Oak acorn crop and Google search volume predict Lyme disease risk in temperate Europe
Introduction
Lyme borreliosis is the most common zoonosis and a major health concern in the northern hemisphere (Barbour and Fish, 1993, Pfäffle et al., 2013). It is caused by Borrelia burgdorferi sensu lato transmitted by hard–bodied ticks (genus Ixodes), and the abundance of infected tick nymphs is a good predictor of the disease risk in humans (Barbour and Fish, 1993, Ostfeld, 2010, Stafford et al., 1998). In the eastern United States temporal fluctuations in oak (Quercus spp.) acorn production (mast seeding: Crone and Rapp, 2014, Kelly, 1994) cause temporal variation in food provision for granivorous rodents, which shapes their population dynamics (McShea, 2000, Wolff, 1996). This in turn influences the abundance of infected tick nymphs and the risk of Lyme disease in humans (Jones, Ostfeld, Richard, Schauber& Wolff 1998a; Ostfeld, 2010, Ostfeld et al., 2006, Ostfeld et al., 1996, Schauber et al., 2005). However, the connection between masting and Lyme disease in the eastern United States has been disputed, given the variety of factors that may contribute to the disease risk (Estrada-Pena, 2009, Randolph, 1998).
The ecological link between mast seeding and Lyme disease in eastern United States is as follows. In summer after a good acorn year, host-seeking larvae of the blacklegged tick (Ixodes scapularis) co-occur with high abundance of rodents caused by previous year surplus food supply (Jones et al., 1998a, Ostfeld et al., 2006, Ostfeld et al., 2001). High rodent numbers increase larval feeding chances and survival, which leads to high nymph densities next season (Keesing et al., 2009, Ostfeld, 2010, Ostfeld et al., 2006). Moreover, high reservoir-competence of rodent hosts increases B. burgdorferi prevalence among tick nymphs (Ostfeld et al., 2001, Ostfeld et al., 2006). Next spring, numerous infected nymphs search for vertebrate hosts, including humans (Schauber et al. 2005). Nymphs are mostly responsible for transmitting Lyme disease to humans because their small size makes them difficult to detect. Moreover, their summer peak in activity coincides with the peak of human outdoor activity (Ostfeld, 2010, Barbour and Fish, 1993). Thus, oak acorn production in year T influences rodent abundance next year (T + 1), which subsequently affects infected nymph abundance and Lyme disease risk in year T + 2 (Barbour and Fish, 1993, Ostfeld et al., 2001, Ostfeld et al., 2006, Schauber et al., 2005).
Fluctuations in oak acorn production have similar effects on wildlife in Europe (Bogdziewicz et al., 2016, Jędrzejewska and Jędrzejewski, 1998, Pucek et al., 1993), but the effects of oak masting on the incidence of Lyme disease have not been studied. Seed fall in autumn determines rodent abundance next summer (Jensen, 1982, Pucek et al., 1993, Zwolak et al., 2016), and small mammals are a good reservoir for the Borrelia spirochete (Franke et al., 2013, Gern, 2008, Michalik et al., 2003, Siński et al., 2006). Moreover, the sheep tick (Ixodes ricinus), ecological equivalent of I. scapularis in the European host–tick–pathogen system, has similar ecology to the blacklegged tick (Barbour and Fish, 1993, Beytout et al., 2007, Hubálek et al., 2003, Korenberg et al., 2002, Siński et al., 2006). Thus, we hypothesize that a similar chain of ecological events linking acorn production and Lyme borreliosis risk might occur in Europe.
Traditional methods of gathering ecological data can be supplemented with new technologies. Temporal fluctuations in Google search volume and Wikipedia logs have been used to forecast influenza, dengue or tuberculosis outbreaks (Generous et al., 2014, Ginsberg et al., 2008, McIver and Brownstein, 2014). In a recent study, Google Trends were successfully used to collect national–scale data on fluctuations in rodent numbers, to study the role of rodent predation pressure in wood warbler (Phylloscopus sibilatrix) habitat selection (Szymkowiak & Kuczyński 2015). Indices of rodent abundance obtained using Google search engine were positively validated with field data (see details in Szymkowiak & Kuczyński 2015). Here, we used temporal changes in Google search volume to trace ecological chain reactions linking acorn production with Lyme disease risk. We assumed that people use the Internet as a source of information about ticks and Lyme disease. Hence, an increase in number of interactions between humans and ticks should lead to increase in the search volume of focal keywords. We selected two unambiguous keywords i.e., “kleszcz” (which stands for tick in Polish) and “borelioza” (Lyme disease), and calculated search volumes for each year of the study period. Moreover, we calculated search volumes for the term “na myszy” (in Polish: something for/against mice, hereafter “mice”) for each year as this appears to provide a reliable index of annual fluctuations in rodent numbers (cf. Szymkowiak & Kuczyński 2015).
