Potential impacts of climate change on dengue fever distribution using RCP scenarios in China

FAN Jing-Chun,LIU Qi-Yong*

a School of Public Health,Gansu University of Chinese Medicine,Lanzhou,730000,China

b State Key Laboratory of Infectious Disease Prevention and Control,Collaborative Innovation Center for Diagnosis and Treatment of Infectious Diseases,National Institute for Communicable Disease Control and Prevention,Chinese Center for Disease Control and Prevention(China CDC),Beijing,102206,China

c Shandong University Climate Change and Health Center,School of Public Health,Shandong University,Jinan,250012,China

AbstractThisstudy projected denguedistribution risk map using representativeconcentration pathways(RCP2.6,RCP4.5,RCP6.0,RCP8.5)in China in 2020s,2030s,2050s and 2100s.Based on the biological characteristics of Aedes albopictus and the dengue epidemic process,dengue transmission biological model was developed to project the risk epidemic areas.Observational temperature data in 1981-2016 at 740 stations and grid data of 0.5°×0.5°(15°ˉ55.5°N,70°ˉ140.5°E)under selected RCPs in 2020s,2030s,2050s and 2100s were used.Relative to 142 counties and 168 million people living in the projected high risk areaof dengue in theclimatecondition of 1981-2016,dengue high risk areasin Chinawould expand under same RCPscenariosin the21st century with timepast except RCP2.6 with aturning down point in 2050s.Especially under RCP8.5 which global mean temperature would increase by 4.9°C till 2100s,the high risk area and population for dengue transmission would expand additional 34 counties(20 million)in 2020s,114 counties(60 million)in 2030s,208 counties(160 million)in 2050 and 456 counties(490 million)in 2100srespectively than thoseof 1981-2016.For RCP8.5 in 2100s,thepopulation and expanded high risk areaswould increase4.2-fold and 2.9-fold than the 1981-2016 mean.Thenewly added high risk areasshould preparefor controlling and preventing dengue in different period according to projected dengue risk map.

Keywords:Dengue fever;Climate change;Aedes albopictus;Representative concentration pathways(RCPs);Risk distribution

1.Introduction

Being the fastest spreading mosquito-borne disease in the world,dengue fever(DF)distribution range has expanded remarkably and the case incidence has also multiplied 30-fold since 1960s(A~nez et al.,2017).DF is a febrile disease that affects infants,children,and adults,with symptoms appearing 3-14 d after the infective bite.Currently,no effective vaccine or any specif ic medicine is available to treat dengue.According to the World Health Organization(WHO)estimation,30.0%-54.7%of theworld'spopulation(2.05-3.74 billion)is living in areaswheredenguevirusescan betransmitted(Brady et al.,2012).DF is a vector-borne disease and susceptible to climate variables,as the growth and development of mosquitoes are inf luenced by factors such as temperature and humidity(Bhatt et al.,2013).Climate change could have a direct effect on the vector's survival condition,and the extreme weather caused by climate change may have an indirect effect on the outbreak of DF.Previous studies have indicated that changes in duration and patterns of dengue transmission may result in a potential increase in the latitudinal and altitudinal range of dengue outbreaks(Col′on-Gonz′alez et al.,2018).WHO hasidentif ied f ivemajor health consequences of climate change,one among which was that changing temperaturesand patternsof rainfall would alter thegeographical distribution of insect vectors that spread infectious diseases.Of these diseases,malaria and dengue are of the greatest public health concern(WHO,2008).

Dengue outbreaks have been recorded almost every year in China since f irst outbreak in Foshan city,Guangdong province,in 1978,and have further been subsequently recorded in other provinces,including Hainan,Guangxi,Fujian,Zhejiang,Jiangsu,Yunnan,and Henan.Aedes aegypti and Aedes albopictus are the two major vectors responsible for the spread of DF;however,DF in China is primarily attributed to A.albopictus(Guo et al.,2018).From RCP2.6 to RCP8.5,the global average temperature will increase by 1.5-4.9°C from 2010 till 2100;however,under RCP2.6,before 2100,the temperature will reach a peak by midcentury and then decline(Wayne,2013;IPCC,2015).Due to the predicted climate warming in the future,a major concern is the potentially expanded distribution of dengue in China.DF is inf luenced by several factors such as meteorological variables,vegetation and geological features,and man-made environments,for example,urbanization trends and human movement patterns(Messina et al.,2015).However,temperature is the most signif icant variable that affects dengue distribution under climate change(Friedrich,2018).

