Prevalence and prediction of Lyme disease in Hainan province.

Lyme disease (LD) is one of the most important vector-borne diseases worldwide. However, there is limited information on the prevalence and risk analysis using correlated factors in the tropical areas. A total of 1583 serum samples, collected from five hospitals of Hainan Province, were tested by immunofluorescence assay (IFA) and western blot (WB) analyses using anti-Borrelia burgdorferi antibodies. Then, we mapped the distribution of positive rate (by IFA) and the spread of confirmed Lyme patients (by WB). Using ArcGIS, we compiled host-vector-human interactions and correlated data as risk factor layers to predict LD risk in Hainan Province. There are three LD hotspots, designated hotspot I, which is located in central Hainan, hotspot II, which contains Sanya district, and hotspot III, which lies in the Haikou-Qiongshan area. The positive rate (16.67% by IFA) of LD in Qiongzhong, located in hotspot I, was higher than that in four other areas. Of confirmed cases of LD, 80.77% of patients (42/52) whose results had been confirmed by WB were in hotspots I and III. Hotspot II, with unknowed prevalence of LD, need to be paid more attention considering human-vector interaction. Wuzhi and Limu mountains might be the most important areas for the prevalence of LD, as the severe host-vector and human-vector interactions lead to a potential origin site for LD. Qiongzhong is the riskiest area and is located to the east of Wuzhi Mountain. In the Sanya and Haikou-Qiongshan area, intervening in the human-vector interaction would help control the prevalence of LD.

Introduction

Lyme disease (LD) is caused by the tick-borne spirochete Borrelia burgdorferi, and has been reported to be widely distributed across China mainland[1-2]. Human LD generally occurs in stages, from the early localized stage of erythema migrans, fatigue, chills, and fever, to a late disseminated stage of intermittent bouts of arthritis with severe joint pain and swelling and neurological symptoms [3-4].

While culture of B. burgdorferi from patient specimens is difficult and single serological tests are partially hampered by the disease stage at which antibodies start to appear, leading to insufficient results; thus, patients may still be seronegative in the early stages of infection[4-5]. In China, without surveillance, Lyme patients often don’t find themselves ill until late symptoms occur. Therefore, we only collected serum samples from patients with arthritis and neurological symptoms, as these two symptoms were common in lyme patients[6]. Nowadays, the US-CDC diagnosed LD by serological tests in a two-step process, including a screening test (enzyme immunoassay (EIA) or immunofluorescence assay (IFA)) and confirmation test (western blotting, WB)[4].

LD is known to be a vector-borne disease distributed by ticks[7]. The spirochete preserved in nature depends on transmission through the host-vector route[8-9]. Tick bites are the most common way by which humans acquire LD[10]. The abundance of tick communities has deep implications for LD[11]. The risk of acquiring this disease depends on the encounter of ticks that are bacterial carriers with human beings at the appropriate site[12]. This process involves several complicated factors, including the presence of a suitable habitat for tick survival, breeding, and questing, and a proper area where humans can come into contact with questing ticks. While the relationship between the incidence and density of infected nymphs is sometimes weak [13,14], it is due to the complicated ecosystem for their unknown interactions. This gap could be partly filled by predictive models. Currently, there are no vaccines that can protect humans from LD; therefore, the appropriate action is necessary to control the risk of infection to avoid tick exposure [15,16]. To effectively target prevention and control, public organizations need spatial estimates of LD risk, and correlated factors should be compiled in order to model disease prevalence.

LD infection mostly occurs after a tick bite. In rural areas, locals are bitten by ticks in farmland, grassland, bush, and forest in their daily life[17]. The risk of acquiring LD depends on ticks that may or may not be carrying the pathogen B. burgdorferi. Hence, it is clear that ticks probably carry the borrelia burgdorferi through contact with their hosts[18]. A previous study revealed that small mammals are the most important hosts of ticks, and play an essential role in persistence of B. burgdorferi [19]. In China, B. burgdorferi strains have been isolated either from ticks or from the urinary bladders or kidneys of rodents of the family Muridae [20,21].

Hainan (also Hainan Island) is located in the South China Sea. It is separated from Guangdong’s Leizhou Peninsula to the North by Qiongzhou Strait. The island is renowned for its tropical climate, which is completely different from that of the Chinese mainland. Due to its unique geographical situation and climate, it is important to study the distribution of LD in Hainan Province, as the epidemiology, pathogenesis, and management of LD in Hainan would be different from those in the Chinese mainland. Tick borne disease spotted fever was firstly discovered since 1990s in Hainan[22]. However, there is still lack of data about LD in Hainan Province. Then, we investigated and confirmed that LD existed in Hainan Province since 2014[23]. All six patients were distributed in the south coast of Hainan Province. Furthermore, the investigation was insufficient for the lack of research in the central and northern regions of Hainan Province. Therefore, the serums in our investigation were from 5 hospitals which were located in the central and north Hainan (Fig 1). Temperature regulation of the development rates and mortality of ticks depends on water loss, as these factors are regulated by the relative humidity and the air saturation deficit[24]. Therefore, from September to November, an increasing number of ticks appear outside and the natural environment during this time in Hainan is favorable to ticks. Normally, along with growth in number of ticks, patients with LD show a similar increasing a same trend with a time lag[11]. To study the distribution of patients with LD and the potential areas that are at risk for acquiring LD, and to understand the spread of LD in Hainan Province, we conduct investigations in suspected patients presenting with LD-related symptoms. Here, from September to November, we collected blood samples from patients in five hospitals presenting with arthritis and nervous system disease which were all the large general hospitals in local area. Then, we analyzed the spatial pattern of LD risk using the knowledge based method with the correlated factors.

Fig 1

Distribution of 1583 serum samples from five hospitals.

This map was plotted by a combination of ArcGIS software version 10.1 (Esri, USA) and Photoshop CS 2.0 software (Adobe Systems, USA).

Methods

Ethics statement

This study and the research protocol were reviewed and approved by the Ethical Committee (Institutional Review Board, IRB) of People’s Hospital of Sanya (License number: ME[2018]-3). All patients gave written informed consent for participation in this study with their identifiable information, and the legal guardians of young children (less than 12 years of age) provided informed consent on their behalf; in accordance with the Declaration of IRB approval.

Blood samples

We collected 1583 serums from patients with arthritis or neurological symptoms from 5 hospitals in Hainan province of China from September to November in 2015. Five hospitals were Wenchang People’s Hospital, Dan zhou People’s Hospital, Dongfang People’s Hospital, Qiongzhong People’s Hospital and The Third People’s Hospital of Haikou (Fig 1). These samples were examined for the presence of B. burgdorferi serologically by the two-step tests (IFA and WB).

Immunofluorescent assay (IFA)

IFA method was used to screen serum samples for anti-Bb (Borrelia burgdorferi) antibodies by antigen slides as a surrogate antigen[1]. Serums were initially screened at dilutions of 1:128 for IgG antibody and 1:64 for IgM antibody. Titers of positive serum samples were subsequently determined to end point.

Antigen used to test anti-B. burgdorferi s. l. IgG antibodies was prepared from a Chinese human B. garinii isolate, PD91. The PD91 isolate was cultivated in BSK-media at 33°C for a week, harvested and washed in phosphate-buffered saline. Antigen was spotted onto the wells of microtiter slides and fixed with acetone. A titer of ≥1/128 for IgG or 1:64 for IgM was considered positive [23].

Western Blot (WB)

With the positive or equivocal results of IFA test yielded, we conducted both the IgG and IgM Western Blot (WB) assay for further confirmation.

