Effects of rodent abundance on ticks and Borrelia: results from an experimental and observational study in an island system

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ Effects of rodent abundance on ticks and Borrelia: results from an experimental and observational study in an island system © The Author(s) 2024 Published version Kiran, Nosheen; Brila, Ilze; Mappes, Tapio; Sipari, Saana; Wang, Yingying; Welsh, Erin; Kallio, Eva R. Kiran, N., Brila, I., Mappes, T., Sipari, S., Wang, Y., Welsh, E., & Kallio, E. R. (2024). Effects of rodent abundance on ticks and Borrelia: results from an experimental and observational study in an island system. Parasites and Vectors, 17, Article 157. https://doi.org/10.1186/s13071-024- 06130-x 2024 Kiranetal. Parasites & Vectors (2024) 17:157 https://doi.org/10.1186/s13071-024-06130-x RESEARCH Open Access © The Author(s) 2024. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Parasites & Vectors Effects ofrodent abundance onticks andBorrelia: results fromanexperimental andobservational study inanisland system Nosheen Kiran1*, Ilze Brila1,2 , Tapio Mappes1 , Saana Sipari1 , Yingying Wang1 , Erin Welsh1 and Eva R. Kallio1 Abstract Background Lyme borreliosis is the most common tick-borne disease in Europe and is often caused by Borrelia afzelii, which is transmitted by Ixodes ricinus ticks. The prevalence and abundance of infected ticks fluctuate in time and space, influencing human infection risk. Rodents are reservoir hosts for B. afzelii and important feeding hosts for larval ticks. In the study reported here, we examined how variation in rodent abundance is associated with B. afzelii infection prevalence in ticks, the density of nymphs (DON) and the density of infected nymphs (DIN) in the following year. We further analysed the relationships between the abundance of infected rodents and nymphal infection prevalence (NIP) and DIN. Methods We conducted a study that combined experimental and observational approaches on 15 islands (10 small islands and 5 large islands) in Finland. On all of the islands, ticks and rodents were monitored and sampled during the summer of 2019, with the monitoring of tick abundance and sampling continuing into the spring of 2020. On five of the 10 small islands, captured rodents were removed from the island (“removal” islands), and on the other five small islands, captured rodents were released back to the trapping site after marking and sampling (“control” islands). On the five large islands, captured rodents were released back to the trapping site after marking and sampling. The presence of B. afzelii from nymph and rodent samples was examined. Results The results of the experimental study showed that neither treatment (removal), rodent abundance index nor abundance index of infected rodents in 2019 was associated with DON, NIP or DIN in 2020. Based on data from the observational study, the NIP in 2020 decreased with increasing rodent abundance index and abundance index of infected rodents in 2019. However, the DIN in 2020 was not associated with the rodent abundance index or the abundance index of infected rodents in 2019. In addition, in the observational study, DON in 2020 increased with increasing rodent abundance index. Conclusions Our results suggest that low rodent abundance during the tick activity period is not sufficient for reducing the disease hazard and, hence, rodent removal may not be a feasible control measure in natural ecosystems. Keywords Ixodes ricinus, Borrelia burgdorferi, Borrelia afzelii, Nymphs, Infection prevalence, Host abundance *Correspondence: Nosheen Kiran [email protected] Full list of author information is available at the end of the article Page 2 of 12 Kiranetal. Parasites & Vectors (2024) 17:157 Background Tick-borne diseases are a source of serious health concerns in Europe [1–3]. The most common tick-borne disease in the Northern Hemisphere is Lyme borreliosis (LB) [4, 5]. LB is caused by spirochete bacteria belonging to the Borrelia burgdorferi sensu lato (B. burgdorferi s.l.) complex that are transmitted among vertebrate hosts byixodid ticks [6]. Ixodes ricinus is the main vector of B. burgdorferi s.l. in Europe [7]. The life stages of this tick differ in terms of host preferences, with rodents being common hosts for larvae, nymphs parasitizing a wider range of hosts and adults preferring large hosts, such as deer [8–10]. The nymphal stage is considered to be the life stage posing the highest risk to humans due to its small size of nymphs, their potentially high abundance and the potential for nymphs to carry pathogens [11]. Therefore, the local abundance of infected nymphs is considered