It is reported that a higher temperature would signif icantly increasethedomestic dengueoutbreaksrisk in Koreabased on RCPscenarios(Lee et al.,2018).Similar resultswere reported in coastal Kenya,air temperature increase under climate changewould oppose effects on malaria directly and indirectly(Le et al.,2019).

Another research from China demonstrated that temperature and precipitation would strongly drive dengue outbreaks in Guangzhou(Xu et al.,2017).However,there was a few researches about the distribution of dengue risk in future China under climate change.In this study,we used temperature data and the regional model under RCP scenarios to project the dengue distribution risk in 2020s,2030s,2050s,and 2100s.The results would provide the scientif ic basic to control and prevent dengue in different areas in different years.

2.Materials and methods

2.1.Digital map and temperature data

A digital map of China(scale:1:1,000,000)was obtained from the Chinese State Bureau of Surveying and Mapping.The National Climate Center provided observational temperature data in 1981-2016 at 740 stationsand the data from ECHAM5 model of 0.5°×0.5°(15-55.5°N,70-140.5°E)under RCP2.6,RCP4.5,RCP6.0 and RCP8.5 in 2020s,2030s,2050s,and 2100s across China.Interpolation by Kriging Technique in the ArcGIS 10.0 software was used to convert area temperature data at the county level to raster temperature data.

2.2.Dengue case data

To construct the national dengue risk map,the DF and dengue hemorrhagic fever cases at the county level from 2001 to 2016 in China were extracted from communicable disease information systemsin the Chinese Center for Disease Control and Prevention(China CDC).The cases from 1990 to 2000 in Guangdong province provided by Guangdong CDCwere used to test the reliability of the model.

2.3.Biological model

2.3.1.Thecritical suitabletemperaturefor DF transmission

In the Ross-Macdonald model(Macdonald,1956),the prerequisite for DF transmission is that the virus'infectious lifespan(Li)is at least 1 d.Theoretically,mosquitoes need to survive for more than 1 d after being infected to transmit pathogens,and temperature is a critical factor for the extinctive incubation period and also the infectious lifespan.Within the range of suitable temperature,a higher temperature is associated with reduced extinctive incubation period.According to Watts et al.(1987),Eq.(1)can be used to calculate the incubation period of dengue virus within the mosquito.

Where K and C are the minimum accumulated temperature(165.2°C?d)and the minimum developed temperature(11.9°C)(Yu et al.,2005)that the dengue virus to develop within the mosquito,T is the actual temperature,n is the extrinsic incubation period,respectively.

In general,the daily survival rate of A.albopictus is reported as88%(Hawley,1998)and 95%(Lacroix et al.,2009).Almeida(Almeida et al.,2005)investigated the bioecology of A.albopictus(Diptera:Culicidae)in Macao,China,and calculated that itsdaily survival rate ranged from 91%to 97%,and 91.2%further applied this range to estimate other biological parameters.In this study,we used 88%,91%,and 95%as the daily survival rates of A.albopictus to calculate the infectious lifespan Li(Eq.(2))at different temperatures(Table 1).In Eq.(2),p is the daily survival rates of A.albopictus and n is the extrinsic incubation period.

According to the Ross-Macdonald model,the prerequisite of the virus'infectious lifespan is at least 1 d for DF transmission,theinfectiouslifespan is1.0 d when thedaily survival rate is 91%and the temperature is 18.5°C.Therefore,we assumed that the minimum temperature suitable for dengue transmission is 18.5°C,which is similar to the result reported by Lambrechts(Lambrechts et al.,2011).

Table 1Infectious lifespan(unit:d)of an infected mosquito at different temperatures and survival rates.