Antigen strips (provided by China CDC, use B.garinii PD91 as an antigen) were put into the seropositive samples by IFA, which diluted with PBS-T by 1:25 for both IgG and IgM, then incubated at room temperature in Orbital Shaker for 4 hours. After five washes of at least 10 min each, the strips then were incubated over night with horseradish peroxidase-conjugated rabbit anti-human IgG (1:1500) and IgM (1:2500) antibodies (Sigma), respectively. After washing for 10 min with PBST, we used TIANGEN Enhanced HRP-DAB Chromogenic Substrate Kit to develop the color to identify the bands. The color development was stopped by purred water when the positive control serum sample reached a defined intensity.

Three pathogenic genotypes which are B.garinii, B. afzelii and Borrelia burgdorferi sensu stricto exist in China. According to their different genomic species, the United States[25-26] and Europe[27] separately established criteria for a positive WB result. In China, criteria for a positive diagnosis of B.garinii were established in 2010 as B.garinii was the predominat genotype[21,28]. It was at least one band of P83/100, P58, P39, OspB, OspA, P30, P28, OspC, P17, P66 and P14 in the IgG test and at least one band of P83/100, P58, P39, OspA, P30, P28, OspC, P17 and P41 in the IgM test[29-30].

Spatial data

We compiled three types of spatial data for each of the 19 counties of Hainan province. These were host data, vector data and socioeconomic ecological data. Here, all data were used in subsequent analyses.

Host data

The wild species list (Muridae) is based on widely used database GBIF(Global Biodiversity Information Facility, https://www.gbif.org/) (Muridae Information see S1 Data). We chose one Familie of rodentia in this study, which is Muridae as the rodents of the family Muridae were the most important reservoir of B. Burgdorferi in China[31-34]. We removed a specimen data Chiropodomys gliroides, which were historically distributed in Hainan but not reported in nowadays. All the rest records have explicit geographic coordinates. We carefully checked geographic and taxonomic accuracy for each species and excluded those species without geographic information.

Vector data

Data on vector (Ixodidae) distribution were derived from a previous study (Ixodidae Information see S2 Data)[35]. We chose Ixodidae as vector in our prediction because most ixodes need more than one host to finish their life cycle. This habitual nature would facilitate the transmission of disease between hosts[36]. We carefully checked geographic accuracy for each species and deleted repetition data. We mapped the distribution of Ixodidae according to the longitude and latitude. We also calculated the Distance Band and mapped the hotspots of Ixodidae in Hainan Province. The methods are same as Muridae hot spot prediction.

Socioeconomic ecological data

Based on the 2019 Hainan province Statistical Yearbook, we collected county-level data on area of cultivated land, number of cattle in stock, number of goat in stock, area of forest, area of grassland, total value of farming, total value of husbandry, and total value of forestry, area of forest, rural population density. Data on land-cover types were retrieved from Resource and Environment Science and Data Center (http://www.resdc.cn/data.aspx?DATAID=335). The land cover data were derived from the 2020 Landsat8 Land Cover product, Land Cover Type Yearly National 1km, which classifies land cover as 6 primary categories and 25 secondary categories (S1 Table). Land cover types were combined and reclassified into 10 land-use categories, which were rice paddy field, dry land, grassland, bush, forest land, wet land, rural area, urban and construction land, unused land and sea (S1 Table).

We generated LUmix layer to demonstrate the area of Muridae and Ixodidae contact. We reclassified the land-use layer which the possible contact area: bush as “9”, grassland as “8”, dry land as “7”, rice paddy field as “6”; the rare contact area: rural area as “5”, forest land as “4”, urban and construction land as “3”, unused land as “2”, while we assigned grid value of impossible contact area-lake and sea -as “1”. In order to display where the farming activity happens on map, we generated raster of farmland (LUfa) using reclassify tool in ArcGIS. As 68% rice paddy fields were three ripe cycle land (wet-wet-dry) in Hainan, also based on 8-hour working/about 40 minutes time spent on the way, we assigned the value of the dryland and rural area grids to “12”, whereas rice paddy field grids were “4” and other land-use type grids were “1”. We also created raster of grassland and bush (LUh) layer, where grassland, bush and rural area grids were reclassified as “12” and other grids were “1”. The raster of forests(LUfo) layer was the reclassified land-use layer where forest land, bush and rural area grids value were assigned as “12” and other grids were “1”.(Land-use Reclassification Remap Table see S2 Table)

Analyses

According to the addresses of suspected cases, we found the latitude and longitude of all 1583 patients we examined. Then we added these data as points into ArcGIS using Display XY Data tool (Fig 1). All layers were using the geographic coordinate system GCS_WGS_1984. It should be noticed that we might need to project all layers using the identical projected system to avoid the distance errors when facing the bigger extent of study area, like country or even continent. As long as the projected system were equal for all layers in risk analysis, using which one doesn’t change location of highly risk areas, but only the visual sizes of them displayed on the map. To visualize the positive rate of Lyme disease in 5 study areas, we typed IFA positive rate into the attribute table as a new field of the county-level vector layer of Hainan (Fig 2A). Then we added WB positives as points into the map using the tool Display XY Data(Fig 2B).

Fig 2

Distribution of IFA- and WB-positive cases in Hainan Province.

(A) Distribution of the IFA-positive rate in our research in Hainan Province. (B) There were a total of 52 confirmed patients among the 94 IFA-positive cases in Hainan.This map was plotted using a combination of ArcGIS software version 10.1 (ESRI, USA) and Photoshop CS 2.0 software (Adobe Systems, USA).

Two primary risk assessment analysis were conducted. The first was about Muridae-Ixodidae interactions, incorporated factors relating to the muridae hotspot, and the Ixodidae hotspot. We then generated a risk map incorporated the anthropogenic factors by combining the layers related to Lyme risk that rural people might face tick. Finally, we calculated the sum of these two risk to evaluate the lyme risk in Hainan province.(Models information see S1–S3 Figs)

To eliminate the interference of dimensional differences on the assessment, the area of farmland (Afa), the intensity of farming activity (Ifa), the raster of farmland (LUfa), the area of grassland (Ah), the intensity of husbandry (Ih), the density of cows (Dcw), the density of goats (Dgt), the raster of grassland and bush (LUh), the intensity of forestry (Ifo), the raster of forests (LUfo), and the densities of rural population (Dpr) were normalized using Fuzzy Membership tool in ArcGIS, respectively. While the area of forests (Afo) cannot be normalized using Fuzzy Membership as the counties with no forests were exported to no data. Therefore, we used their own value divided by the maximum value.

Host hot spot analysis

As research investigation data points (presence data) could not thoroughly reflect the animals’ distribution, we generated the hot spot map of Muridae and Ixodidae to demonstrate the clusters of their distribution considering spatial autocorrelation. We created a Grid feature using Create Fishnet tool. The grid size of this feature was 10×10km, as the range of rodent’s activity were relatively wide driven by food searching (as host movements contribute importantly to tick distribution, especially in lyme disease [37-39], we therefore set the same grid size 10×10km as Muridae in Fishnet Feature). Then we calculate the Muridae and Ixodidae distribution in every fishnet cell using Spatial Join tool, respectively. At the same time, we calculated the Distance Band using Multi-Distance Spatial Cluster Analysis tool (Ripleys K) to the point layer of distribution of Muridae and Ixodidae, respectively. Based on the Spatial Join of Muridae and Ixodidae to fishnet layers, we mapped the hotspots of Muridae () and Ixodidae () by using the Hot Spot Analysis tool to calculate the Getis-Ord Gi* statistic in ArcGIS respectively (details see ArcGIS Resource Center. http://resources.esri.com/help/9.3/ArcGISEngine/java/Gp_ToolRef/Spatial_Statistics_tools/how_hot_spot_analysis_colon_getis_ord_gi_star_spatial_statistics_works.htm). The Getis-Ord General G tool is an inferential statistic and is most appropriate looking for unexpected spatial spikes of high values. We used the largest DiffK value (Distance Band exported, S3 Table) as the appropriate Fixed Distance when running the Hot Spot Analysis, as the largest DiffK value reflecting the distance where spatial processes promoting clustering are most pronounced. where x is the total number of Muridae (Ixodidae in tick Hot Spots Analysis) distribution for each fishnet feature cell j, w is the spatial weight between feature i (the feature class for which Hot Spots Analysis will be performed) and j (Muridae distribution feature, Ixodidae distribution feature when calculating tick hotspots), n is equal to the total number of features and:

Lyme risk from host-tick interactions

Contacts between hosts and vectors are principal pathways of Borrelia burgdorferi transmission [7,8], as well as most of tick-borne diseases. We assessed Lyme risk from potential contact between Muridae and Ixodidae by mapping the distribution hotspots of these two genera.

