The influence of enhanced vegetation index, proximity to national borders, proximity to protected areas, and proximity to water on malaria case prevalence in Sub-Saharan Africa, 2000-2020

INTRODUCTION

In the last two decades, malaria remains a major global public health challenge, especially in low- and middle-income countries.¹ Over 210 million estimated cases of malaria were reported in 2015 with over 450,000 deaths worldwide.^2,3 The number of malaria cases rose to about 300 million in 2018 and in 2019, the estimated malaria cases from 87 countries were nearly 230 million.^3,4

Globally, there was a decline in the number of deaths that resulted from malaria cases from about 730, 000 in 2000 to over 400,000 in 2019.^3,5 Out of the 430,000 deaths resulting from malaria in 2017; Sub-Saharan Africa (SSA) accounted for about 90% of the global malaria deaths with about 260,000 being children under five years of age.^2,6 The burden of malaria cases among children under five years differs across the countries in SSA. More than 30% and 10% of children under five mortalities in Nigeria and Tanzania respectively were a result of malaria cases alone and it is the second largest contributor to childhood morbidity and mortality.^2,5,7 Ethiopia is amongst the countries with the highest under-five mortality in SSA, with most cities and villages being malaria endemic.^8,4 Malaria infection is said to be more prevalent in rural areas than in urban centers.⁹

Malaria was believed to be a rural disease because the transmitting vectors are said to breed more in rural areas.⁹ However, malaria had remained a serious public health concern in urban areas as observed by.¹⁰ Unfortunately, this is not the situation for most African countries with limited resources to provide adequate infrastructure amenities that cope with the rate of urbanization experienced, resulting in poor housing, sanitation, and, drainage systems, which could increase the vector breeding and human contact.⁹ Research scientists have attributed the prevalence of malaria in SSA to several factors including medical conditions, seasonal influences, age, gender, pregnancy, socioeconomic status, and demographic and environmental factors.^4,11

METHODS

Data source

This study involved data that were pooled from the Demographic and Health Surveys (DHS) conducted in 32 Sub-Saharan African countries spanning 2000 to 2020 with a total population of 736,487. These countries included Ghana, Nigeria, Kenya, Senegal, Tanzania, Burkina Faso, and Zambia. The DHS data are generated from household surveys conducted in over 90 countries every three to five years. These surveys are nationally representative and primarily provide data for monitoring and impact assessment of population, health, and nutrition indicators for individual countries as well as for cross-country comparative analyses. These surveys use standard data definitions and data collection procedures across countries.

Geospatial covariates

The following geospatial covariates were obtained from the Spatial Data Repository under the DHS program (https://spatialdata.dhsprogram.com/covariates/):enhanced vegetation index, proximity to national borders, proximity to protected areas, proximity to water.

Sampling design

The demographic and health survey was based on a multi-stage stratified cluster sampling. The countries were first stratified by regions and counties and urban and rural areas. Sample enumeration areas (EA’s) were selected independently in each stratum in two stages. In the first stage of obtaining the samples, EA’s (residential households) were selected with probability proportional to the EA size (number of residential households) and with independent selection of each of the sampling stratum. The EA sizes were obtained from the immediate past population and housing census of each country. The sampling frames for the survey were obtained by conducting household listing operations in all the selected EAs, and the resulting lists of households served as a sampling frame for the selection of households in the second stage. In the second stage of the selection, a fixed number of households per cluster (EAs) were selected with an equal probability systematic selection from the newly created household listing. The trained interviewers visited and interviewed only the selected households.

Statistical methods

The study employed five stages of pooling the data together. Stage 1 involved appending all kids datasets from different countries for different survey periods. Stage 2 involved appending all environmental datasets for different survey periods for different countries. In stage 3, the kids’ dataset and environmental datasets were merged. Stage 4 involved merging datasets obtained in Stage 3 with women 14 to 49 population dataset. In the final stage (i.e., stage 5), the dataset obtained from stage 4 was merged with women sample dataset. The final data analyses were carried out using the final merged dataset obtained in stage 5.

Outcome variables

The study investigated malaria case prevalence in SSA. Children who had fever for the past two weeks were used as proxies for malaria cases and classified as such.

Primary exposure variables

The primary exposure variables involved in the analyses were enhanced vegetation index, proximity to protected areas, proximity to national borders, and proximity to water.