2.3.2.The duration of dengue epidemic process

Indigenous dengue cases are always triggered by imported cases under favorable conditions in China(Cheng et al.,2016).Wecalculated thedurationof thedengueepidemic processasthe time period between the f irst imported dengue case who was bitten by a mosquito and the appearance of the f irst indigenous case.Thistimeperiod wasdivided intotwophases,theextinctive incubation period(3-15 d,7 d in general)and the internal incubationperiod(3-15d,7d ingeneral);thus,thedurationof the dengueepidemic processis14 d in average(Fig.1).

2.3.3.Potential transmission index(PTI)of dengue

We assumed the minimum temperature required for Aedes mosquito development as T0and the daily mean temperature as Tmean.The accumulated degree-days(ADD)was calculated as the difference between Tmeanand C(11.9°C)summed over the extinctive incubation period of dengue virus within the mosquito(ADDDV)or the development period for A.albopictus(ADDa.a.).The values of ADD can be calculated according to Eq.(3)and Eq.(4),the interval of ADDDVis from the f irst day of dengue virusdeveloped in the mosquito till the mosquito has infectious capacity;the interval of ADDa.a.is the period that A.albopictus develop from an egg till an adult mosquito.

According to Hawley et al.(1989)the literature,the minimum temperature required for Aedes mosquito development(T0)is 11°C,the minimum accumulated temperature required for A.albopictus development(ADDa.a.)is 980°C?d(Kobayashi et al.,2002).ADDDVis the same as K in Eq.(1),165.2°C?d,and C is 11.9°C.

For each observing meteorological station i,the ADDDFand ADDa.a.were calculated using the respective daily mean temperature data(Tmean(i))or the predictive temperature increases(Tp(i))according to Eq.(5)and Eq.(6)(Zhou et al.,2008).

Fig.1.The duration of the dengue epidemic process.

According to Eq.(1),we found that the extinctive incubation period of dengue virus in the mosquito is 12.5 d at 25°C and 36.7 d at 16.4°C,as the lifespan of A.albopictus is generally 30-40 d.Therefore,themaximum duration required for A.albopictus development at the minimum temperature is 30 d,and the maximum duration required for dengue virus development is also 30 d.

The PTI was calculated at each of the 740 meteorological stations(i)in different years,according to Eq.(7).Only those PTI values above 1 were considered where DF transmission could potentially occur.

The PTI values for dengue virus and A.albopictus were calculated by Eq.(8)and Eq.(9),respectively.

2.4.DF risk mapping

The ArcGIS 10.0 software was used for mapping and analysis.First,we worked out the daily average temperature for each meteorological station for 1981-2016 and projected the daily averagetemperature under RCP2.6,RCP4.5,RCP6.0,and RCP8.5 in 2020s,2030s,2050s,and 2100s.In case the daily average temperature was≥18.5°C in consecutive 14 d,we conf irmed that the area was suitable for DF transmission in these 14 d and imported the data into the GIS software and obtained the dengue transmission suitable temperature map.Second,we worked out the PTI(i)of each station,and if PTI(i)≥1,weconsidered that denguecould potentially occur in that area and imported the data into the GIS software and obtained the dengue PTI map.Third,by taking into account the dengue transmission suitable temperature and the PTI,we obtained the ultimate dengue risk map.According to the previous epidemic situation of DFand thedistribution of A.albopictus in China,in general,during January to March,no indigenous case was reported from Guangdong,which wasa high epidemic area in China;there was also no indigenous case reported in Zhejiang province from November to next April.We def ined the time period of more than 9 months suitable for dengue transmission ashigh risk,5-8 monthsasmoderaterisk,1-4 months as low risk,and 0 as no risk.

2.5.Reliability test for forecast

Kappa agreement was used to test the reliability between the actual distribution and the projection.We collected the indigenous dengue case records in Guangdong province in 1990-2016 and constructed the GIS-based geographic distribution map of indigenous dengue casesat the municipal level.We also constructed the projected dengue risk map at the municipal level in Guangdong using thetemperature data from 1981 to 2016.The results demonstrated that there were indigenous cases from 18 cities in actual distribution and 17 citiesin the projection.According to Landisand Koch(1977),when the Kappa statistic ranges from 0.61 to 0.80,there is substantial agreement.By calculating the Kappa coeff icient,we obtained the agreement rate as 92.7%and the Kappa coeff icient as 0.71.The results revealed a high agreement between the projection for dengue transmission and the actual indigenous dengue case in Guangdong.