Contact rate between Muridae and Ixodidae are difficult to quantify and there have been few attempts to do so. While we could assess the risk by overlapping the distribution hotspots of Muridae and Ixodidae, which assuming that interlap of these two genera possible distribution revealed the most risk of contacts. In the grass land and forest area, small mammals and ticks probably come into contact through tick life history. The land-use layer was used to indicate the spatial patterns of these potential contact zones. Therefore, we overlapped the hotspots of Muridae (), the hotspots of Ixodidae () and reclassified land-use layer (LUmix) using Fuzzy Overlay tool in ArcGIS.

To assess the Lyme risk from the contact between Muridae and Ixodidae, we generated the Muridae-Ixodidae layer, in which the risk value of ith cell, Rmix, was calculated as: where is the Getis-Ord Gi* statistic for Muridae Hot Spot within the ith cell and is for Ixodidae Hot Spot, respectively; and LUmix is the potential contact area of Muridae and Ixodidae within the ith cell.

Lyme risk from human-tick interaction

Lyme disease caused by the tick bite is usually happened in bush, grass land or forest area[37]. People who lived in rural area might encounter ticks in their daily life. Due to cultivation, grazing and forestry activity, close contact between ticks and humans can occur on farms, grass land, or forest respectively [40-42]. We carefully classified the kind of human-tick contact in rural, according to people’s career and their workplace. These were farming activity, husbandry and forestry. Corresponding to these activities, we used primary variables to generate three secondary variables: (i) coefficient of farming, (ii) coefficient of husbandry, (iii) coefficient of forestry.

With farming activity, farmland served as human-tick contact place. We used area of farmland to quantify the chance of contact, which assuming that the larger area might lead to the bigger chance of contact. The intensity of farming activity in one place revealed the ratio of the value of farming to the sum value of farming, husbandry and forestry. In order to demonstrate the area where human-tick contact happened when farming activity occur, we combined the Lufa layer in ArcGIS. To eliminate the interferences of dimensional differences on the assessment, the area of farmland, the intensity of farming activity and LUfa layer were normalized using fuzzy Membership Tool in ArcGIS. Then we generate the coefficient of farming activity (Cfa) by combining these three layers using the fuzzy overlay tool in ArcGIS.

To assess the risk from farming activity, we generated the coefficient of farming activity layer, in which the value of ith cell, Cfa, was calculated as: where Afa is the area of farmland within the ith cell and Ifa is the intensity of farming activity within the ith cell, which is the ratio of the total value of farming to the sum value of farming, husbandry and forestry; LUfa is the land-use layer to map where is farming activity happened.

With husbandry, we also calculated the area of grassland and bush [43], and the intensity of husbandry, and reclassified the land use. Moreover, considering the cows and the goats served as the host of ticks [44], we added the density of cows and goats to the equation. We also deal with all 5 layers using Fuzzy Membership, and then we overlapped these layers to calculate the coefficient of husbandry activity (Ch).

To assess the risk from husbandry, we generated the coefficient of husbandry layer, in which the value of ith cell, Ch was calculated as: where Ah is the area of grassland within the ith cell and Ih is the intensity of husbandry within the ith cell, which is the ratio of the total value of husbandry to the sum value of farming, husbandry and forestry; Dcw and Dgt are the density of cows and goats in stock within the ith cell respectively; LUh is the land-use layer to map where the grassland and bush are.

With forestry industry, we also calculated the area of forest and the intensity of forestry, and reclassified the land use. We also used Fuzzy Membership to normalize all these layers, and then we overlapped these layers to calculate the coefficient of forestry activity (Cfo).

To assess the risk from forestry, we generated the coefficient of forestry layer, in which the value of ith cell, Cfo was calculated as: where Afo is the area of forest within the ith cell, containing broadleaved evergreen forest, needle leaved evergreen forest, and Ifo is the intensity of forestry within the ith cell, which is the ratio of the total value of forestry to the sum value of farming, husbandry and forestry; LUfo is the land-use layer to map where the forest are.

Human population density in rural area and distribution of Ixodidae were used to estimate the intensity of these contacts, assuming that higher densities probably result in higher intensities of contacts. In rural area, people will have a greater chance of getting contact with ticks when field operations like farming, husbandry or forestry activities go through.

To assess the Lyme risk from the contact between human and Ixodidae, we generated the human-tick layer by Fuzzy overlaying rural population density layer, sum of Cfa, Ch and Cfo layer and hotspots of ixodidae layer, in which the risk value of ith cell, Rpr, was calculated as: where Dpr is human population density in rural area within the ith cell and Cfa, Ch and Cfo is the coefficient of farming activity, the coefficient of husbandry and the coefficient of forestry activity respectively within the ith cell, and is the Getis-Ord Gi* statistic for Ixodidae Hot Spot within the ith cell.

Lyme risk from human-nature interactions

Host communities are thought to be one of the most essential factors on the tick life cycle[45]. They have deep implications for the circulation of tick-transmitted pathogens, because the relative abundance of potential reservoir hosts may produce large variations in the prevalence of such pathogens[46]. For persistence of lyme borreliosis in the field, a seasonal synchronicity among the questing stages of vectors and the abundance of reservoir hosts is also necessary[47]. The risk to humans of lyme disease is proportional to the risk of contact, which not only depends on human-tick contact, but also on the host-tick contact. Therefore, we demonstrate the lyme risk from human-nature interactions, RR, was calculated as: where Rpr is risk from small mammal-tick interactions within the ith cell and Cfa, Ch and Cfo is risk from human-tick interactions within the ith cell.

Results

Origin of suspected patients

A total of 1583 blood samples were collected from five hospitals and added to a map using ArcGIS according to patients’ addresses. Our samples were distributed mostly in the north and central all across Hainan Island (Fig 1).

IFA test results

A total of 1583 serum samples obtained from patients with arthritis or neurological symptoms were examined for B. burgdorferi using the two-step test. A total of 94 serum samples were found to be positive by IFA (IFA titer, 128 for IgG and 64 for IgM); therefore, an overall infection rate of all five hospitals in Hainan Province was 5.94%. The median age of the 94 patients was 53.17 years (range, 13–86 years) and 57.45% of the patients were male. In all the five hospitals, Qiongzhong People’s Hospital, which is located in the center of Hainan Province, was the area with the greatest prevalence of LD in our investigation, with 39 IFA-positive cases (positive rate, 16.67%). Then, the Third People’s Hospital of Haikou had the second highest prevalence, with a positive rate of 5.88%, followed by Dongfang People’s Hospital, Danzhou People’s Hospital, and Wenchang People’s Hospital with positive rates of 4.62%, 3.36%, and 1.99%, respectively (Table 1).

Table 1

IFA results of the tested serum samples.

