The Role of Environmental Factors in Lyme Disease Transmission in the European Union: A Systematic Review

Background: Lyme disease (LD) is an emergent vector-borne disease caused by Borrelia spp. and transmitted through infected ticks, mainly Ixodes spp. Our objective was to determine meteorological and environmental factors associated with LD transmission in Europe and the effect of climate change on LD. Materials and methods: A systematic review following the PRISMA guidelines was performed. We selected studies on LD transmission in the European Union (EU) and the European Economic Area (EEA) published between 2000 and 2022. The protocol was registered in the PROSPERO database. Results: We included 81 studies. The impact of environmental, meteorological or climate change factors on tick vectors was studied in 65 papers (80%), and the impact on human LD cases was studied in 16 papers (19%), whereas animal hosts were only addressed in one study (1%). A significant positive relationship was observed between temperature and precipitation and the epidemiology of LD, although contrasting results were found among studies. Other positive factors were humidity and the expansion of anthropized habitats. Conclusions: The epidemiology of LD seems to be related to climatic factors that are changing globally due to ongoing climate change. Unfortunately, the complete zoonotic cycle was not systematically analyzed. It is important to adopt a One Health approach to understand LD epidemiology.


Introduction
Lyme disease (LD) is a common bacterial vector-borne disease in countries of the European Union (EU) and European Economic Area (EEA).Its pathogenic agent, Borrelia sp, is transmitted through the bite of infected ticks, mainly Ixodes spp.[1][2][3].Other ticks that are present in Europe, like Dermacentor spp.and Haemaphysalis spp., have been identified as carriers of Borrelia spp.spirochete, although little is known about their vector competence [4][5][6].Sporadic infection and transmission by Rhipicephalus spp, Hyalomma spp.and Amblyomma spp.have been reported [7,8].Deer and rodents are common animal hosts, crucial for the maintenance of the zoonotic cycle in the wild [1][2][3]9].
In the EU/EEA, where it is considered endemic in many places, with more than 360,000 cases reported over the last two decades, the main vector of LD is Ixodes ricinus and the main pathogens are Borrelia afzelii, B. garinii and B. burgdorferi [1][2][3].In 2018, Lyme neuroborreliosis was included on the list of diseases under EU epidemiological surveillance due to increasing trends in the diagnosis of LD cases and vector presence [9,10].Areas of presence for LD vectors were mapped based on updated information up to March 2022 and

Results
The systematic search strategy yielded 1218 references.After screening the titles and abstracts, we retained 113 articles for full-text screening.Eighty-one articles met all inclusion criteria (Figure 1).(Version 2023, Sydney, Australia; https://www.canva.com/,accessed on 30 June 2023) for mapping and creating explanatory figures.

Results
The systematic search strategy yielded 1218 references.After screening the titles and abstracts, we retained 113 articles for full-text screening.Eighty-one articles met all inclusion criteria (Figure 1).

Descriptive Characteristics of the Studies
Most papers (n = 59) were published between 2011 and 2022.Six studies covered more than one country and three studies focused on the whole European continent (Figure 2).Most research was carried out in Germany (n = 15), France (n = 12) and Belgium (n = 11).The most frequently used types of analyses were association/correlation analyses (n = 56), predictive models (n = 23) and spatial models (n = 15).Twelve studies used two or more different types of modeling approaches (Table 1).Definitions of analyzed variables are specified in Supplementary Materials: Table S2.
Overall, the studies were of medium or good quality (average 11.42 points).The main reasons for scoring lower were the improper identification of the sources for data or the data collection procedures and unclear results (Supplementary Materials: Figure S1).

Descriptive Characteristics of the Studies
Most papers (n = 59) were published between 2011 and 2022.Six studies covered more than one country and three studies focused on the whole European continent (Figure 2).Most research was carried out in Germany (n = 15), France (n = 12) and Belgium (n = 11).The most frequently used types of analyses were association/correlation analyses (n = 56), predictive models (n = 23) and spatial models (n = 15).Twelve studies used two or more different types of modeling approaches (Table 1).Definitions of analyzed variables are specified in Supplementary Materials: Table S2.
Overall, the studies were of medium or good quality (average 11.42 points).The main reasons for scoring lower were the improper identification of the sources for data or the data collection procedures and unclear results (Supplementary Materials: Figure S1).

Lyme Disease Vectors
Sixty-five studies addressed the impact of environmental factors on LD vectors.Twenty-four of these studies additionally analyzed Borrelia sp.infection in ticks, which ranged from 0.25% in Germany [60] to 38.0% in Italy [76] (Tables 1 and 2, Supplementary Materials: Table S3).