Enhanced vegetation index

This measures the density of green leaves within a specified geographic region.

Proximity to national borders

This measures how close people in a geographical area are to a national border. It is a measure of straight-line distance to the nearest international border.

Proximity to protected areas (meters)

This is a straight-line distance to the nearest protected area.

Proximity to water (meters)

This is a straight-line distance to the nearest major water body.

All the primary exposure variables were grouped into 4 quantiles (i.e., 1^st, 2^nd, 3^rd, and 4^th quantiles). The 1^st quantile of the enhanced vegetation represents low density of green leaves in a particular geographic area whiles 4^th quantile represents high density of green leaves in a particular geographic area. The 1st quantile of proximity to national borders represents how close households are to national or international borders and the 4th quantile represent how far households are to national or international borders. The 1^st quantile of proximity to protected area represents how close households are to the nearest protected area and the 4^th quantile represents how far households are to the nearest protected area. Similarly, the 1^st quantile of proximity to water represents how close households are to the nearest major water body and the 4^th quantile represent how far households are to the nearest major water body.

The countries were also grouped into four regions for descriptive analysis as follows: East Africa, West Africa, Central Africa, and Southern Africa. Angola, Madagascar, Kenya, Rwanda, Tanzania, and Uganda were grouped as East Africa. West Africa group was made up of Benin, Burkina Faso, Ivory Coast, Gambia, Ghana, Guinea, Liberia, Mali, Nigeria, Senegal, and Togo. The Central Africa group was made up of Burundi, Cameroun, Chad, The Democratic Republic of Congo, and Gabon. Estwani, Lesotho, Malawi, Mozambique, Namibia, South Africa, Zambia, and Zimbabwe were also grouped as Southern Africa.

Potential confounders

Factors such as age of household head, place of residence, household utility, wealth index, main floor material, multiple births, sex of child (male or female), mothers age at first birth, parity, mothers’ highest educational level, currently using contraceptive, total children ever born, place of delivery, currently breastfeeding, type of delivery, number in co-wives, Antenatal Care (ANC) visits, household ownership of bed net, sanitary facility, wealth index, main floor material and properties (i.e. electricity, water, car, and motorbike), have been identified to independently influence malaria prevalence.^12,13 Other studies have linked environmental factors such as rainfall to malaria.¹⁴ These factors were grouped into household characteristics, child-related characteristics, maternal factors, and environmental factors (see supplementary file).

Statistical analyses

The analysis was restricted to women who had children in the last 5 years preceding the survey. The sample contains the total number of children under-five born in the 5 years preceding the survey, and the data contain information on their respective mothers and households. Such data include place of residence, household wealth, and household size.

Although all statistical analyses for survey data adjusted for weighting, clustering, and stratification at the individual analysis, traditional regression models assume that both geospatial and non-geospatial covariates are identically and independently distributed over the geographical area. The study employed three-staged statistical analyses. In the first stage, the women’s standard weights were denormalized to obtain one pooled dataset for all the demographic and health surveys (2000-2020) since the study pooled data from different DHS data from different countries at different survey periods in SSA. The second stage involved bivariate analyses of potential confounders. Simple logistic regression was used to determine potential confounders for malaria case prevalence. Thirdly, each covariate’s effect was assessed by fitting a logistic regression model that includes the primary exposure variables adjusted for potential confounders and the country-fixed effect. Two different logistic regression models were fitted to assess the multivariable effect of the primary exposure variables on malaria prevalence. Model 1 assessed the effect of only primary exposure variables on malaria prevalence and model two assessed the effect of primary exposure variables adjusted for maternal factors, child-related factors, household factors, and environmental factors on malaria prevalence.

Mean and standard deviation was reported for continuous variables whiles proportions were reported for categorical variables. All statistical analyses were conducted using Stata 14 (StataCorp, College Station, Texas, USA) and p-value less than 0.05 were considered statistically significant. The final interpretation of the results was based on the adjusted models, however, in some instances, references were made to the unadjusted models to highlight those interesting findings.