We used our dataset to test the predictions derived from Ostfeld et al.’s works (referenced above). First, acorn production should positively influence rodent numbers (e.g. Pucek et al. 1993). Thus, we tested whether acorn production in year T positively correlates with Google index of mice abundance in year T + 1. Second, the positive effect of acorn production (year T) on rodent numbers (year T + 1) should translate into increase in intensity of tick-human interactions and, consequently, Lyme disease cases (in year T + 2). Thus, we tested whether Google index of rodent abundance in year T + 1 correlates positively with Google search volume for “tick” in year T + 2. Next, we tested whether acorn production (year T) correlates positively with Lyme disease-related key words (“tick” and “Lyme disease”), and Lyme disease casesin year T + 2. To our knowledge, this is the first study linking temporal fluctuations in acorn production with Lyme disease risk in Europe. Our research answers the call of Jones et al. (1998b), “Nevertheless, we hope that epidemiologists will test the power of acorn production as a predictor of Lyme disease risk in European oak forests”, which stayed ignored for almost 20 years.
Section snippets
Data collection
We extracted the epidemiological data on the number of reported cases of Lyme disease in Poland from annual (2006–2013) reports on infectious diseases (Fig. 1B). These data were collected by the Provincial Sanitary–Epidemiological Stations and provided by the National Institute of Public Health–National Institute of Hygiene in Poland (http://www.pzh.gov.pl/).
Data on annual acorn production of pedunculate (Quercus robur) and sessile oak (Q. petraea) were collected in years 2005–2011 for the
Results
As hypothesized, acorn production in year T was a good predictior of the number of Lyme disease cases in year T + 2 (Table 1 and Fig. 2C). We also found a postivie correlation between acorn production in year T and Google search volume for “tick” in year T + 2 (Table 1 and Fig. 2B). Moreover, there was a positive correlation between acorn production in year T and Google search volume for “mice” (T + 1) and “Lyme disease” (T + 2) (Fig. 2A and D), but both relationships were not significant at α = 0.05
Discussion
Risk of acquiring Lyme disease by humans is directly linked with the abundance and distribution of questing ticks. Thus, it is essential to understand factors shaping tick population dynamics. We used a combination of three sources of the data: National Institute of Hygiene in Poland on Lyme disease incidence, Polish State Forests on oak acorn crop, and Google Trends, and demonstrated that oak acorn production is associated with the Lyme disease risk in humans. The increase in Google searches
Acknowledgements
We would like to thank Rafał Zwolak and Anonymous Reviewers for critical comments on previous versions of our manuscript. The work was done while MB was supported by Polish National Science Centre grant “Harmonia” no. 2012/04/M/NZ8/00674, Adam Mickiewicz Foundation Scholarschip awarded in 2014, and NCN “Etiuda” grant no. 2015/16/T/NZ8/00018. JSZ was supported by Polish National Science Centre grant “Preludium” no. 2012/07/N/NZ8/00129. We are grateful to Jerzyna Przypaśniak from the State
References (53)
- et al.
Exploring gaps in our knowledge on Lyme borreliosis spirochaetes–updates on complex heterogeneity, ecology, and pathogenicity
Ticks and Tick-Borne Diseases
(2013) - et al.
European Reservoir Hosts of Borrelia burgdorferi sensu lato
Zentralblatt für Bakteriologie
(1998) - et al.
Mast seeding and Lyme disease
Trends in Ecoogy & Evoution
(1998) The evolutionary ecology of mast seeding
Trends in Ecology & Evolution
(1994)- et al.
Empirical evidence for key hosts in persistence of a tick-borne disease
International Journal for Parasitology
(2003) - et al.
The ecology of tick-borne diseases
International Journal for Parasitology
(2013) Mighty theories from little acorns grow: Is Lyme disease risk predictable from mast-seeding by oak trees?
Trends in Ecology & Evolution
(1998)- et al.
Beech masting modifies the response of rodents to forest management
Forest Ecology and Management
(2016) - et al.
The biological and social phenomenon of Lyme disease
Science
(1993) - et al.