to be a primary environmental risk for humans [12, 13]. Rodents not only provide a blood-meal source for immature I. ricinus ticks but also contribute to the propagation of horizontally transmitted tick-borne pathogens; uninfected larval ticks may acquire pathogens from infected reservoir rodents that acquire the infection from infected nymphs [14–16]. In Europe, one of the most common etiological agents of Lyme borreliosis is Borrelia afzelii (a genospecies of B. burgdorferi s.l.), which is maintained by rodent hosts such as mice and voles [15, 17–24]. The role of rodent abundance in the prevalence of B. burgdorferi s.l. infection in nymphs has been examined in several studies carried out in North America [25–27] and Central Europe [28, 29], with the results suggesting that a higher abundance of reservoir rodents translates into a higher infection prevalence in nymphs the following year. However, the Lyme disease system in North America and Central Europe differs from the system in Northern Fennoscandia in many ways. For example, occasional masting increases rodent densities in temperate Europe, whereas in northern Fennoscandia, vole-dominated rodent populations show seasonal and multiannual density fluctuations characterized by very low densities over extended periods (over the breeding season) [30, 31]. It remains to be quantified whether these very low rodent densities affect the dynamics of rodent-associated tick-borne pathogens and infection prevalence in ticks, as is expected because the rodent density fluctuations have been shown to translate into human Lyme disease incidence with time delays in northern Europe [32, 33]. Most studies addressing the role of rodent density in determining the prevalence of B. burgdorferi s.l. in ticks are observational or theoretical in design [26, 27, 33, 34]. Only recently has an experimental study, conducted in small outdoor enclosures in temperate Europe [29], shown that an increase in rodent density increases the prevalence of rodent-associated tick-borne pathogens in ticks the following year. Our study aimed to quantify whether rodent density variations, especially low rodent abundance, are associated with changes in B. afzelii infection prevalence in ticks the following year. In our study, we utilized a natural island system and incorporated both experimental and observational study designs. We investigated the associations between rodent abundance and the abundance of infected rodents in year one (t) and the density of nymphs (DON), the prevalence of B. afzelii in nymphs (referred to as nymph infection prevalence [NIP]) and the density of infected nymphs (DIN) in the subsequent year (t+1). We hypothesized that: (i) rodent removal decreases NIP the following year (NIPt+1) and (ii) the higher the rodent abundance index or abundance index of infected rodents, the higher NIPt+1. We also predicted that: (iii) a higher rodent abundance index will also result in a higher density of nymphs in the following year (DONt+1) because rodents are considered to be important feeding hosts of larvae. Consequently, we expect that: (iv) the rodent abundance index would increase the DIN the following year (DINt+1), which in turn determines the level of B. afzelii hazard to humans. Methods Study area anddesign The study was conducted on 15 islands in the Porvoo archipelago in southern Finland, of which 10 islands were small (0.5–4ha) and five were large (63–966ha) (Fig.1). An experimental study was conducted on the 10 small islands which consisted of: (i) the removal of all captured rodents from five of the islands (treatment: removal; designated “removal” islands); and (ii) the release of all captured rodents back to the trapping site (approximately) after sampling and marking on the remaining five small islands (designated “control” islands). Each pair of small islands was randomly allocated to removal and control treatments before the study. The observational study was carried out on the five small “control” islands and the five large islands where the rodents were released after the sampling and marking. Tick collections were also carried out on all islands during the same time period and also during the following year in the spring. An overview of the study design is shown in Fig.2. Island size, study area and tick collection information are available in Additional file1: TableS1 and Additional file1: TableS2. Dung surveys were also carried out at the same time as rodent trappings to estimate the abundance of cervids (Additional file2: Text S1); however, these surveys had no significant association with the study design, and the dung data are addressed only in