RESULTS

Descriptive statistics

The average age of the head of household was approximately 41±12.9 years (range = 12-98), and females headed 17.3% of households with an average household size of 2.1±0.8. The percentage of households in rural areas was 69%. Male and female children born in the 5 years preceding the survey have almost equal percentages (male = 50.6%, female = 49.4%). The prevalence of multiple births was 3.5% and 54.8% of women had 4 or more Antenatal Care (ANC) visits. The percentage of women who did not deliver in a recommended facility (home delivery) was 47.5% and about 4% of all delivery were conducted through caesarian section. Fifty-one percent (51%) of the women breastfed their children after 1 hour after delivery.

Approximately 60% of households have mosquito bed nets for sleeping. The percentage of children under-five who did not sleep under mosquito bed net the night preceding the survey was 59.7%. Thirty-three percent (33%) of respondents had no formal education and 3% completed higher level of education. The percentage of households that belonged to the poorest category of wealth index was 22.9%, 20.1% belonged to the middle category and 16.2% belonged to the richest category of the wealth index.

DHS Demographic and Health Survey

West Africa recorded the highest temperature (27.4^oC±1.1). This was followed by Central Africa (24.6^oC±2.6) with Southern Africa recording the least temperature (21.7^oC±2.4). East Africa recorded 1124.6mm±363 of rainfall, 1217.2mm±635.8 of rainfall was recorded in West Africa, and 846.4mm±263.1 of rainfall was recorded in Southern Africa over the period (2000-2020) (Table 2).

The description of the data source used for the study can be found in Table 1.

The influence of enhanced vegetation index, proximity to national borders, proximity to protected areas, and proximity to water on malaria cases prevalence in SSA from 2000 to 2020

Table 3 shows the factors that influence malaria case prevalence in SSA from 2000 to 2020. Enhanced vegetation index has a significant association with malaria case prevalence. In the unadjusted model, children belonging to households with higher vegetation indices have significant risks of suffering from malaria. Children in households in the 3^rd quantile have a 15% higher risk of suffering from malaria [COR=1.15, 95%CI: 1.09-1.22; p value<0.001] compared to children belonging to households with lower vegetation index (1^st quantile) and children in households in the 4^th quantile have 34% higher risk of suffering malaria [COR=1.34, 95%CI: 1.27-1.41; p-value<0.001] compared to children belonging to households with lower vegetation index (1st quantile). After adjusting for other covariates, only children in households belonging to the 3^rd quantile of enhanced vegetation index have a significant risk of suffering from malaria. The risk of children suffering malaria in this household is 27% less compared to children that belong to households that belong to the 1^st quantile of vegetation index [aOR=0.73, 95%CI: 0.55-0.99; p-value=0.050].

OR Odds Ratio; aOR Adjusted Odds Ratio; NB: significant p-values are in bold

Among households that are closer to national borders (i.e., those in the 2^nd quantile), the risk of children suffering malaria is 9% higher compared to children who belong to households that are very close to national borders (1^st quantile) [COR=1.09, 95%CI: 1.02-1.15; p-value=0.005] in the crude analysis. However, after adjusting for potential confounders, the risk of suffering malaria between children of households in the 2^nd quantile and children of households in the 1^st quantile is the same [aOR=1.00, 95%CI: 0.81-1.14; p-value=0.974].

In the unadjusted model, children who belong to households that are not close to protected areas (3^rd and 4th quantiles) have significant risks have suffering malaria. Those in the 3^rd quantile have 13% less risk compared to households in the 1^st quantile [COR=0.87, 95%CI: 0.83-0.92; p-value<0.001], and children belonging to households in the 4^th quantile have 10% less risk of suffering malaria compared to households in the 1^st quantile [COR=0.90, 95%CI: 0.85-0.94; p-value <0.001]. After adjusting for other covariates, the risks were almost the same as previously determined in the unadjusted model, however, these estimates were no longer significant; children belonging to households in the 3^rd quantile have 14% less risk of suffering malaria compared to children belonging to households in the 1^st quantile [aOR=0.86, 95%CI: 0.68-1.10; p=0.243] and those belonging to household in the 4^th quantile have 18% less risk of suffering malaria compared to children belonging to households in the 1^st quantile [aOR=0.82, 95%CI: 0.62-1.07; p-value=0.145].