Hospital	Serum sample	IFA-positive	Positive rate (%)
Wenchang People’s Hospital	351	7	1.99
Danzhou People’s Hospital	327	11	3.36
Dongfang People’s Hospital	195	9	4.62
Qiongzhong People’s Hospital	234	39	16.67
Third People’s Hospital of Haikou	476	28	5.88
Total	1583	94	5.94

WB test results

All the 94 IFA-positive serum samples were tested by a standardized Western immunoblot, which is known to have high specificity for LD. Of the 94 IFA-positive samples, 52 were identified positive by WB; thus, the average positive rate was 3.28%. The median age of the 52 patients was 53.88 years (range, 17–85 years), and 57.69% of the patients were male. Among all the five hospitals, Qiongzhong People’s Hospital had the highest positive rate of 11.97% according to the two-step testing approach used in our research (28/234) (Table 2).

Table 2

Results of the IFA-positive serum samples tested by WB.

Hospitals	Serum samples	IFA-positive	WB-positive	Two-step positive rate (%)
Wenchang People’s Hospital	351	7	4	1.14
Danzhou People’s Hospital	327	11	6	1.83
Dongfang People’s Hospital	195	9	2	1.03
Qiongzhong People’s Hospital	234	39	28	11.97
Third People’s Hospital of Haikou	476	28	12	2.52
Total	1583	94	94	3.28

LD risk analysis

The calculation of hot spot of Muridae or Ixodidae was based on z-value and p-value in Getis-Ord statistics. A positive and significant z-value indicates spatial clustering of high density of Muridae and Ixodidae (S3 and S4 Datas). On the interaction assessment map of Muridae-Ixodidae (Rmix) in Hainan Province, areas with the highest risk (red cells) were relatively concentrated in the central areas of Hainan, which contained the areas north of Qiongzhong Li and Miao, south of Danzhou, east of Baisha Li, and west of Tunchang(Fig 3A).

Fig 3

Prediction maps (Rmix, Rpr, and RR) of LD risk in Hainan Province.

This map was plotted by a combination of ArcGIS software version 10.1 (Esri, USA) and Photoshop CS 2.0 software (Adobe Systems, USA). The land-use raster was generated from land cover layer (http://www.resdc.cn/data.aspx?DATAID=335).

The human-Ixodidae interaction assessment map (Rpr) revealed that in the influence of anthropogenic factors, the highest risk of LD may occur in the intermediate zone of Hainan, and also in the northeast corner of Hainan, which contained provincial capital- Haikou and Qiongshan District. (Fig 3B).

The combination of Rmix and Rpr, designated RR, revealed the risk of humans acquiring LD from human-nature interactions, revealed two hotspots (hotspot I, II and III)(Fig 3C). Hotspot I was located in the central region of Hainan Province, close to the Limu and Wuzhi mountains where the landcover is primarily composed of bush and forests. Hotspot I covered five counties. Hotspot II was in the south Hainan, which contained Sanya city and the west part of Lingshui. To the north and south of hotspot I, there is a relatively high-risk area corridor (salmon color). Hotspot III occurred in the north of Hainan Province, which is located around Haikou and Qiongshan, and also covered five counties.

Patient distribution and risk analysis

Among the 94 IFA-positive patients, 38 were from Qiongzhong and 25 from Haikou City, and the patients from these two cities comprised 67.02% of our study population (Fig 2A). Among all the 52 WB-positive patients, 28 were confirmed from Qiongzhong People’s Hospital and 12 from the Third People’s Hospital of Haikou (Fig 2B), which had positive rates of 11.97% and 2.52%, respectively.

The prediction of LD (Fig 3C) can be matched with the overall distribution of IFA-positive cases. Qiongzhong with the highest IFA-positive rate; this region is located in hotspot I. Haikou and Qiongshan, which have lower IFA-positive rates, were in hotspot III. 80% cases in Danzhou (8/10) were discovered in Hotspot I, even though Danzhou was the second lowest-positive rate region. The positive rate of Dongfang (which is not in the hotspots of Lyme disease) was lower than that of Haikou and Qiongshan. Finally, Wenchang had the lowest positive rate, and was not in the LD risk hotspots.

The most confirmed patients (42/52, 80.77%) in all the five counties in our analysis were distributed in or around these two hotspots. Among these 42 patients, 30 WB-positive patients were located in/around hotspot I and 12 patients were in hotspot III. Another 10 cases that were WB-positive were distributed in Wenchang (4/10), Dongfang (2/10), Qiongzhong (2/10), and Danzhou (2/10)(Fig 3C).

Discussion

LD has been studied in China for at least 30 years since the 1980s [48-49]. LD research is being carried out in Hainan Province since 2005, and was neglected for hospitals in Hainan for they still cannot do the lyme test. Clinical test of Lyme disease could tell how many Lyme patients were there in Hainan and what were their distributions. In our investigation, there were 94 positive LD cases by IFA, suggesting that 5.94% of patients with arthritis and neurological diseases were suspected to have LD. A total of 52 positive cases were identified by WB, confirming that LD is present in central and north Hainan Province. There were Lyme diseases in all 5 areas in our investigation. Lyme disease mostly distributed in temperate zone, was tested prevalent in tropical area in China. Clinicians should pay more attention of Lyme disease in this area.

According to this survey, the results showed that the infection rate (5.94%, 94 of 1583 serum samples tested positive by IFA) of LD was close to the average positive rate (5.06% [23]) of B. burgdorferi in the Chinese mainland. The age interval was 13–86, and there were 54 men and 40 women. The results of the WB assay verified 52 positive cases, including 30 men and 22 women (17–85 years). The results showed that LD was present in patients with arthritis or neurological symptoms in Hainan, and this disease should be commonly considered by clinicians with patients presenting with these symptoms.

In our study, the positive rate of LD was high in the central regions of Hainan and relatively low in the northern region of Hainan (Fig 2A). Of all the five hospitals included in this research, Qiongzhong People’s Hospital, located in Qiongzhong City in the center of Hainan Province, had the highest positive rates of LD. In our research, we assumed that the LD infection rate of these five hospitals could represent local conditions as all these five hospitals were the typical ones in their respective local areas. Therefore, Qiongzhong had the highest prevalence of LD in Hainan Province in our research. Haikou city, which is located in the north of Hainan Province, came second in terms of the B. burgdorferi positive rate, followed by Danzhou and Dongfang, and Wenchang, which had the lowest prevalence of LD.

The distribution of Muridae, which is the most important reservoir of B. burgdorferi, could lead to a potential LD epidemic in the region[50]. Simultaneously, the interactions between Muridae and Ixodidae could help the antigen persist in its natural environment. The prediction map (Rmix) was overlapped by the hotspot distribution of Muridae and Ixodidae. The overlapping areas of distribution of these two genera facilitate their interactions, leading to the persistence and prevalence of B. burgdorferi.

Although host and vector can influence the distribution of the epidemic area[51], variation in human behavior across different landscapes may partly explain the human infection pattern[52]. An important detail in this study is determining the most probable location where people became infected. Previous research suggests possible relationships between LD risk and human incidence, and identified areas of substantial uncertainty in relationships between tick density and human exposure to infection. Therefore, in risk analysis of the interaction between human and Ixodidae, we classified human behavior into three activities: farming, husbandry, and forestry. Rural people who engage in these activities might encounter ticks in the farmland, grass and bush, and forest areas. We calculated the sum of these risky behaviors to map the interaction between human and Ixodidae. Rpr reflected the risk from human contact with ticks. In Hainan, the midline region from the central to the south and the Haikou-Qiongshan area were found to be associated with the highest risk of LD infection.

The overall Lyme risk map (RR) that result from our analyses revealed high-risk areas for LD in the Wuzhi Mountain (hotspot I), Sanya (hotspot II) and Qiongshan (hotspot III) areas, which results from the convergence of several known risk factors, including distribution hotspot of Muridae and Ixodidae; dense rural populations; area of farmland, grassland and bush, and forest; and a dense population of cows and goats.