Analyzed Species Countries
Analyzed Borrelia species
Saturation deficit, which describes the functional relationship between saturation vapor pressure, temperature and relative humidity and provides an integrated measure of the drying power of the atmosphere, was analyzed in nine studies in relation to tick abundance and density [48,49,53,62,68,78,86,89,91].Seven studies observed a positive relationship between I. ricinus abundance [48,53,78,86,91] or density [89] and saturation deficit.
Soil-related variables were analyzed in twelve studies in relation to tick abundance [15,29,39,40,48,49,52,55,56,63,74] and tick bites [80], of which, only three studies found positive relationships [49,63,74].Clay and silt soils were related to a higher abundance of nymphal ticks, and sandy soil was related to higher nymphal infection prevalence [49].Lower nymphal infection prevalence values, however, were observed with silt soil [49].Moder humus, i.e., a kind of forest floor in deciduous and mixed-wood forests characterized by a thick layer of fragmented leaves, was strongly associated with nymph abundance in France [48].Limestone [74] and increased soil water content [63] were also related to increased tick abundance.
Both medium-low-and high-emission scenarios [98] and future climate change projections [41] were positive predictors for current and future I. ricinus densities in Scandinavia [98] and Europe [41].
Figure 3 shows the significant effects of the analyzed environmental variables on LD vector abundance and density.

Lyme Disease in Human Hosts
Sixteen studies addressed the impact of environmental variables on human LD cases (Supplementary Materials: Table S4).Two studies focused on specific forms of LD infection, i.e., human neuroborreliosis [99] and erythema migrans (EM), a pathognomonic skin

Lyme Disease in Human Hosts
Sixteen studies addressed the impact of environmental variables on human LD cases (Supplementary Materials: Table S4).Two studies focused on specific forms of LD infection, i.e., human neuroborreliosis [99] and erythema migrans (EM), a pathognomonic skin rash that appears following infection in up to 80% of cases [96].The impact of temperature on human neuroborreliosis [99], EM [96] and LD incidence [24,31,32,69,70] was assessed in seven studies.Mean [31,69,70,96,99] and minimum [31] monthly [31,96,99] and weekly [69,70] air [31,69,70,96,99] and soil [31] temperatures were positive predictors for human LD cases.Higher numbers of winter days with an average temperature below 0 • C in Sweden were related to lower numbers of reported EM cases in the study region [96].Growing degree days, an indicator of heat accumulation, was also positively related to increased human LD incidence [24].Only one study showed no relationship between temperature and human LD incidence [32].
Four studies focused on precipitation and human LD incidence.Mean monthly precipitation was positively associated with increases in neuroborreliosis cases [99].A reduced number of frost days was also positively related to increased human LD incidence because of its critical effect on small mammals, the main hosts for questing larvae and nymphs.This is because of higher host mortality during harsh winters.Ticks are therefore unable to find suitable hosts to survive, which then reflects on lower human LD incidence [31].However, two other studies found no relationship between precipitation and human LD cases [32,96].
Humidity was assessed in relation to human LD [27,31], neuroborreliosis [99] and EM incident cases [96].Both the annual cumulative Normalized Difference Water Index (NDWI) [27] and the number of summer days with relative humidity above 86% [96] correlated positively with the number of human cases.In contrast, neither mean monthly relative humidity [96,99] nor soil humidity at the end of winter [31] showed any relation to the number of human cases.
Altitude was a positive predictor for human LD cases [46,90], i.e., human LD cases were also registered at higher altitudes.
Regarding NAO (North Atlantic Oscillation), a cyclical meteorological phenomenon, and human LD incidence, one study showed no relationship [32].Another study found a negative correlation between NAO index and the number of human cases and was used to accurately predict human cases in Europe [35].
Three studies analyzed human LD incidence in relation to vegetation.One study found no relationship between the mean monthly NDVI and increases in human LD cases in Slovenia [90], whereas it was a positive predictor for human LD incidence in Belgium [24,26].
Climate change predictions, i.e., warmer temperatures, higher CO 2 emissions and changes in rainfall patterns, among others, showed a positive relationship with human LD incidence in Slovenia as a result of the vector niche shifting to new habitats [90].
Figure 4 shows the significant effects of the analyzed environmental variables on human LD incidence.

Lyme Disease in Animal Hosts
Only one study analyzed animal hosts' infections in relation to land use.The presence of pastures and natural grasslands in Romania was a positive predictor for Borrelia spp.infection in wild boars, roe deer and cattle.Most infections were due to B. afzelii, B. burgdorferi sensu stricto and B. garinii, although B. valaisiana, B. spielmanii and B. bavariensis were also detected [59].