Lyme borreliosis incidence in two French departments: Correlation with infection of Ixodes ricinus ticks by Borrelia burgdorferi sensu lato
Vector-Borne and Zoonotic Diseases
(2007)
How do vertebrates respond to mast seeding?
Oikos
Time series analysis: Forecasting and control
Multiple causes of variable tick burdens on small-mammal hosts
Ecology
Relating increasing hantavirus incidences to the changing climate: The mast connection
International Journal of Health Geographics
Introductory time series with R
Resource depletion, pollen coupling, and the ecology of mast seeding
Annals of the New York Academy of Sciences
Using internet search behavior to assess public awareness of protected wetlands
Conservation Biology
Tick-borne pathogens, transmission rates and climate change
Frontiers in Bioscience
Global disease monitoring and forecasting with wikipedia
PLoS Computational Biology
Borrelia burgdorferi sensu lato, the agent of lyme borreliosis: Life in the wilds
Parasite
Detecting influenza epidemics using search engine query data
Nature
Association of Borrelia afzelii with rodents in Europe
Parasitology
Body mass and sex-biased parasitism in wood mice Apodemus sylvaticus
Oikos
Longitudinal surveillance of the tick Ixodes ricinusfor borreliae
Medical and Veterinary Entomology
Risk indicators for the tick Ixodes ricinus and Borrelia burgdorferi sensu lato in Sweden
Medical and Veterinary Entomology
Predation in vertebrate communities: The Bialowieza Primeval Forest as a case study
Cited by (22)
Evaluating sampling schemes for quantifying seed production in beech (Fagus sylvatica) forests using ground quadrats
2021, Forest Ecology and ManagementCitation Excerpt :This reproductive strategy, which is typical of many anemophilous tree species or those dispersed by food-hoarding animals (Bogdziewicz et al., 2018a), has cascading effects on overall ecosystem functioning. Indeed, masting affects plant regeneration (Cutini et al., 2015), tree species composition (Lichti et al., 2014; Yang et al., 2019), and reverberates across trophic levels, driving population dynamics from primary resource consumers (Bisi et al., 2018; Cutini et al., 2013) to top predators (Masaki et al., 2020), also including those carrying human infectious diseases (Bogdziewicz and Szymkowiak, 2016; Rubel and Brugger, 2021). Therefore, the understanding of masting patterns has relevant management implications (Ascoli et al., 2020; Azad et al., 2017; Cutini et al., 2015; Wagner et al., 2010).
Borrelia prevalence and species distribution in ticks removed from humans in Germany, 2013–2017
2020, Ticks and Tick-borne DiseasesCitation Excerpt :The density of infected nymphs is usually regarded as more important for Lyme borreliosis risk than the infection prevalence per se (Ostfeld et al., 2006). Nevertheless, a similar relationship has been detected between acorn production and Lyme borreliosis incidence in humans in Poland (Bogdziewicz and Szymkowiak, 2016). Both the detected Borrelia prevalence and Borrelia species distribution in the human-derived ticks in the present study are comparable to data on questing ticks collected in Germany (Tappe et al., 2014; Blazejak et al., 2018) and elsewhere in Europe (Kjelland et al., 2010; Lommano et al., 2012).
Predicting tick-borne encephalitis using Google Trends
2020, Ticks and Tick-borne DiseasesCitation Excerpt :Thus, combining traditional field techniques with Internet based data could be a promising avenue for future investigation on tick-borne conditions. For example research by Bogdziewicz and Szymkowiak (2016) found a correlation between acorn masting and Google searching on ticks over a two year period. Ecological processes and human activities are important in tick-borne conditions; considering these factors is important.
Effects of land use change and Quercus rubra introduction on Vaccinium myrtillus performance in Pinus sylvestris forests
2019, Forest Ecology and ManagementCitation Excerpt :However, the presence of an understory rich in plants was the necessary condition, whereas forests with Q. rubra are almost totally lacking of herb and moss layers (Woziwoda et al., 2014a, 2017). Additionally, forests with deciduous trees are perceived as habitats with higher risk of lyme disease occurrence (Bogdziewicz and Szymkowiak, 2016), and for this reason are also avoided by visitors. Our study for the first time quantified an impact of land use change (ancient vs. recent forests) and alien (invasive) tree (Q. rubra) introduction on biomass and population dynamics of bilberry (V. myrtillus) – a dominant component of the Scots pine forest understory and a very important ES provider.