The three hotspots that emerged from our analysis differed in their size and overall severity. Hotspot I contained parts of five counties, which were north of Qiongzhong Li and Miao, south of Danzhou, southwest of Chengmai, west of Tunchang, and east of Baisha. Hotspot I was associated with the Wuzhi and Limu mountains, which are characterized by high densities of human, Ixodidea, and Muridae populations. In addition, hotspot I have extensive bush, grass, and forest land, which facilitate interactions between vector and potential hosts, as well as human and vector. Although hotspot II and III lack a population belonging to Muridae and Ixodidae, and feature smaller bush, grass, and forest than hotspot I, they emerged as high priority area due to their high rural population density, intensive farming, husbandry, and forestry activities. Furthermore, hotspot II should be paid more attention to as there was lack of investigation of human Lyme disease. Lyme disease might be a severe problem locally.

The current spatial pattern of Lyme risk factors is due to the interaction of humans and natural systems related to the host, vectors, and humans[53]. Therefore, we reason that it is possible to exert some control over the transmission and prevalence of LD by regulating the interaction between the relevant human and natural systems. Instead of classical statistical methods, a geographically explicit, knowledge-based method was applied here. This method was motivated by the lack of comprehensive LD data and inexplicit of complicated mechanisms of interaction of host-vector-human. Furthermore, this method not only serves as a substitute for the traditional investigation of LD, but it is also an additional tool to map risks, and it may help efforts to contain disease outbreaks and transmission. As our understanding of LD ecology improves, we can introduce new risk factors to improve our predictions of risks in space.

Model Information of Rmix calculations.

(TIF)

Click here for additional data file.

Model Information of Rpr calculations.

(TIF)

Click here for additional data file.

Model Information of RR calculations.

(TIF)

Click here for additional data file.

The attribute of Landcover map and reclassification remap table of Landcover layer.

(DOC)

Click here for additional data file.

Reclassification remap table of Land-use layer.

(DOC)

Click here for additional data file.

The Distance Band results table using Multi-Distance Spatial Cluster Analysis tool (Ripleys K).

(DOC)

Click here for additional data file.

Information of Muridae distribution in all 19 cities.

(XLS)

Click here for additional data file.

Information of Ixodidae distribution in all 19 cities.

(XLS)

Click here for additional data file.

Results of Muridae Hotspot analysis.

(XLS)

Click here for additional data file.

Results of Ixodidae Hotspot analysis.

(XLS)

Click here for additional data file.

26 Feb 2020

Dear Dr Zhang,

Thank you very much for submitting your manuscript "Prevalence of Lyme Disease in Hainan Province" for consideration at PLOS Neglected Tropical Diseases. As with all papers reviewed by the journal, your manuscript was reviewed by members of the editorial board and by several independent reviewers. In light of the reviews (below this email), we would like to invite the resubmission of a significantly-revised version that takes into account the reviewers' comments.

Please carefully address all the reviewer comments.

We cannot make any decision about publication until we have seen the revised manuscript and your response to the reviewers' comments. Your revised manuscript is also likely to be sent to reviewers for further evaluation.

When you are ready to resubmit, please upload the following:

[1] A letter containing a detailed list of your responses to the review comments and a description of the changes you have made in the manuscript. Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

[2] Two versions of the revised manuscript: one with either highlights or tracked changes denoting where the text has been changed; the other a clean version (uploaded as the manuscript file).

Important additional instructions are given below your reviewer comments.

Please prepare and submit your revised manuscript within 60 days. If you anticipate any delay, please let us know the expected resubmission date by replying to this email. Please note that revised manuscripts received after the 60-day due date may require evaluation and peer review similar to newly submitted manuscripts.

Thank you again for your submission. We hope that our editorial process has been constructive so far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Peter J. Krause

Deputy Editor

PLOS Neglected Tropical Diseases

Andrew Azman

Deputy Editor

PLOS Neglected Tropical Diseases

***********************

Please carefully address all the reviewer comments.

Reviewer's Responses to Questions

Key Review Criteria Required for Acceptance?

As you describe the new analyses required for acceptance, please consider the following:

Methods

-Are the objectives of the study clearly articulated with a clear testable hypothesis stated?

-Is the study design appropriate to address the stated objectives?

-Is the population clearly described and appropriate for the hypothesis being tested?

-Is the sample size sufficient to ensure adequate power to address the hypothesis being tested?

-Were correct statistical analysis used to support conclusions?

-Are there concerns about ethical or regulatory requirements being met?

Reviewer #1: See summary and general comments below

Reviewer #2: The authors need explain carefully on how you linked hospital data to your environmental predictors, which is very important to evaluate the performance of your predictive models. The present methods only focused much on the process of testing the disease in the lab. What is even more important is to provide the steps linking these test data with environmental climatic niches and human activities, which are unfortunately lacking here.

Risk factors

There is no any fundamental information on how you selected the 17 cities as your study sample areas. The explicated geographical positions and occurrence of the Lyme disease reporting should also be provided here.

The authors need explain the reason why you chose the three factors here. In addition to human activities and host characteristics, fundamental niches such as climate and microhabitat factors determine the suitability of the disease survival (e.g., Guernier et al. 2004, Liu et al. 2013). Just as what I have suggested on the Introduction section, the authors need strengthen the hypotheses around disease prevalence.

Ref.

Guernier V, Hochberg ME, Guegan JFO. (2004) Ecology drives the worldwide distribution of human diseases. PLoS Biol. 2, 740–746.

Liu X, Rohr JR, Li Y (2013) Climate, vegetation, introduced hosts and trade shape a global wildlife pandemic. Proceedings of the Royal Society B: Biological Sciences, 280: 20122506.

“were no data…” should be revised to “included no data…”

[27] red color?

One major issue in the Data analysis and Result sections is that the authors did not provide any statistical analyses to evaluate the performance of your predictive model. Although the GIS spatial analysis is fine with me, there are at least some approaches such as randomization test to validate the robustness of your predictor selection and overall predictive power of your model.

Fig. 3 please add the place name in the map to facilitate the understanding of spatial distributions of Lyme hotspots.

Reviewer #3: The study makes claims beyond its testable hypothesis. The design is inappropriate to the stated objectives -- or rather, a risk index could be valuable but this one is slapdash. Limitations on the population and sampling distribution are not addressed. No statistical analysis supports the conclusions drawn (see Summary & General Comments response).

--------------------

Results

-Does the analysis presented match the analysis plan?

-Are the results clearly and completely presented?

-Are the figures (Tables, Images) of sufficient quality for clarity?

Reviewer #1: See summary and general comments below

Reviewer #2: I suggest shorting the paragraph on Lyme test, which is not related with your main model analysis. By contrast, the authors should pay more attentions to clarify the relationship between your hospital test and modeling analyses.

Weather should be whether

The paragraph of “as we know” is weak in logic and utility, which should be modified using more clear hypotheses driving the outbreak and prevalence of the Lyme disease. The present writing did not provide readers enough information on the background of factors important to Lyme prevalence in your study area, China and other regions worldwide.

It is better to introduce a series of abiotic, biotic and human activities in determining the disease prevalence. This information is very helpful to represent the three predictors used in the following GIS models. In addition, the authors need strengthen the distributional patterns and crucial predictors of disease in a general way (i.e., not specific to the Lyme but other related neglected tropical bacterial diseases worldwide before talking about the Lyme disease.

