 m asl, 13 km South of the South England collection site; Clermont Ferrand, 331 m asl, 25 km E of the French collection site. The monthly and annual mean temperatures at each tick sampling site (Table 1; Fig. 3) were estimated by adjusting the data from these weather stations to account for differences in altitude by subtracting 6.4°C for every 1000-m elevation gain, which is the lapse rate under normal atmospheric conditions (Moore 1956).
It was important to confirm that the temperatures we used in the experiments coincided with the temperatures associated with ticks naturally increasing their questing at each site. Therefore, for each site, we used our results to predict which months had temperatures associated with an increase in questing from 20% to 80%, and compared this with our experimental temperatures. This also helped us assess how robust our questing season estimates were for each site.
Exploring the impact of climate change
To estimate how climate change may affect future temperatures experienced by each tick population, we adjusted the estimated 1971–2000 mean monthly temperatures (estimated as described above for each sampling location) in accordance with the IPCC temperature change projections for each region we sampled (Christensen et al. 2007). We used the IPCC-projected temperature changes 1980–1999 and 2080–2099, averaged over 21 models (Christensen et al. 2007). For northeast Scotland and North Wales, these temperature change projections are +2.5°C in winter (December, January, February) and +2°C in summer (June, July, August; Christensen et al. 2007). We therefore assumed the midpoint of +2.25°C for spring and autumn (March, April, May, September, October, and November). For South England, these temperature change projections are +2.5°C in both winter and summer (Christensen et al. 2007) and we assumed also for spring and autumn. For central France, these temperature change projections are +2.5°C in winter and +4°C in summer (Christensen et al. 2007). We therefore assumed the midpoint of +3.25°C for spring and autumn.
We then explored the estimated impact of climate change on the proportion of ticks questing by applying the results of our questing experiments to the projected changes in monthly mean temperatures at each sampling location. For this, we graphically compared the estimated proportion of I. ricinus nymphs questing for estimates of the 1971–2000 monthly mean temperatures with those questing for our estimates of the temperatures in 2080–2099 from the IPCC temperature change projections as described above (Christensen et al. 2007). For months when the temperatures tend to be colder or warmer than our experimental temperatures, our estimated proportions of nymphs questing during those months had to be estimated by extrapolation and are therefore less robust.
To account for nymphs found dead in each tube at the end of the experiment, and in the absence of information on when each tick died, we assumed a constant death rate over the course of the experiment (after Van Es et al. 1999). Tick questing activity was then calculated at each observation as a proportion, that is, the number of ticks observed questing divided by the estimated total number alive inside the tube at that time.
A generalized linear mixed model was used to analyze the proportion of ticks in each tube that were questing, using the GLIMMIX procedure in SAS version 9.1.3 (SAS Institute 2002). This procedure was chosen in order to allow for the binomial nature of the response variable (proportion of ticks questing) as well as the presence of random effects. The model included tick population (i.e., site name), temperature, temperature2, and their interactions (tick population*temperature, tick population*temperature2) as fixed effects, and random effects were “block” (representing the two incubators used simultaneously for the UK populations plus the separate trial (albeit using the same incubator) for the French population) and tube (as 50 count observations were made for each tube for each population). The quadratic term, temperature2, was included in the model because there was a clear curved relationship between temperature and the proportion of ticks questing. Type 1 tests of the fixed effects showed temperature2 explained more variation than did temperature; therefore, final models included temperature2 rather than temperature per se.
Generalized linear mixed models were then run for each temperature (6–15°C) individually with proportion of questing as the response variable, tick population as the fixed effect and block and tube entered as random effects. This enabled us to identify the temperatures at which the proportion of ticks questing varied significantly between populations. Post hoc analyses, using Tukey–Kramer comparisons of the least squares means, were performed to indicate the populations between which there was significant variation in the proportion of ticks questing at each temperature.
Although tick populations were collected from climates along a thermal climate gradient due to differences in latitude, each site differed in other ways, such as rainfall, snow cover, habitat, and hosts. For this reason, and to explore the nature of any relationship between questing and thermal climate, we reran these generalized linear mixed models, but used climate parameters as continuous variables as fixed effects in place of site name (a categorical variable). Climate parameters used were mean air temperature in May (because our results suggested that this month experienced temperatures associated with the steepest increase in nymph questing) and the number of weeks per year with a mean temperature of at least 10°C (because our results suggested that most ticks quested above 10°C in all populations). These climate parameters were entered separately and sequentially in separate models. For these models, we used questing data from 9°C, because the results of our experiments indicated this temperature to be closest to the temperature at which 50% of nymphs quested for all populations (8.1°C, 8.5°C, 9.1°C, and 10.2°C for Scottish, Welsh, English, and French nymphs, respectively).