3.Results

3.1.Dengue risk in current China

According to the population density in 2010,there were 168 million people living in 142 high-risk counties of dengue transmission under temperature of 1981-2016 mean,as shown in Fig.2.The dengue risk levels in the projected map roughly agreed with actual situation,the high risk areas in the projected map weretropical climatezone,such as Guangdong,Yunnan,Guangxi,Fujian,Hainan and Taiwan,which were the general epidemic provinces in China.The areas with none dengue risk centered in the high latitude areas with low annual mean temperature,such as Xinjiang,Tibet,Inner Mongolia,Heilongjiang,etc.

3.2.Projected dengue risk under RCP scenarios

3.2.1.The secular trend of RCP2.6

For RCP2.6,the high risk area(population)would increase from 146 counties(172 million)in 2020sto 344 counties(278 million)in 2050s.But after 2050s there would be a turning point,the high risk area would decrease from 344 counties in 2050s to 277 counties in 2100s,so would the high risk population from 278 million to 233 million(Figs.3 and 4).

3.2.2.The secular trend of dengue distribution in different scenarios at the same period

Comparing with the climate condition of 1981-2016,the distribution range of dengue would increase in same period from RCP2.6 to RCP4.5,RCP6.0 and RCP8.5,which was positively correlated with temperature rise.In the short-term projection,such as 2020s,the expanded counties of dengue high risk areas were 21 in RCP4.5 and 26 in RCP8.5.The expanded trend of denguein 2020swasnot distinct in different scenarios.In the long-term projection,such as 2100s,the expanded areas(population)would be 135 additional counties(186 million population)in RCP4.5 to 456 new counties(335 million population)in RCP8.5.The distribution areas and populations would be 4.2-fold and 2.9-fold of 142 counties/168 million population in denguehigh risk map in 1981-2016 respectively increase under RCP8.5 in 2100(Figs.3 and 4).

3.2.3.The secular trend of dengue distribution at different period in the same scenarios

For different period in the same scenario,the dengue high risk area would also expand from 2020s to 2100s in comparison to the climate condition of 1981-2016 except RCP2.6.The high riks areas of dengue in RCP 4.5 in 2020s were 163 counties(188 million population)while 418 counties(354 million)in 2100s.For RCP8.5,the increase of dengue high risk area would be 176 counties(192 million)in 2020s and 598 counties(493 million)in the 2100s.Under RCP8.5,the expanded areas and population in 2100s would be 3.4 times and 2.6 times of those in 2020s.

4.Discussion

Fig.2.The projected dengue risk map under the temperature of 1981-2016 mean.

Fig. 3. The projected dengue risk map under RCPs in China.

It is now widely accepted that climate change has occurred and it may affect our health through a range of pathways(Shindell et al.,2012).In this study,we developed the biological model for dengue transmission both considering the biological character of A.albopictus,the main vector of DFin China,and the transmission process to explore the changes of distribution risk map of potential dengue transmission in future China in RCPs.Historical evidence conf irms the associations between climatic conditionsand vector-bornediseases(Cabrera et al.,2018).Our resultsthusconf irmed the evidence that a signif icant expansion of potential dengue risk regions would occur in thefuture Chinawith climatechange.Therank of DF risk under the current condition in our projected map was consistent with the geographical distribution of dengue incidence in China(Xu et al.,2017).With climate and time change,the distribution of the potential dengue transmission areain Chinawould expand from south to north and shift from low-latitude areas to high latitude areas.These results were conf irmed based on evidence shown by Sun et al.(2017)who reported that thenumber of dengue caseshasincreased and the affected areas have expanded in recent years in China.These results are also consistent with studies from other countries such as the U.S.(Butterworth et al.,2017),Tanzania(Mweya et al.,2016),and Korea(Lee et al.,2018).Temperature is the decisive factor for dengue transmission.From present research results,the expanded trends of dengue distribution were almost same in different countries all over the world.With climate and time change,the distribution of the potential dengue transmission area in China would expand from south to north and shift from low-latitude areas to high latitude areas.These results were conf irmed based on evidence shown by Sun et al.(2017)who reported that the number of dengue cases has increased and the affected areas have expanded in recent years in China.These results are also consistent with studies from other countries such as the U.S.(Butterworth et al.,2017),Tanzania(Mweya et al.,2016),and Korea(Lee et al.,2018).Temperature is the decisive factor for dengue transmission.From present research results,the expanded trends of dengue distribution were almost same in different countries all over the world.