the additional files Page 3 of 12 Kiranetal. Parasites & Vectors (2024) 17:157 Fig. 1 Map of the study sites showing rodent removal, control and large islands. Grey islands have not been included in the study Fig. 2 Study design including a collection of ticks and rodents in the experimental (yellow) and observational (green) studies to examine the associations between rodents, ticks and Borrelia afzelii. The crossover indicates that no rodent trapping was done in May 2020. The distance in meters indicates the tick collection distance on the islands Page 4 of 12 Kiranetal. Parasites & Vectors (2024) 17:157 (Additional file1: TableS3; Additional file1: TableS4) and the dung index is not included in other models (see sectionStatistical Analyses). Rodent trapping Rodents were trapped in May, June, July and late August/early September in 2019 using Ugglan live traps (Grahnab AB, Gnosjö, Sweden). The number of trapping squares (4 traps per square, with an inter-trap distance of 20m) used depended on the island size. On the small islands, we used between two and seven trapping squares (ntraps = 8–28 per island) to cover the whole island area, while on large islands we used four trapping squares (ntraps = 16 per island) placed 50–100 m apart from each other; see Additional file1: TableS1 for more details. Traps baited with sunflower seeds and potatoes were set in the late afternoon at pre-marked locations on each island and checked on two consecutive mornings. All trapping was done within approximately 1 week (May–July: 6 days; August/September: 10days). In total, we captured three rodent species: bank vole (Clethrionomys glareolus), field vole (Microtus agrestis) and yellownecked mouse (Apodemus flavicollis). We calculated the rodent abundance index per 100 traps as the sum of bank voles and field voles relative to the trapping effort on the islands (the number of traps per island multiplied by the 2 trapping nights) × 100 to control for the differences in trapping effort between islands. Bank voles and field voles were chosen because they were the most common rodent species on the islands and as their presence varied on different islands, with some islands not having any field voles or bank voles. Yellow-necked mice were not included in the analysis due to the low number of individuals trapped across all islands (n = 3). We determined the species of each rodent and recorded the individual’s characteristics and presence of ectoparasites. An ear biopsy (diameter: 2 mm), taken from one ear of each animal for B. afzelii screening, was stored in 70% ethanol at - 20°C until further processing. Animals captured on the control and large islands were marked with a microchip (Trovan Unique™; Microchips Australia, Keysborough, VIC, Australia) at the first capture and released near their trapping location after sampling. The same data and samples were taken from recaptured animals in the following trapping months (sessions). All animals captured on removal islands were euthanized using cervical dislocation after data collection and sampling. Tick collection Ticks were collected during all rodent trapping sessions and in May 2020. In Finland, tick activity is absent during the winter frost and snow periods. In Finland, depending on the region, life stage and biotope, I. ricinus is mostly active from April/May to October/November [35–38]. Both I. ricinus larvae and nymphs have been found to be abundant in vegetation in May–June and in August–September in southwest and central Finland [35, 38]. Thus, most of the activity period of the larvae and nymphs was covered by the study period. However, this was the first study to observe the activity dynamics of I. ricinus on this island system. Ticks were collected by dragging a 1-m2 cotton flannel flag through the vegetation. After every 10m, the flag was checked for ticks, and all ticks on the flag were recorded, removed, placed in 70% ethanol and stored at −20°C until further processing. The dragging distance depended on the size of the island and the session. On the large islands, the dragging distance was 400m in all sampling sessions and on all islands in May 2019 and 2020. On the small islands in other sessions (July–August/September), the dragging distance was 100m to avoid the removal of too many ticks, thus potentially affecting tick abundance and tick-borne pathogen circulation (Fig.2). Details on the tick collections and infection prevalence per session per island are available in Additional file1: TableS2. All collected ticks were determined to be I. ricinus using a light microscope and morphological identification keys [39–41]. We estimated the average DON per 