Proximity to water is significantly associated with malaria case prevalence. In the unadjusted model, children belonging to households that are not closer to water (i.e., those belonging to 3^rd) have 10% less risk of suffering malaria compared to children in households closer to water (1^st quantile) [COR=0.90, 95%CI; 0.85-0.95; p-value<0.001] and those belonging to the 4^th quantile (i.e., far from water) have 17% higher risk of suffering malaria compared to households in the 1^st quantile [COR=1.17, 95%CI=1.10-1.24; p-value<0.001]. After adjusting for potential confounders, only households in the 4^th quantile remained significant. Children belonging to this household, have 33% higher risk of suffering malaria compared to households that are closer to water [aOR=1.33, 95%CI:1.08-1.63; p-value=0.007].

DISCUSSION

This study sought to determine the factors that influenced malaria case prevalence in sub-Saharan Africa from 2000 to 2020. We were specifically interested in the role environmental factors such as enhanced vegetation index, proximity to water, proximity to national borders, and proximity to protected areas played in increasing or decreasing malaria prevalence in sub-Saharan Africa. This objective was achieved by modeling the joint effect of these environmental factors using the logistic regression approach. This is critical to understanding other environmental factors that promote or prevent malaria infections among households and communities in sub-Saharan Africa.

The prevalence of malaria cases was 23.1%. The plausible reason for this high prevalence could be that some households may not have enough ITNs for all the children under five, and therefore priority is given to the youngest among all the children. Another explanation that could be given is that information regarding malaria prevention may not be properly understood by the households because it may not be culturally sensitive and may not be based on already existing positive beliefs and behaviour.

Enhanced vegetation index was found to significantly influence malaria morbidity in SSA. The findings suggest that an enhanced vegetation index tends to negatively influence malaria morbidity in SSA. For instance, people located in areas with higher vegetation index have 27% less risk of recording malaria cases. This finding is interesting as opposed to the findings of other studies that suggest that higher vegetation index is positively associated with malaria cases.^15,16 The present finding that suggests a negative association could be because the plants located in the area may have some mosquito repellant properties as identified by.¹⁷ This would mean that as such plants increased in population, the whole geographical area will be mosquito free and this may result in low malaria morbidity.

The study found higher odds of malaria infection among households who live far from major water bodies. For example, those who live close to major water bodies have 22% higher risk of malaria infection compared to those who live closer to major water bodies though not significant. This odds increased to 1.33 among those who live far from major water bodies compared to those who live closer to major water bodies. This is consistent with other studies that found presence of water bodies to be significantly associated with malaria transmission in Africa.¹⁸ However, the interesting thing is that whiles this present study found that the longer the distance from major water bodies, the higher the risk of malaria infection, other studies have reported that the longer the distance from major water bodies, the less the risk of malaria infection.^19,20 Further studies in the form of evidence synthesis are however necessary to establish whether proximity to major water bodies results in a higher or lower risk of malaria infection.

Proximity to national borders was significantly associated with malaria prevalence in the absence of confounders. When the final analysis was adjusted for confounders, those who live far from national borders have a 3% reduced risk of being diagnosed with malaria compared to those who lived closer to national borders, though not significant. This is contrary to similar studies done in Africa and elsewhere. For example, a study conducted in The Gambia that sought to map malaria transmission risk using satellite imagery showed that malaria case prevalence was high among residents of villages that are closer to The Gambia- Senegal border compared to villages that were far away from the border.²¹ Other publications have also indicated that the increasingly high incidence of malaria along the Indian national borders is creating real challenges in eliminating malaria in those areas.²² The plausible explanation for this contradiction could be that the period within which data was collected for one study may be the peak season for malaria cases at the national and international borders although this theory seems unlikely. Further studies are thus needed to establish whether proximity to national borders increases or decreases malaria case prevalence.

Contrary to other studies, this study showed no significant influence of proximity to protected areas on malaria prevalence even though those who live far from protected areas have an 18% reduced risk of being diagnosed with of malaria. Even though not enough studies have been conducted on the effect of closeness to protected areas on malaria prevalence, some publications available suggest that proximity to protected area may be used to predict the rates of malaria especially in areas where the disease exist. Other studies have actually determined the influence of protected areas on malaria case prevalence, and the results from these studies are contrary to each other.^23–25 Whiles one study reports that proximity to protected area leads to a reduced risk of malaria infection others have reported otherwise.²³ Some of the protected areas that are forest reserves may have some reserves species with mosquito-repellant properties that led to reduce risk of malaria infection.¹⁷ While other protected areas that are forest reserves may have ornamental plants such bromeliads which can serve as a good breeding place for mosquitoes leading to a higher risk of malaria infection.²⁶

