The variations in tick activity across space and time can be classified into 4 levels: 1) Baseline abundance variation; 2) Sub-daily variation; 3) Seasonal variation; 4) Inter-annual variation. The baseline abundance reflects the suitability/carrying capacity of the habitats for the populations, both land cover and climatic conditions18. The sub-daily variation arises from the immediate responses of questing ticks to the environment, such as temperature, relative humidity, or photoperiod48, reflecting the instantaneous probability of questing. While seasonal and inter-annual variation reflects the long-term cumulative results of the questing behaviour as well as reproduction, development, and mortality rates on the tick population3. However, previous studies on the effects of the environment on I. ricinus nymph activities in Europe did not capture all the aforementioned levels of variation. They largely focused on either a short-term period for multiple sites19–22 or a long-term period for a few areas13–16. The present study attempted to explore the baseline abundance, seasonal, and inter-annual variations of I. ricinus nymph activity by expanding the observation across a wide range of geographical/climatic areas over a long period, involving a total of 11 sampling sites from 7 tick observatories in France over 8 years. With the repeated removal sampling design, we controlled the uncertainty on the proxy measure of tick abundance (nymph counts per 100 m2) that arises from the variability of sampling rate across different sites and times24,49. Furthermore, all transects were covered with low vegetations, which were shown to maximize sampling rates compared to other types of vegetative floors24, resulting in higher accuracy in estimating the proxy of tick abundance. To avoid the uncertainty caused by the variations of the environment within a sampling site, we repeated tick sampling from the same transects throughout the observation period. Furthermore, we strived to minimize the interference of tick populations and their environment during the long-term observation period by using a one-month sampling interval. As all of the transects were open areas, ticks were free to relocate in and out of the transects, either directly or through hosts, and compensate for the monthly removals of tick populations. Nevertheless, the sub-daily variations in questing behaviour or immediate response to meteorological conditions among nymphs from different regions were not captured by this study design.
We observed the phenological patterns of nymph activity tended to follow the characteristics of meteorological conditions: 1) Pattern 1, a unimodal pattern with a broad summer peak and a winter pause in colder climates (Saint-Genès-Champanelle, and Velaine-en-Haye); 2) Pattern 2, a bimodal pattern with spring and autumn peaks in intermediate climates with dry summer (Les Bordes A to C, and Etiolles); 3) Pattern 3, a unimodal pattern with an early spring peak without a winter pause in warm and humid climates (Carquefou and Gardouch). However, despite the phenological patterns 1 and 3 being clearly distinct, we found that the existence of the autumn peaks in pattern 2 was ambiguous for the following reasons: 1) Most sites with phenological pattern 2 had a low baseline abundance level. One additional nymph count due to variability of the sampling could lead to a significant false signal for the autumn activity; 2) Compared to other patterns, tick activity at most sites with phenological pattern 2 was observed for a shorter period (a smaller sample size). A longer observation period is required to confirm the presence of bimodal patterns at these sites. Furthermore, we observed different phenological patterns at La Tour de Salvagny A and B, despite these two sites being approximately 2-km apart. The heterogeneity of the phenological patterns could have arisen from the following reasons: 1) The phenological pattern 2 observed from a short-term period at La Tour de Salvagny A was ambiguous, as previously discussed; 2) Different landscapes and host densities influenced tick population dynamics and apparent tick activities differently.
The overall phenology patterns observed in our study followed a climatic gradient previously reported on I. ricinus adults across northern Africa to eastern Europe, with a unimodal pattern with a summer peak and a winter pause in freezing climates (Karelia, Russia), a bimodal pattern with spring and autumn peaks in intermediate climates (European Russia's temperate zone), and a unimodal pattern with a late winter peak in warm climates (Algeria)17. Also, a similar phenology trend observed in one mountain region in Switzerland was reported to follow an elevation gradient where the spring-autumn bimodal phenology shifted to a late-spring unimodal pattern at a higher elevation, presumably due to a temperature gradient16. Therefore, it is conceivable to hypothesize that a gradually warming climate will result in the following phenological changes to I. ricinus: 1) At colder climates, the activity peaks could have shifted towards late-winter or early-spring; 2) At warmer and drier climates, the habitats would no longer support the physiological needs of I. ricinus. However, a spring-autumn bimodal phenology was also observed in a freezing climate in Finland50, where the population genetics structure was completely different from that of I. ricinus in western Europe51. This suggested that I. ricinus from different climatic regions could have adapted/evolved to questing for a host at different temperature thresholds52 and responded differently to climate change. Furthermore, our findings indicated that describing phenological patterns solely based on their shape, such as unimodal or bimodal, was insufficient. Rather, the timing of the phenology peak and the winter traits should also be addressed. As a consequence, in order to better understand the relationship between climatic gradient and phenological patterns, we recommend longitudinal studies conducted over large geographic areas to report the phenological pattern of each sampling location separately. Moreover, determining the age of nymphs through fat content analysis has been demonstrated to guide formulating hypotheses about the mechanisms for the population dynamics of each phenological pattern16.