100 m2 per site per session as the number of nymphs collected/dragging distance × 100. Detection ofB. afzelii inticks androdents DNA from nymphs (n = 1709) was extracted using the ammonium hydroxide (NH4OH) (Sigma Aldrich, St. Louis, MO, USA) method [42] with slight modifications. Each nymph was taken from the 70% ethanol in which it was stored, air dried, and placed in a 2-ml safe lock microtube (Sarstedt, Numbrecht, Germany) containing 200µl of 1.25% NH4OH solution. Each sample was heated for 3 min at 100°C, crushed using the Qiagen TissueLyser (Qiagen, Hilden, Germany) for 2 min and then incubated at 100°C for 20min. After quick cooling, the microtubes were left open and incubated at 100°C for 15min to evaporate the ammonia (approx. 50% of the starting volume). Negative controls (containing only 1.25% NH4OH solution) were included after every five samples. Total DNA was extracted from rodent ear biopsy samples using the protocol of Laird etal. [43] with negative controls included after every five samples. All extracted DNA was stored at −20°C until further analysis. The DNA samples from the ticks and rodents were screened for B. burgdorferi s.l. using real-time PCR (Bio- Rad CFX96 Touch Deep Well Real-Time PCR Detection System; Bio-Rad Laboratories, Hercules, CA, USA). Ticks were screened for B. burgdorferi s.l. by targeting 23S Page 5 of 12 Kiranetal. Parasites & Vectors (2024) 17:157 ribosomal RNA (rRNA) [44] in most cases, but the flagellin gene [45] was targeted for ticks captured from specific islands in May and June 2019 (Additional file1: TableS2) as these samples were also included in another study (in preparation). All tick and ear biopsy samples positive for B. burgdorferi s.l. were screened using B. afzelii-specific primers [46]. The list of primers and probes used is available in Additional file1: TableS5, while detailed protocols for the PCR assays are provided in Additional file2: Text S2. Based on the tick data we calculated the percentage NIP as (the number of quantitative PCR-positive nymphs/total number of screened nymphs) × 100, and the density of infected nymph (DIN per 100 m2) as DON × NIP per site per session. For rodents, we calculated the abundance index of infected rodents as (the number of infected animals captured per island divided by trapping effort) × 100 (Additional file1: TableS1). Statistical analyses First, we evaluated whether rodent removal was successful at controlling rodent abundance. We fitted generalized linear mixed models (GLMMs) with a Poisson distribution using the glmer function in the R package lme4 [47] to examine whether rodent abundance depended on sessions and treatment. The model included the number of rodents captured as the response variable; the trapping effort as an offset; and session, treatment and their interaction as explanatory variables. Island identity was included in the model as a random effect to control for the potential correlation among observations from the same island. An equivalent model was used to examine the role of treatment and session on the abundance of infected rodents in the experimental study. We further studied the response of rodent abundance and the abundance of infected rodents to session and the size category (small vs large) of the island in an observational study. The level of tick infestation of rodents in 2019 was examined in relation to the treatment of the island (removal, control, large), trapping session (May, June, July and August/September) and rodent species (bank voles vs field voles) with individual level rodent trapping data. A GLMM approach with negative binomial distribution and trapping island as the random effect was performed using the glmmTMB package in R [48]. An overview of all the tested models is presented in Additional file1: TableS6. On the experimental islands, we investigated whether tick-related measures, i.e. NIPt+1, DONt+1 and DINt+1, in May 2020 were associated with the rodent removal treatment (removal vs control) or rodent abundance index or abundance index of infected rodents in 2019. We used generalized linear models (GLMs; NIPt+1) with binomial error distributions or linear models (LMs; DONt+1 and DINt+1). With respect to the NIPt+1 and DINt+1 models, we used either rodent removal versus control treatment; the rodent abundance index in 2019 (rodent abundance index of each session tested separately); or the abundance index of infected rodents (each session tested separately). The DONt+1 model included the rodent removal treatment versus control treatment, or the rodent abundance index in 2019 (the rodent abundance index with each session tested separately). We tested