One main strength of the study is that the survey is nationwide and population-based with internationally approved survey methods. The procedures used to carry out the survey combined with large samples drawn nationally allow for the findings to be generalized to the population of children under-five in SSA, as well as other similar population worldwide. Sampling weights were adjusted for in the analysis, making the estimates more reliable and representative of the entire population. Despite these strengths, the study has some limitations which should be considered in interpreting the results. The data used in the study was based on cross-sectional survey design and cannot be used to conclude causative effect between the outcome and predictors. Just as with all other studies, this study did not account for all the factors that could predict malaria case prevalence among children under-five.

CONCLUSION

This study has a direct bearing on achieving SDG Target 3 which relates to a reduction in malaria incidence and malaria mortality rate to at least 90% each and to end the epidemic of malaria in at least 35 countries by 2030. The study has highlighted environmental factors as predictors of malaria incidence and demonstrated the utmost need to include these factors in the fight against malaria in SSA. Intervention should be targeted at households that are closer to water with more children under five to ensure full access and use of ITNs among all children under five as part of the overall goal of achieving health-related SGDs. Also, the contradictory findings relating to proximity to water, proximity to protected areas and proximity to national borders warrant further studies such as evidence synthesis to establish the actual effect of these environmental factors on malaria case prevalence.

ACKNOWLEDGEMENTS

We would like to thank Demographic and Health Surveys Program for making the data available for this study.