Lyme Disease Risk and Expansion
Eight studies analyzed and predicted an expansion of LD to other regions both within countries and cross-border because of the influence of environmental variables [26,41,43,44,58,72,90,98], of which, one study modeled the ecological risk of LD in Europe focusing on the whole transmission cycle under future climate change scenarios.It considered species distribution mapping for animal hosts, i.e., deer, rodents and birds, as well

Lyme Disease in Animal Hosts
Only one study analyzed animal hosts' infections in relation to land use.The presence of pastures and natural grasslands in Romania was a positive predictor for Borrelia spp.infection in wild boars, roe deer and cattle.Most infections were due to B. afzelii, B. burgdorferi sensu stricto and B. garinii, although B. valaisiana, B. spielmanii and B. bavariensis were also detected [59].

Lyme Disease Risk and Expansion
Eight studies analyzed and predicted an expansion of LD to other regions both within countries and cross-border because of the influence of environmental variables [26,41,43,44,58,72,90,98], of which, one study modeled the ecological risk of LD in Europe focusing on the whole transmission cycle under future climate change scenarios.It considered species distribution mapping for animal hosts, i.e., deer, rodents and birds, as well as ticks and human LD risk [43].Human LD cases will expand to western regions of Slovenia, especially those of lower altitude and rich in wood forests under future climate change scenarios [90], and to southern and northeastern Belgium under the influence of an enhanced NDVI [26].LD vectors are expected to expand to northern regions of Italy, especially the Piedmont [72], and to northwestern Germany [58] and large regions in Scandinavia as far as 70 • N [41,43,44,98].This expansion will be exacerbated by the presence of coniferous and deciduous forests [72] and black alder trees [98], as well as increases in mean temperatures [41,44,58] and future climate change scenarios [41,43].According to two studies, by 2030, vectors will have expanded to Nordic countries and central Europe because of increases in temperature and NDVI and future climate change scenarios.However, some models predict that by 2050, LD transmission may be disrupted in some areas of southern Europe because of decreased suitability and no niche overlap between ticks and hosts, due to future predictions of climate change and the transformation of forests into crops [43,44].
We observed contradictory results for some meteorological variables [15,34,37,42,44,45,48,52,58,60,61,[63][64][65]81,[84][85][86]88,92,95,97].This might have been a result of different climate zones in Europe, different vegetation cover and different suitability for ticks: southern Europe is characterized by a subtropical climate, where increasing temperatures may even be a limiting factor for suitable tick habitats, whereas higher precipitation may favor tick establishment.However, most parts of Scandinavia present a cold climate, whereas the climate in central Europe is temperate maritime in the west and temperate transitional in the east [104].Not only the environment, such as certain flora and vegetation cover found in northern latitudes [105], but also fauna, such as Cervus elaphus and Capreolus capreolus, which are widely present in central and northern Europe and act as hosts for adult ticks, play a major role in LD epidemiology [106].In addition, some studies analyzed the meteorological or environmental variations within a season, while others focused on several years and analyzed the interannual variations and provided climate change projections.These different methodological approaches may explain some of the observed differences.
Those studies that considered human LD cases mainly focused on LD incidence [21,24,26,27,31,32,35,46,69,70,75,90,96,99], and only two studies analyzed the seroprevalence of LD antibodies in humans [23,83].This might have been due to the lack of prevalence studies in endemic regions, difficulties in diagnosing LD, the development of new diagnostic methods and the lack of standardized diagnostic protocols and the lack of routine screening for LD antibodies [107,108].Higher temperature [31,69,70,96,99] and less precipitation [31,99] were associated with an increase in human LD cases.This might be due to human outdoor activity and, thus, exposure to infected ticks, being higher on warm and sunny, non-rainy days, as people might go to the countryside or parks or perform outdoor activities [109].In addition, the number of questing ticks is higher during summer [110].Given that extremely high temperatures and droughts were registered during the summer of 2023 in southern and southwestern Europe [111], this may influence future LD epidemiology, leading to higher human LD incidence.In the case of human exposure to LD vectors, additional factors, i.e., animal host abundance or human social behavior, may be important, since the degree of human activity in nature varies and may be affected by the environment.Therefore, both the environment and human behavior have important effects on the whole zoonotic cycle [112].
The most frequent limitations identified by the studies' authors were the lack of the analysis of other variables that may influence LD dynamics (n = 44) and concerns about the study and/or model accuracy (n = 11).This might compromise the results of the papers and highlights the importance of developing comprehensive and holistic models when analyzing other variables.