several NIPt+1, DONt+1 and DINt+1 models using the rodent abundance index in each trapping session separately, as the rodent abundance index estimates at different sessions were correlated and, thus, cannot be included in the same model. To take into account multiple testing for each response variable (rodent abundance index at four sessions using the same data), the threshold significance level was determined based on Bonferroni correction (0.05/4 = 0.0125), i.e. P-value ≤ 0.0125 was considered to indicate a significant result. We also examined DON, NIP and DIN in 2019 to ensure that the treatment (removal vs control) did not differ among the islands before treatment (Additional file2: Text S3; Additional file1: TableS6). On observational islands, we used the same approach as with the experimental islands. We used GLMs and the LM model for testing NIPt+1 and DINt+1, respectively, by including the size category (small vs large) of the islands, either the rodent abundance index in 2019 (tested each session separately) or the abundance index of infected rodents in 2019 (each session tested separately) as fixed factors. DONt+1 models were analysed using LM including the rodent abundance index in 2019 (tested each session separately) as a fixed factor (Additional file 1: TableS6). To take into account the multiple testing for each response variable, the Bonferroni corrected P-value was used in the interpretation of the results (threshold P-value = 0.0125). All data analyses and model building were performed in R (4.2.2) [49]. Results Rodents andtick infestation ontheexperimental andobservational islands in2019 Rodent abundance did not differ between the control and removal islands at the beginning of the study (May 2019) (Fig.3; Additional file1: TableS7). Rodent abundance increased on the control islands during the breeding season, being significantly higher in the August/September trapping session than in the May trapping session (Fig.3, Additional file1: TableS7). Rodent abundance was lower on the removal islands than on the control islands, but the difference was statistically significant only in July Page 6 of 12 Kiranetal. Parasites & Vectors (2024) 17:157 2019 (P = 0.04; Additional file1: TableS7). Additionally, the abundance of infected rodents did not differ between treatments (Additional file1: TableS7). Neither rodent abundance nor the abundance of infected rodents differed between the small (control) and large islands at the beginning of the study (May 2019) (Fig.3; Additional file1: TableS8). However, the abundances were higher on the large islands than on the small control islands in the other trapping sessions (Fig. 3, P < 0.001; Additional file1: TableS8). The mean tick infestations on the rodents were 7.6 larvae and 0.56 nymphs per rodent. Of the 281 rodents examined, 246 were infested by ticks (87.5%) and 74 of these (30%) were infested with both larvae and nymphs. Larval tick infestation load on the rodents did not differ between the treatments or the trapping session, but was significantly higher in bank voles than in field voles (Additional file1: TableS9). Ticks andB. afzelii inticks in2020 In 2020, a total of 550 nymphal ticks were collected. The estimated mean (± standard deviation [SD]) density of nymphs (DONt+1) per 100 m2 was 6.1 ± 1.79 on the rodent removal islands, 8.09 ± 2.77 on the small control islands and 22.9 ± 7.19 on the large islands (Fig.4). In total, 414 of the collected nymphs from May 2020 were tested for B. afzelii. In May 2020, the infection prevalence in nymphs (NIPt+1) was 15.4% (95% confidence interval [CI]: 3–34%) on the rodent removal islands, 21.2% (95% CI: 11–32%) on the small control islands and 12.7% (95% CI: 4–21%) on the large islands (Fig.4). The estimated mean (± SD) density of infected nymphs (DINt+1 = NIPt+1*DONt+1) per 100 m2 in May 2020 was 1.36 ± 0.75 on the rodent removal islands, 1.70 ± 0.63 on the small control islands and 2.95 ± 1.55 on the large islands (Fig.4). In 2019, DON, NIP and DIN did not differ between the rodent removal and control islands in any of the sessions, i.e. in May, June, July, and August/September (Additional file1: TableS10). Experimental study: effects ofrodents on DONt+1, NIPt+1 or DINt+1 On the experimental study islands (control and removal islands), there was no effect of rodent removal treatment in the first study year (2019) on DONt+1, NIPt+1 and DINt+1 (May 2020) (P > 0.5; Additional file1: TableS11). Within the experimental islands, neither DONt+1 (P > 0.5; Additional file1: TableS12) nor NIPt+1 (P > 0.5; Additional file1: TableS13) nor DINt+1 (P > 0.5; Additional file1: TableS14) was associated with the