As the disease widely distributes across the mainland of China, the following paragraph should stress more on the important implication of conducting Lyme disease risk analyses in Hainan province of China. I am not able to follow the present description only addressing the unique geographical position and climate. The authors need explain more on the unique opportunity of conducting this research in Hainan province, China. Only the lack of related study in Hainan is not a reasonable ground of a novel study. Furthermore, the authors need to provide readers some background knowledge on the differences of disease prevalence between mainland and island, which will make the present work more interesting and general.

Reviewer #3: The results and figures are not presented clearly (see Summary & General Comments response).

--------------------

Conclusions

-Are the conclusions supported by the data presented?

-Are the limitations of analysis clearly described?

-Do the authors discuss how these data can be helpful to advance our understanding of the topic under study?

-Is public health relevance addressed?

Reviewer #1: I'd suggest being explicit about the limitations.

Reviewer #2: The present conclusion is too descriptive and should be strengthened by more important implications based on their main findings.

Reviewer #3: More ambitious conclusions are claimed than can be supported by the data and analysis (see Summary & General Comments response).

--------------------

Editorial and Data Presentation Modifications?

Use this section for editorial suggestions as well as relatively minor modifications of existing data that would enhance clarity. If the only modifications needed are minor and/or editorial, you may wish to recommend “Minor Revision” or “Accept”.

Reviewer #1: See summary and general comments below

Reviewer #2: I have no specific comments on this field.

Reviewer #3: Significant grammar and spelling errors throughout.

--------------------

Summary and General Comments

Use this section to provide overall comments, discuss strengths/weaknesses of the study, novelty, significance, general execution and scholarship. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. If requesting major revision, please articulate the new experiments that are needed.

Reviewer #1: The investigators have undertaken a Lyme seroprevalence study in patients with neurological and/or arthritic symptoms in Hainan Province China. The findings —as reported— suggest modest prevalence

Language, phrasing, grammar, spelling etc.

I strongly suggest review editorial review. Unfortunately, this detracts from the manuscript considerably. Parts of the manuscript are unclear. Areas that warrant attention include —but are not limited to the following—

• Abstract: significant language issues

• “Lyme disease is one of the most important neglected tropical bacterial diseases” is it a tropical disease?

• nuerological symptoms: change “neurological”

• “single serologic test is partially hampered by the occurrence stage of antibodies”

• detecting, leading to insufficient results,

• “Weather there is risk”

• “eaqual”

• “avtivity”

• “As we all know,” would remove non-contributory statements

Methods

• How were the hospitals selected? Include brief description of the size and scope of those hospitals. Similarly, it is important to understand the representativeness of the sampling approach. How was the sample size determined?

• Need more detail about the selection/inclusion criteria i.e. “arthritis and nervous system disease”? How long had they been symptomatic? Any further clinical detail e.g. history of rash? Duration of symptoms also impacts interpretation of the results

Discussion

• Provide more information on local tick vectors and reservoir hosts

• Add a limitations section

• Could provide a few sentences on the clinical/public health implications of Lyme disease, if left neglected

• Try to contextualize the prevalence findings by offering a few sentences on the regional and/or international estimates

• Is there evidence of other tick-borne illnesses in Hainan? Have other agents been investigated locally and/or in China e.g. Babesia, anaplasma, Borrelia miyamotoi?

• Resolution of Fig 1 map needs to be improved

Tables and Figures

• Figures need legends

• Supplementary table needs a legend and explanation

Reviewer #2: Zhang et al. used 1,583 serum samples across 5 hospitals of Hainan province, China to explore the prevalence and potential predictors of Lyme disease. The topic fits well within the scope of PNTD, and the sampling effort at a province scale is impressive. In addition, Hainan province is very unique in its biodiversity and biogeography in China, and thus understanding the prevalence and the risks of important neglected tropical bacterial diseases is very timely and crucial for the prevention and control of the disease. However, my major concerns are that the current manuscript is too descriptive and should be strengthened especially for the hypothesis and additional data analyses to validate the performance of your predictive models. In addition, the scientific writing could be improved with an aid of a native speaker.

Reviewer #3: Software and risk factors

paragraph 1: More needs to be said about “all data were entered into ArcGIS as raster layers.” One almost never creates raster layers from scratch. Were they derived from satellite imagery? Converted from vector data? What does “entered” mean here?

paragraph 2: The authors likely mean “land cover” instead of “land use.” The source of the land cover data should be provided, as should a source or method of generating the “tick species distribution” layer. A description of what went into the “Tick Active Area” and “Tick Risk Area” layers is needed, as is some rationale supporting the reclassification into 9 classes. Why, for example, do the authors consider farmland a greater risk than bush? Why does seaside wetland appear to be rated higher risk than forest? It is probably not necessary to list the number of pixels containing each land cover type.

paragraph 3: “As tick questing ask for a certain height.” is a typo? Last sentence is virtually a duplicate of the last sentence of paragraph 2 and is unnecessary.

Paragraph 4: The reclassification scheme could be better conveyed using a remap table. What made the authors decide that “more tick species could increase the risk of lyme disease” – as opposed to greater tick population or greater chance of human exposure? Should be supported with citation to past research if this is true.

Analyses

This section needs considerable expansion, perhaps to include any meaningful spatial analysis at all. How was the Tabulate Area output used and interpreted? What were the grounds for deciding that those were high risk areas? What was the means of inference to other areas? Are “cities” the polygonal regions shown in the map?

Figure 1

This is certainly not a map of 1583 individuals. There are about 200 points in the map, and they are not well-distributed throughout the island. Perhaps the large point symbols obscure multiple individuals, in which case a smaller symbol should be used. Also some reference to the underlying population distribution would be helpful. Why are the areas with nearly no points so sparse – does no one live there? Also a description of the process used to geocode addresses is needed.

Figure 2 – A

Because of sampling bias, this map could easily be interpreted merely as a map of which districts have a testing hospital in them. The legend is badly done. It is misleading to show 6 decimal places on numbers with so little precision. The legend should be more explicit as to the quantity represented by the color ramp (it’s the IFA positive rate). Why map this instead of the 2-step positives?

Patient Distribution and Risk Analysis

Paragraph 2: “According to the survey, there were 12 species ticks in Qiongzhong (Supplemental material, S1 Table ), which is the area with most tick species in Hainan.” According to the S1 Table, there were also 12 tick species in Ledong and 10 in Sanya, both of which have no positive cases (and nearly no samples). How do the authors account for these counter examples to their argument (probably sampling bias)? Would they expect high values here if they could now acquire samples? Perhaps that would be way to test their theory.

“In all 5 areas in our research, the spread of Lyme diseases were basically matched with the tick distribution.” The word “spread” implies an expansion over time, which is unsupported by the rest of the work. A map of the tick distribution (the number of species present in each district?) might help.

Figure 3

There does not seem to be much predictive about Figure 3. We see the same 9 classes described above . . . are we just seeing land cover with numeric labels? If not, what went into this index? Is it an average of the rather arbitrary scores assigned to land cover, the reclassified number of tick species per district and . . . . what? Still don’t know at this point what “Tick Active Area” and “Tick Risk” are.

Nothing in the paper supports the drawing of those arrows, as if the authors intend to show past or future change in the spatial pattern of the disease.

Discussion

paragraph 6: Similarly, there is nothing to support the verb “radiated” implying spread over time “The risk of lyme disease in Hainan was radiated from Wuzhi Mountain in the east, south and west directions . . . ”

“In the northwest of Wuzhi Mountain, there is Limu Mountain which hampered the distribution of lyme disease. “ If mountain barriers are this effective, perhaps they should be included in the risk index.

--------------------

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

Figure Files:

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org.

Data Requirements:

Please note that, as a condition of publication, PLOS' data policy requires that you make available all data used to draw the conclusions outlined in your manuscript. Data must be deposited in an appropriate repository, included within the body of the manuscript, or uploaded as supporting information. This includes all numerical values that were used to generate graphs, histograms etc.. For an example see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5.