Funding: No funding sources

Conflict of interest: None declared

Ethical approval: The study was approved by the Institutional Ethics Committee

REFERENCES

1. 1. Bennett A, Bisanzio D, Yukich JO, Mappin B, Fergus CA, Lynch M, et al. Population coverage of artemisinin-based combination treatment in children younger than 5 years with fever and Plasmodium falciparum infection in Africa, 2003-2015: a modelling study using data from national surveys. Lancet Glob Health. 2017;5(4):e418–27.
2. 2. Dawaki S, Al-Mekhlafi HM, Ithoi I, Ibrahim J, Atroosh WM, Abdulsalam AM, et al. Is Nigeria winning the battle against malaria? Prevalence, risk factors and KAP assessment among Hausa communities in Kano State. Malar J. 2016;15(1):351.
3. 3. WHO. World malaria report 2020: 20 years of global progress and challenges. 2020. Available at: https://www.who.int/teams/global-malaria-programme/reports/world-malaria-report-2020. Accessed on 25 August 2022.
4. 4. Aychiluhm SB, Gelaye KA, Angaw DA, Dagne GA, Tadesse AW, Abera A, et al. Determinants of malaria among under-five children in Ethiopia: Bayesian multilevel analysis. BMC Public Health. 2020;20(1):1468.
5. 5. WHO. World Malaria Report, 2014. Available at: https://www.bing.com/ck/a?!&&p=0cd24d5ee5c9a90bJmltdHM9MTY5NDczNjAwMCZpZ3VpZD0wYWU3MzkxMC04YzdjLTYyZGItM2I3Yy0yYmNiOGRjZTYzMzcmaW5zaWQ9NTIwNA&ptn=3&hsh=3&fclid=0ae73910-8c7c-62db-3b7c-2bcb8dce6337&psq=World+Malaria+Report+2014.&u=a1aHR0cHM6Ly93d3cud2hvLmludC9wdWJsaWNhdGlvbnMtZGV0YWlsLXJlZGlyZWN0Lzk3ODkyNDE1NjQ4MzA&ntb=1. Accessed on 19 September 2022.
6. 6. UNICEF. Malaria in Africa, 2020. Available at: https://data.unicef.org/topic/child-health/malaria/. Accessed on 26 October 2022.
7. 7. Adinan J, Damian DJ, Mosha NR, Mboya IB, Mamseri R, Msuya SE. Individual and contextual factors associated with appropriate healthcare seeking behavior among febrile children in Tanzania. Jhaveri R, editor. PLOS ONE. 2017;12(4):e0175446.
8. 8. Alene M, Yismaw L, Berelie Y, Kassie B. Health care utilization for common childhood illnesses in rural parts of Ethiopia: evidence from the 2016 Ethiopian demographic and health survey. BMC Public Health. 2019;19(1):57.
9. 9. Alemu A, Tsegaye W, Golassa L, Abebe G. Urban malaria and associated risk factors in Jimma town, south-west Ethiopia. Malar J. 2011;10(1):173.
10. 10. Baragatti M, Fournet F, Henry MC, Assi S, Ouedraogo H, Rogier C, et al. Social and environmental malaria risk factors in urban areas of Ouagadougou, Burkina Faso. Malar J. 200;8(1):13.
11. 11. Bassey S, Izah S. Some determinant factors of malaria prevalence in Nigeria. J Mosquito Res. 2017;(31):7.
12. 12. Dwomoh D, Agyabeng K, Agbeshie K, Incoom G, Nortey P, Yawson A, et al. Impact evaluation of the free maternal healthcare policy on the risk of neonatal and infant deaths in four sub-Saharan African countries: a quasi-experimental design with propensity score Kernel matching and difference in differences analysis. BMJ Open. 2020;10(5):e033356.
13. 13. Obasohan PE, Walters SJ, Jacques R, Khatab K. A Scoping Review of Selected Studies on Predictor Variables Associated with the Malaria Status among Children under Five Years in Sub-Saharan Africa. Int J Environ Res Public Health. 2021;18(4):2119.
14. 14. Tottrup C, Tersbol BP, Lindeboom W, Meyrowitsch D. Putting child mortality on a map: towards an understanding of inequity in health. Trop Med Int Health. 2009;14(6):653-62.
15. 15. Adeola AM, Botai JO, Mukarugwiza Olwoch J, De W. Rautenbach HCJ, Adisa OM, De Jager C, et al. Predicting malaria cases using remotely sensed environmental variables in Nkomazi, South Africa. Geospatial Health. 2019;14(1).
16. 16. Amadi JA, Olago DO, Ong’amo GO, Oriaso SO, Nanyingi M, Nyamongo IK, et al. Sensitivity of vegetation to climate variability and its implications for malaria risk in Baringo, Kenya. Shaman J, editor. PLOS ONE. 2018;13(7):e0199357.
17. 17. Tisgratog R, Sanguanpong U, Grieco JP, Ngoen-Kluan R, Chareonviriyaphap T. Plants traditionally used as mosquito repellents and the implication for their use in vector control. Acta Trop. 2016;157:136-44.
18. 18. Kibret S, McCartney M, Lautze J, Nhamo L, Yan G. The impact of large and small dams on malaria transmission in four basins in Africa. Sci Rep. 2021;11(1):13355.
19. 19. Kibret S, Lautze J, McCartney M, Wilson GG, Nhamo L. Malaria impact of large dams in sub-Saharan Africa: maps, estimates and predictions. Malar J. 2015;14(1):339.
20. 20. Mangani C, Frake AN, Chipula G, Mkwaila W, Kakota T, Mambo I, et al. Proximity of residence to irrigation determines malaria risk and anopheles abundance at an irrigated agroecosystem in Malawi. Am J Trop Med Hyg. 2022;106(1):283–92.
21. 21. Bøgh C, Pinder M, Jawara M, Dean A, Thomas CJ, Clarke SE, et al. High spatial resolution mapping of malaria transmission risk in the gambia, west africa, using landsat tm satellite imagery. Am J Trop Med Hyg. 2007;76(5):875-81.
22. 22. severemalaria.org. Malaria and migration in India. Available at: https://www.severemalaria.org/countries/india/malaria-and-migration-in-india. Accessed 6 June 2023.
23. 23. Taber ED, Smithwick EAH. Influence of protected areas on malaria prevalence in Sub-Saharan Africa. Appl Geogr. 2015;64:35–45.
24. 24. Do Protected Areas Have an Effect on Malaria Prevalence?. Available at: https://www.geographyrealm.com/do-protected-areas-have-an-effect-on-malariaprevalence/#:~:text=Likewise%2C%20proximity%20to%20protected%20areas,absent%20at%20a%20given%20site. Accessed 6 June 2023.
25. 25. Valle D, Clark J. Conservation efforts may increase malaria burden in the Brazilian Amazon. Eisele T, editor. PLoS ONE. 2013;8(3):e57519.
26. 26. Multini LC, Wilke ABB, Marrelli MT. Neotropical Anopheles (Kerteszia) mosquitoes associated with bromeliad-malaria transmission in a changing world. Acta Trop. 2020;205:105413.