rodent abundance index in any of the trapping sessions in 2019. The abundance index of infected rodents in any of the trapping sessions in 2019 were associated with NIPt+1 on the experimental study islands (Additional file1: TableS13). Observational study: effects ofrodents andisland size on, DONt+1, NIPt+1 or DINt+1 On the observational islands (small control and large islands), DONt+1 showed a positive association with the rodent abundance index in September (Fig.5; Additional file1: TableS12), whereas no association with the rodent Fig. 3 Mean (± standard deviation) rodent abundance index (a) and abundance index of infected rodents (b) on the removal islands, control islands and large islands during the trapping sessions in 2019 Page 7 of 12 Kiranetal. Parasites & Vectors (2024) 17:157 abundance index was observed in the other study sessions (Additional file1: TableS12). Moreover, NIPt+1 was negatively associated with the rodent abundance index and the abundance index of infected rodents in July 2019 (P = 0.011; Fig. 6; Additional file1: TableS15), but no significant associations were observed during the other trapping sessions. However, DINt+1 was not associated with the rodent abundance index in any trapping session in 2019 (P > 0.5; Additional file1: TableS16). Neither DONt+1 nor DINt+1 was associated with the island type (Table1). NIPt+1 was associated with Fig. 4 Mean (+/- standard deviation) (a) density of nymphs (DON), (b) nymph infection prevelance (NIP) and (c) density of infected nymphs (DIN) in May 2020 in study islands (removal, control and large). (a) mean DONt+1 per 100 m2, (b) NIPt+1 per 100 m2 (c) and DINt+1 per 100 m2 in the subsequent year (2020) in removal, control and large islands Fig. 5 Predicted (line with 95% confidence interval in shading) and observed (points, with different colours indicating the small and large islands, respectively) density of nymphs (DONt+1) per 100 m2 in May 2020 in relation to the rodent abundance index in August/September 2019 in the observational study. The figure show the relationship between the rodent abundance index in year t (2019) and DONt+1 in year t+1 (2020) Page 8 of 12 Kiranetal. Parasites & Vectors (2024) 17:157 island type, with a lower NIPt+1 value on large islands (P = 0.046; Table1). Discussion In this study, we examined the association between the abundance of the reservoir rodent host and the abundance of nymphal I. ricinus ticks and a rodent-associated tick-borne pathogen, B. afzelii, in ticks. We observed that rodent removal in the experimental study did not translate into changes in the abundance of ticks, tick infection prevalence or the abundance of infected ticks in the following year. Neither NIPt+1 nor DINt+1 was associated with the abundance index of rodents or infected rodents on the small experimental islands. In the observational component of this study, we aimed to test whether NIPt+1, DONt+1, or DINt+1 was associated with rodent density and island type. Contrary to our hypothesis, we found a negative association between NIPt+1 and the abundance index of rodents and infected rodents. However, we found a positive association between rodent abundance index and DONt+1. NIPt+1 inrelation topreceding rodent abundances Our experimental study revealed no association between rodent abundance at t (2019) and NIPt+1 (2020). This finding contrasts with the findings of an earlier experimental study by Krawczyk etal. [29] which showed that Fig. 6 Predicted (line with 95% confidence interval) and observed (points, with different colours indicating the small and large islands, respectively) nymph infection prevalence (NIPt+1) in May 2020 in relation to the rodent abundance index in July 2019 in the observational study. The figure show the relationship between the rodent abundance index in year t (2019) and DONt+1 in year t+1 (2020) Table 1 The estimated effect of island size on NIPt+1, DONt+1, and DINt+1 in relation to island type (small control islands vs large islands) *Statistically significant a NIPt+1 (nymph infection prevalence in following year [2020]) was estimated using generalized linear models (GLMs). DONt+1 (density of nymphs in following year [2020]) and DINt+1 (density of infected nymphs the following year [2020]) were estimated using linear models (LM) b Small control islands are included in the intercepts Response variableaExplanatory variablebEstimate (standard error) Z-value P-value NIPt+1 Intercept − 1.33 (0.22) − 6.02 < 0.001 Large island − 0.63 (0.31) − 1.99 0.046* DINt+1 Intercept 1.71 (1.18) 1.44 0.19 Large island 1.25 (1.67) 0.75 0.48 DONt+1 Intercept 8.10 (5.45) 1.49 0.18 Large island 14.8 (7.70) 1.92 0.09