Reproducibility:

To enhance the reproducibility of your results, PLOS recommends that you deposit laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see https://journals.plos.org/plosntds/s/submission-guidelines#loc-methods

Submitted filename: review.pdf

Click here for additional data file.

6 Aug 2020

Submitted filename: Responses.doc

Click here for additional data file.

8 Sep 2020

Dear Dr. Hao,

Thank you very much for submitting your manuscript "Prevalence and prediction of Lyme Disease in Hainan Province" for consideration at PLOS Neglected Tropical Diseases. As with all papers reviewed by the journal, your manuscript was reviewed by members of the editorial board and by several independent reviewers. In light of the reviews (below this email), we would like to invite the resubmission of a significantly-revised version that takes into account the reviewers' comments.

We cannot make any decision about publication until we have seen the revised manuscript and your response to the reviewers' comments. Your revised manuscript is also likely to be sent to reviewers for further evaluation.

When you are ready to resubmit, please upload the following:

[1] A letter containing a detailed list of your responses to the review comments and a description of the changes you have made in the manuscript. Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

[2] Two versions of the revised manuscript: one with either highlights or tracked changes denoting where the text has been changed; the other a clean version (uploaded as the manuscript file).

Important additional instructions are given below your reviewer comments.

Please prepare and submit your revised manuscript within 60 days. If you anticipate any delay, please let us know the expected resubmission date by replying to this email. Please note that revised manuscripts received after the 60-day due date may require evaluation and peer review similar to newly submitted manuscripts.

Thank you again for your submission. We hope that our editorial process has been constructive so far, and we welcome your feedback at any time. Please don't hesitate to contact us if you have any questions or comments.

Sincerely,

Peter J. Krause

Deputy Editor

PLOS Neglected Tropical Diseases

Andrew Azman

Deputy Editor

PLOS Neglected Tropical Diseases

***********************

Reviewer's Responses to Questions

Key Review Criteria Required for Acceptance?

As you describe the new analyses required for acceptance, please consider the following:

Methods

-Are the objectives of the study clearly articulated with a clear testable hypothesis stated?

-Is the study design appropriate to address the stated objectives?

-Is the population clearly described and appropriate for the hypothesis being tested?

-Is the sample size sufficient to ensure adequate power to address the hypothesis being tested?

-Were correct statistical analysis used to support conclusions?

-Are there concerns about ethical or regulatory requirements being met?

Reviewer #2: The revised method section is much more improved. Especially it has been much clearly for the part of spatial analyses. My final suggestion is that the authors need explain a bit more on the screening process of the host distribution data from GBIF. GBIF database includes occurrences in suspicious locations (e.g., on the grounds of a museum, the epicenter of a city, etc.), which are needed to remove. These records are usually incomplete sightings/specimens that are re-labelled later by people who don’t know where the sighting/collection took place.

Reviewer #3: The revised analysis is more sound than the original, but some methodological ambiguities and weaknesses remain. The objectives are articulated and the study design is appropriate, but details of the execution may lead to inaccurate results.

-- One flaw remaining in the spatial analysis portion of the revised paper is the source of the land use data. The authors cite Wikipedia and claim it was downloaded from there. I am unable to find any land use raster at the link provided, and Wikipedia is generally not in the business of doing land use classification of satellite imagery. If the authors found a dataset there, it must have come with a reference to a primary source. This matters for two reasons: for confidence in the risk analysis, it can be helpful to know how the land uses were derived and what categories were contemplated, and it is essential to know the cell size of the original to evaluate how much resampling was needed in the 10km grid analysis (for example in the computation of Cfa, Ch, and Cfo). Land use is usually derived from land cover, and categories like “sparse woods, seaside wet lands, slope grassland” sound more like land cover. The authors might consider a description of the land use / land cover data (but applicable to whatever they in fact used) like “The land use data were derived from the 2020 MODIS Terra+Aqua Combined Land Cover product, Land Cover Type Yearly L3 Global 500m, June 2020, which classifies land cover according to the International Geosphere-Biosphere Programme 17-class scheme. Land cover types were combined and reclassified into [some number of] land use categories following [some citation to the work used to inform their land use classification].”

— The separate components of the LUmix layer are clear enough, but include the method of their final combination into one layer. A map of the LUmix layer would be good to include, as it is an important component of the spatial analysis.

— The authors make a somewhat light argument in passing for the use of a 10km grid for the rodents (searching for food), but they don’t defend its use for ticks. Is it a reasonable grid size on which to see tick clustering and habitat extent? Some supporting citation to ecological literature on gridding species ranges would be valuable here.

— The hotspot analysis seems fairly sound, but the authors might explain what hot spots/Getis G contributes that their original spatial join count of observations to grid cells does not. Why is the degree of spatial clustering more informative than just high incidence in this case? Similarly, they should explain why Fuzzy Membership is used instead of, say, simply converting to z-scores for commensurability.

-- It would be more informative to the user to identify maximum DiffK value produced by the Ripley’s K analysis, maybe even reproduce the output table. Some explication of the meaning or interpretation of the Getis-Ord statistic, beyond the basic formula, would be useful. Also cite the ESRI documentation from which this section derives, and perhaps include the output so the reader can determine if statistically significant clusters were found at this resolution. The authors appear to be using a particular (but valid) special case of the Getis-Ord Hotspot Analysis, in which the spatial clustering is solely location-based, not weighted by the distribution of any particular attribute. Therefore, "where xj is the attribute value of Muridae" is probably not accurate. “[H]ow you construct the Analysis Field determines the types of questions you can ask. Are you most interested in determining where you have lots of incidents, or where high/low values for a particular attribute cluster spatially? If so, run Hot Spot Analysis on the raw values or raw incident counts. . . . Alternatively (or in addition), you may be interested in locating areas with unexpectedly high values in relation to some other variable. “ http://resources.esri.com/help/9.3/ArcGISEngine/java/Gp_ToolRef/Spatial_Statistics_tools/how_hot_spot_analysis_colon_getis_ord_gi_star_spatial_statistics_works.htm

— It is essential to run the Ripley’s K and Hotspot Analysis in a local projected coordinate system. The authors should disclose the projection in which they conducted their analysis.

— Rmixi =Gm*Gix*LUmix, The authors should consider and explain the effect of negative values and zeroes in this multiplication formula. What happens when Gm is close to -1 (rodent dispersion, not clustering) and Gix is close to -1 (tick dispersion)? A misleading high positive value is produced. (On review of the supporting material, it looks like the authors used Fuzzy Overlay to make this combination, but (1) why? and (2) which? Fuzzy Product sounds problematic in the ESRI documentation and (3) how would Fuzzy Overlay handle the concern about negative values?)

— The authors should consult and cite some resources on designing risk analysis formulae.

— A source is needed for the human population density raster. Also consider that the most densely populated areas (urban) are probably at the least risk.

— “Dpri is risk from small mamal-tick interactions within the ith cell” This seems like a typo. Dpr was population density, and Rpr was a risk index that didn’t involve the mammals (rodent).

-- It would be interesting, but probably not essential, to include some sort of model validation, formalize how or whether the predicted risk aligned with the discovered cases.

--------------------

Results

-Does the analysis presented match the analysis plan?

-Are the results clearly and completely presented?

-Are the figures (Tables, Images) of sufficient quality for clarity?

Reviewer #2: The revised result section is clear and not overstated.

Reviewer #3: The figures are NOT of sufficient quality for clarity. Values in map legends and labels on maps are too blurry to read.

--------------------

Conclusions

-Are the conclusions supported by the data presented?

-Are the limitations of analysis clearly described?

-Do the authors discuss how these data can be helpful to advance our understanding of the topic under study?