As the computation for cumulated nymph counts using areas under the density curve15 requires strictly regular sampling intervals throughout the study period, we used annual maximum nymph counts as a proxy to explore inter-annual variations in questing nymph activity. Directional changes in the annual maximum nymph counts were observed only in Gardouch Outside but not Gardouch Inside, despite being located in the same area. We hypothesized that it could have occurred as a result of 1) A possible decrease in roe deer population outside of the experimental station, while the number of roe deer was always maintained vastly high inside the experimental station; or 2) The periods of heatwaves that impacted this area in summer 2019 and summer 2020 affected the tick survival, while the high host density in Gardouch Inside compensated the surge in tick mortality. Furthermore, a recent study suggested that the inter-annual fluctuation of nymph abundance could be explained by the level of tree seed production, and thus the abundance level of small mammal hosts, in previous years53. However, retrospective data on seed production were not available for all of our sampling sites during the study period, which precluded us from exploring this hypothesis in this study.
The regression analysis suggested that high nymph counts were non-linearly associated with different interval-average variables, including lower previous season temperature \({T}_{M}^{3:6}\), higher half-year average minimum relative humidity \({U}_{N}^{0:6}\), and a range of optimal one-month average temperature \({T}_{M}^{0:1}\). As the number of nymphs collected in each sampling was a product of total population size of nymphs at the time of sampling (true abundance), the proportion of nymphs questing for a host, and the sampling rates, we hypothesized that the meteorological conditions influenced each of these unobserved parameters at different time lags. The population size is determined by the cumulative effect of past meteorological conditions, either through mortality rate or larva-to-nymph developmental rate. While the proportion of questing nymphs tended to respond to short-term fluctuation of the weather conditions54. A longitudinal observation on one site in Germany found similar effects of interval-average variables that explain the phenological pattern and inter-annual variation of I. ricinus activity14, which were current temperature \({T}_{M}^{0:0}\), four-to-six previous months average temperature \({T}_{M}^{4:6}\), and one-month average relative humidity \({U}_{M}^{0:1}\). Furthermore, the best-fitted model also showed a significant interaction between the one-month average temperature and a bioclimatic variable (average annual temperature \({BIO}_{Temp}\)). This result indicated that although I. ricinus from colder climates start questing at lower temperature thresholds52, their activity arrives at its peak at higher temperatures (in summer; temperature ~ 14 to 16°C). While tick activity in warmer climates peaks at lower temperatures (in spring; temperature ~ 11 to 14°C). Presumably, the population size of active nymphs in cold climates is unlikely to reach its peak during a colder spring due to developmental or behavioural diapause. In addition, photoperiod has also been proposed to regulate both behavioural and developmental diapause of ixodid ticks55; therefore, the exclusion of daytime duration from the best-fitted model, which was highly correlated with several meteorological variables, does not rule out the effects of photoperiod on tick activity dynamics.
The regression model also suggested that the baseline abundance level was associated with land cover characteristics. Sampling sites with a transition between moderately fragmented forest and non-forest areas, such as grassland or agricultural areas, support higher baseline nymph abundance (La Tour de Salvagny B, Gardouch Inside, Gardouch Outside, and Carquefou; \({p}_{Forest}\) ~ 17 to 52%). While landscapes with a high proportion of forest covers (\({p}_{Forest}\) > 90%) tended to have intermediate (Velaine-en-Haye and Etiolles) or low (Les Bordes A to C, and Saint-Genès-Champanelle) baseline abundance level. These findings are consistent with previous studies56−58, indicating that forest fragmentation created transitional forest areas known as ecotones, which attract several mammal hosts21 and lead to a higher tick abundance. In addition, it has been suggested that forest fragmentation increases not only tick abundance but also the prevalence of pathogen-infected ticks61 and the risks of human-tick exposure62, which may lead to a higher incidence of tick-borne diseases. We also observed markedly low baseline I. ricinus abundance in the highly fragmented forest with a low proportion of forest-covering area (La Tour de Salvagny A; \({p}_{Forest}\) = 6.5%), possibly due to the lack of large foraging hosts in such a landscape63. Furthermore, we found a remarkably low baseline abundance in all three sampling sites in Les Bordes, which could not be fully explained biologically by our available environmental factors. This could be explained by the unsuitable conditions of Les Bordes forest, where superimposed layers of clay and sand lead to a temporary perched water table in winter, and the low soil water storage capacity leads to arid soil conditions in the upper parts of the soil profile during summer months with low rainfall. Other additional factors could also contribute to explain the low tick abundance at Les Bordes site such as host abundance, competition, or predation of a forest with such high biodiversity.
Desirable model diagnostic results and repeated resampling models using half of the original data produced similar estimate values over 500 iterations, indicating that all explanatory variables in the best-fitted model are robust. Model predictions also successfully predicted baseline abundance, phenology patterns, and inter-annual variations of I. ricinus nymph activity. However, the prediction from the one-month average model still imperfectly captured the autumn peak in the phenology Pattern 2, where the sampling sites with this pattern had both fewer observations and lower abundance. This has brought us to the following inconclusive explanations: 1) The model failed to replicate the apparent autumn activity peak more clearly; 2) The apparent autumn peak was observed at random, and one additional nymph collected could generate a false signal when the sample size and baseline abundance were both small. As a result, it is envisioned that a more extended observation period and more sampling sites encompassing all phenological patterns should be implemented. Finally, although the regression analysis has been primarily used to find robust associations of environmental factors and help formulate ecological hypotheses on tick activity, it is considered a phenomenological model that does not fully explain the mechanisms of the observed phenomena3. In light of growing concerns about the impact of climate change on tick-borne diseases, particularly Lyme borreliosis, the mechanistic modelling framework for I. ricinus population dynamics is still required to investigate these ecological hypotheses.