Limitations
Our study has some shortcomings.We performed a search that was bound to certain inclusion criteria.Other relevant articles might therefore have not been included.However, all included articles were published in English, which is why we believe that most relevant articles were included.We followed the PRISMA guidelines for systematic reviews to limit selection bias.In addition, we observed different methodological qualities in the included studies.Therefore, we used a specific tool to evaluate the studies' quality, which on average scored very high.

Conclusions
LD is expanding across Europe.The epidemiology of LD is related to the presence of vectors, which is related to climate and other environmental factors that are changing globally due to ongoing climate change.The environmental factors that most frequently correlated to changes in LD dynamics were temperature, precipitation, humidity and the incursion of human beings into different natural land habitats.Most studies found a positive relationship, although agricultural habitats were associated with decreased human LD incidence.Unfortunately, the complete zoonotic cycle was not systematically analyzed in most papers.Thus, it is difficult to determine the independent impact of environment on the different components of the transmission cycle.It is important to adopt a One Health approach to understand LD epidemiology and to strengthen the surveillance of this emerging disease and its vector.While temperature is increasing worldwide, the impacts of climate change on precipitation present important geographical variations according to the latest Intergovernmental Panel on Climate Change (IPCC) report [117], and consequently, the global impact of climate change on tick populations and LD epidemiology may present important variations within Europe.

Supplementary Materials:
The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/tropicalmed9050113/s1:Text S1.Inclusion and exclusion criteria.Text S2.Search strategy.Figure S1.Scored points in quality assessment (n = 81).Table S1.Eligibility criteria.Table S2.Definition of analyzed variables.Table S3.Effect of different environmental variables on vectors.Table S4.Effect of different environmental variables on LD in human hosts.

Figure 2 .
Figure 2. Articles by analyzed countries (n = 81).Countries where no studies were performed are displayed in gray.

Figure 2 .
Figure 2. Articles by analyzed countries (n = 81).Countries where no studies were performed are displayed in gray.

Figure 3 .
Figure 3. Significant effects of analyzed environmental variables on LD vector abundance and density (n = 65).Countries with significant negative relationships between environmental factors and LD vector abundance and density are shown in orange (left), and countries with significant positive relationships between environmental factors and LD vector abundance and density are shown in yellow (right).The factors significantly related to vector abundance and density in each country are shown inside the country's shape.The distribution of environmental variables inside each country's shape is arbitrary.Countries with no data are shown in gray.

Figure 3 .
Figure 3. Significant effects of analyzed environmental variables on LD vector abundance and density (n = 65).Countries with significant negative relationships between environmental factors and LD vector abundance and density are shown in orange (left), and countries with significant positive relationships between environmental factors and LD vector abundance and density are shown in Trop.Med.Infect.Dis.2024, 9, x FOR PEER REVIEW 13 of 21

Figure 4 .
Figure 4. Significant effects of analyzed environmental variables on human LD incidence (n = 16).Countries with significant negative relationships between environmental factors and human LD incidence are shown in orange (left) and countries with significant positive relationships between environmental factors and human LD incidence are shown in yellow (right).The factors significantly related to vector density in each country are shown inside the country's shape.The distribution of environmental variables inside each country's shape is arbitrary.Countries with no data are shown in gray.

Figure 4 .
Figure 4. Significant effects of analyzed environmental variables on human LD incidence (n = 16).Countries with significant negative relationships between environmental factors and human LD incidence are shown in orange (left) and countries with significant positive relationships between environmental factors and human LD incidence are shown in yellow (right).The factors significantly related to vector density in each country are shown inside the country's shape.The distribution of environmental variables inside each country's shape is arbitrary.Countries with no data are shown in gray.

Table 1 . Cont. First Author Year of Publication Analyzed Countries Analyzed Vector (Species) Analyzed Reservoirs and Hosts Borrelia Species Study Object Analytical Approach Scored Points at Quality Assessment
AM: association/correlation models; LD: Lyme disease; ND: no data; PM: predictive model; SM: spatial model.Study quality: the quality of the included studies was assessed.Further information can be found in the Material and Methods and Supplementary Materials: FigureS1.

First Author Year of Publication Analyzed Countries Analyzed Vector (Species) Analyzed Reservoirs and Hosts Borrelia Species Study Object Analytical Approach Scored Points at Quality Assessment
burgdorferi s.l.Human cases AM 10 Heylen D [28] 2013 Belgium I. ricinus Animals B. burgdorferi s.l.Tick abundance PM 11 Hönig V [29] 2015 Czech Republic I. ricinus ND B. afzelii, B. garinii, B. burgdorferi s.s., B. Tick abundance AM 11

Table 2 .
Observed Borrelia and Ixodes species in different countries.