-Is public health relevance addressed?

Reviewer #2: The conclusions are supported by the results.

Reviewer #3: The conclusions are improved in scope. They appropriately describe the risk of encounter and its relationship to the observed incidence of disease. If the methodology can be strengthened to give better confidence in the derived results, this should be an informative paper.

--------------------

Editorial and Data Presentation Modifications?

Use this section for editorial suggestions as well as relatively minor modifications of existing data that would enhance clarity. If the only modifications needed are minor and/or editorial, you may wish to recommend “Minor Revision” or “Accept”.

Reviewer #2: I think the manuscript is ready for acceptance after minor revision.

Reviewer #3: The paper is still rife with English grammar errors and needs to be proofread.

"In rural area, people get to contact with ticks most likely happened when farming, husbandry or forestry activities go through."

It's also possible that some of the more grammatical and stylistically unique sentences are taken verbatim from other works and should be encased in quotation marks, e.g.

"For persistence of lyme borreliosis in the field, a seasonal synchronicity among the questing stages of vectors and the abundance of reservoir hosts is also necessary."

--------------------

Summary and General Comments

Use this section to provide overall comments, discuss strengths/weaknesses of the study, novelty, significance, general execution and scholarship. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. If requesting major revision, please articulate the new experiments that are needed.

Reviewer #2: I have no more comments here.

Reviewer #3: I don't think new experiments are needed per se, just some serious strengthening of the execution and documentation of the existing ones. Perhaps the authors might reconsider their multiplicative risk indices based on further review of the literature.

--------------------

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

Reviewer #3: No

Figure Files:

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org.

Data Requirements:

Please note that, as a condition of publication, PLOS' data policy requires that you make available all data used to draw the conclusions outlined in your manuscript. Data must be deposited in an appropriate repository, included within the body of the manuscript, or uploaded as supporting information. This includes all numerical values that were used to generate graphs, histograms etc.. For an example see here: http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001908#s5.

Reproducibility:

To enhance the reproducibility of your results, PLOS recommends that you deposit laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see https://journals.plos.org/plosntds/s/submission-guidelines#loc-methods

18 Dec 2020

Submitted filename: Response Letter.doc

Click here for additional data file.

18 Jan 2021

Dear Dr. Hao,

We are pleased to inform you that your manuscript 'Prevalence and prediction of Lyme Disease in Hainan Province' has been provisionally accepted for publication in PLOS Neglected Tropical Diseases.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. A member of our team will be in touch with a set of requests.

Please note that your manuscript will not be scheduled for publication until you have made the required changes, so a swift response is appreciated.

IMPORTANT: The editorial review process is now complete. PLOS will only permit corrections to spelling, formatting or significant scientific errors from this point onwards. Requests for major changes, or any which affect the scientific understanding of your work, will cause delays to the publication date of your manuscript.

Should you, your institution's press office or the journal office choose to press release your paper, you will automatically be opted out of early publication. We ask that you notify us now if you or your institution is planning to press release the article. All press must be co-ordinated with PLOS.

Thank you again for supporting Open Access publishing; we are looking forward to publishing your work in PLOS Neglected Tropical Diseases.

Best regards,

Peter J. Krause

Deputy Editor

PLOS Neglected Tropical Diseases

Andrew Azman

Deputy Editor

PLOS Neglected Tropical Diseases

***********************************************************

Change "the tropical area" to "tropical areas" in the Background of the Abstract.

Consider reviewer #3 comments.

Consider including one map of China to show Hainan Province in relation to mainland China

Reviewer's Responses to Questions

Key Review Criteria Required for Acceptance?

As you describe the new analyses required for acceptance, please consider the following:

Methods

-Are the objectives of the study clearly articulated with a clear testable hypothesis stated?

-Is the study design appropriate to address the stated objectives?

-Is the population clearly described and appropriate for the hypothesis being tested?

-Is the sample size sufficient to ensure adequate power to address the hypothesis being tested?

-Were correct statistical analysis used to support conclusions?

-Are there concerns about ethical or regulatory requirements being met?

Reviewer #2: The revised methods are clear and robust.

Reviewer #3: The methods of the study are appropriate and have been extensively clarified. One remaining minor criticism is the choice of coordinate system: GCS_WGS_1984 is a geographic coordinate system, not a projected one designed to preserve distance or area. The small extent of the study area makes it unlikely that distance errors would be too substantial, but some geographic dilettantes might dismiss the hotspot analysis entirely based on this detail. I don't think it invalidates the work, but the authors might want to add a comment that indicates they are aware of the issue and made a conscious choice (and why).

**********

Results

-Does the analysis presented match the analysis plan?

-Are the results clearly and completely presented?

-Are the figures (Tables, Images) of sufficient quality for clarity?

Reviewer #2: The revised results are clear.

Reviewer #3: The results are consistent with the analysis. In my pdf reader, the map text was still illegible despite the authors' efforts to increase the resolution.

**********

Conclusions

-Are the conclusions supported by the data presented?

-Are the limitations of analysis clearly described?

-Do the authors discuss how these data can be helpful to advance our understanding of the topic under study?

-Is public health relevance addressed?

Reviewer #2: The revised conclusions are supported by the data.

Reviewer #3: The authors have done a good job in keeping their conclusions modest and consistent with the results. They have made explicit the sources of uncertainty in their analysis.

**********

Editorial and Data Presentation Modifications?

Use this section for editorial suggestions as well as relatively minor modifications of existing data that would enhance clarity. If the only modifications needed are minor and/or editorial, you may wish to recommend “Minor Revision” or “Accept”.

Reviewer #2: I suggest acceptance.

Reviewer #3: The authors might change the NoData color in Fig. 2 Map A, because it is difficult to discern the difference in the two darkest greens. Perhaps NoData could be symbolized in an unrelated hue like gray. There are still some errors in written English, but they are not bad enough to impair communication.

**********

Summary and General Comments

Use this section to provide overall comments, discuss strengths/weaknesses of the study, novelty, significance, general execution and scholarship. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. If requesting major revision, please articulate the new experiments that are needed.

Reviewer #2: The authors have conducted intensive revisions on the manuscript. After careful reading, I find the revised analysis is clear and robust. The scientific writing has also been greatly improved. The previous reviewers' comments have been well addressed and I do not have further comments and questions. I think this is an important and timely topic, and is suitable for publication in PLOS NTDs.

Reviewer #3: (No Response)

**********

PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

Reviewer #3: No

14 Mar 2021

Dear Dr. Hao,

We are delighted to inform you that your manuscript, "Prevalence and prediction of Lyme disease in Hainan province," has been formally accepted for publication in PLOS Neglected Tropical Diseases.

We have now passed your article onto the PLOS Production Department who will complete the rest of the publication process. All authors will receive a confirmation email upon publication.

The corresponding author will soon be receiving a typeset proof for review, to ensure errors have not been introduced during production. Please review the PDF proof of your manuscript carefully, as this is the last chance to correct any scientific or type-setting errors. Please note that major changes, or those which affect the scientific understanding of the work, will likely cause delays to the publication date of your manuscript. Note: Proofs for Front Matter articles (Editorial, Viewpoint, Symposium, Review, etc...) are generated on a different schedule and may not be made available as quickly.

Soon after your final files are uploaded, the early version of your manuscript will be published online unless you opted out of this process. The date of the early version will be your article's publication date. The final article will be published to the same URL, and all versions of the paper will be accessible to readers.

Thank you again for supporting open-access publishing; we are looking forward to publishing your work in PLOS Neglected Tropical Diseases.

Best regards,

Shaden Kamhawi

co-Editor-in-Chief

PLOS Neglected Tropical Diseases

Paul Brindley

co-Editor-in-Chief

PLOS Neglected Tropical Diseases