So there was no one left belonging to the house but the landlord s daughter, who was a good, well-meaning girl, and had taken no part in all the evil doings

Fig.4.The high risk counties and population(million)under different RCPs and years in China.

Severalmethodshavebeenusedtoprojectdenguerisk,suchas machine learning in China(Guo et al.,2017),mechanistic virus transmission model in Australia(Williams et al.,2016),and estimation of dengue epidemic potential in Europe(Liu-Helmersson et al.,2016).Climate and the dengue pathogen were the two perspectives highlighted in these methods.Demographic datawereincluded inother methodstopredictdengue outbreaksin Mexico(S′anchez-Gonz′alezetal.,2018).Thisaspect would beconsidered inour further researchinthefuture.Dengue transmission is affected by comprehensive factors,including environmental and social elements.The problem with these methodsisall thefactorscould notbeinvolved in,our methodsis no exception.However,we still calculated high-risk population based on the data of 6th national population census as supplementary.The results from different methods were used in different areas ref lected the similar outcome,that was dengue would expand withclimatechange.Our study resultssuggestthat the government and the health care department should immediately establishanearly warning systemand improvetheadaptive capacity to control and prevent dengue in different areas in different yearsaccording to our projection,which issolid scientif ic evidence.

An interesting result in our study is there was a turning point occurred 2050s under RCP2.6,that is the high risk counties were 344 counties in 2050s but 277 counties in 2100s under RCP2.6.It is also consistent with RCP2.6 scenario is cumulative emissions of greenhouse gases from 2010 to 2100 would reduce by 70%compared to a baseline scenario(van Vuuren et al.,2011)and there would be a temperature peak in 2050s and decline after that.

A limitation of our study is that it emphasizes the role of temperature but does not take into account other meteorological variables such as humidity,rainfall,and the potential interaction between them.It is diff icult to state whether our results are conservative or whether these additional effects might further amplify the extent of changes projected on the basis of temperature alone.

Another limitation of our study is that we used the passive surveillance data in 1990-2010,which might have introduced measurementand information biases.Thisisbecausethesedata included only the laboratory-diagnosed cases,and the individualswhowereinfected with denguevirusbut experienced only subclinical symptomsand unapparent infections and who did not seek medical treatment might have been underreported(Fan et al.,2014).Hence,we might underestimate the dengue risk when we used the passivesurveillance data.

We have provided the risk maps that were based on suitable durationof denguetransmission.Butthereisstillslightdifference of dengue outbreak risks at the same risk degree.Therefore,a quantif ied analysison the relationship between temperature and dengueand relativerisksmaps(futuredengueincidence/current dengue incidence)for China would be our future research direction.Theseearly projectionscould allow for a proper implementation of targeted public health interventions and valuable buffer time for preventing subsequent large-scale epidemics of dengue locally in different areas in different years(Shang et al.,2010),which issignif icant for developing the strategic plansto control and prevent denguein China.

5.Conclusions

We developed a dengue transmission biological model taking into account thebiological propertiesof thevector and pathogen to project a dengue risk map in future China.In summary,with climate change,from the current day till 2100,the distribution of the potential dengue transmission area in China would expand from south to north and shift from low latitude areas to high latitude areas with climate and time.Under same RCPscenario but different period,the distribution of dengue risk area would expand except from 2050s to 2100s under RCP2.6,the high risk area decreased.In same period,the expanded range increased under different RCP scenario which was positively correlated with global warming in the future.Therefore,the government and the department of disease control and prevention should reduce emissions while strengthening the adaptive capacity and implement targeted measures at different times and in different regions.

Conf lict of interest

The authors declare no conf lict of interest.

Acknowledgments

This study was supported by the National Basic Research Programme of China(2012CB955504)and the National Key Research and Development Project“Biological Security Key Technology Research and Development”Special Funds(2016YFC1200802).