Four online bibliographic databases were searched: PubMed, Scopus, Cochrane Library, and Web of Science. We developed a comprehensive search strategy based on the keywords related to malaria parasite species, human mobility, and molecular/genomic tools used to map the spatial spread of malaria infection. This review was restricted to studies reported in English, without considering any time limitation, and included literature up to the end of November 2024. Full details of the search strategy are available in Supplementary Appendix 1. Furthermore, reference lists of included articles were screened to identify relevant studies that were not found through our initial search.

Articles from online bibliographic databases were stored and combined as EndNote files, where duplicate articles were removed. The titles and abstracts of articles were screened. Abstracts were excluded if they focused exclusively on the clinical aspects of malaria infections, addressed infections other than malaria, were written in languages other than English, or were not relevant to genomic surveillance of malaria. Articles selected for full-text review were independently assessed by two reviewers for final inclusion, and any disagreements were resolved through discussion and consensus. Studies were included if they were peer-reviewed articles, used parasite genetic data, focused on either Plasmodium vivax or Plasmodium falciparum, and included an analytical framework that investigated the origin of infection or explicitly analyzed connections between geographical areas, either at a global or regional scale. Articles were excluded if they only descriptively reported the genetic marker without integrating them into an analytical framework, or if they addressed unrelated topics, as detailed in the PRISMA flowchart (Figure 1).

Data extraction was done from the full text of all eligible articles. Data related to study setting, species of malaria parasites, sample size, and type of genetic markers were extracted. Furthermore, the available data on drug resistance markers, travel surveys, and GPS tracking were recorded for each study. The analytical methods used for analyzing the parasite genetic data were also collected.

Data were synthesized narratively, and descriptive statistics were used to summarize the main study characteristics, such as study setting, sample size, and the number of genetic loci examined. Studies were grouped according to the following three criteria: (1) species of malaria parasites (Plasmodium vivax or Plasmodium falciparum); (2) type of genetic marker (Microsatellite, Single Nucleotide Polymorphism, Whole-Genome Sequencing, or Targeted Sequencing) and (3) analytical approach (genetic diversity, population structure, and relatedness among populations or individuals). The findings were presented and discussed descriptively, categorizing them based on the type of genetic marker used for genotyping and the analytical approach employed to determine the origin of infection. Studies were then categorized by geographical region (continent) and publication timeline. This approach was selected to offer a clear overview of how methods have evolved over time in specific regions, highlighting the number of genetic markers used and changes in study scope.

The eligible articles for this review were published between 2007 and 2024. Only one study was published between 2007 and 2012 (19). However, from 2013 to 2024, there was an increase in the number of publications across all continents (Figure 2). This indicated a rising interest in applying genetic methods to distinguish between indigenous and imported cases, which is particularly important for countries approaching zero malaria cases (20). The highest number of publications were from Asia and South America, each with ten, followed by Africa with nine (Table 1). In contrast, Central America and Oceania had the fewest publications, with just two each. The sample size varied between 45 (21) to 8,654 subjects (22). There were 19 studies focused on P. falciparum and 18 on P. vivax. Additionally, four studies investigated both P. falciparum and P. vivax concurrently. Out of 41 studies, the majority (n = 22) focused on a national scale, using genetic data to finely map imported malaria cases within their respective countries.

We identified seven studies focused on cross-border investigations. Only four studies conducted global-level analyses, collecting data from around the world. The majority of studies (n = 17, 41%) used microsatellite markers. Eight studies used SNP markers/barcodes, and only one combined the data of both microsatellite and SNP markers (23). Seven studies employed whole genome sequencing (WGS): three focused on sequencing the entire Plasmodium mitochondrial genome, while others sequenced the genome of P. vivax and P. falciparum. In addition, we identified eight studies that employed targeted genome sequencing primarily focusing on drug resistance genes (n = 4) and genes involved in the parasite's life cycle and immune evasion (n = 4). Fifteen studies included drug resistance markers in their analysis (Table 1).

More than half of the studies (n = 28, 68%) collected travel survey data using a self-reported questionnaire. Only two studies collected mobile phone data (4, 24). The remaining studies did not explicitly mention the collection of travel survey data, mobile data, or GPS tracking data. Regarding the analytical methods, expected heterozygosity (HE) was the most commonly used to assess genetic diversity, applied in 18 studies (44%). The Admixture model was utilized in 16 studies (39%) to analyze population structure, while Principal Component Analysis (PCA) was employed in 13 studies (32%). Machine learning approaches, such as Naive Bayes and Random Forests, were used in four studies to predict the geographic origin of infections. Finally, only one study applied mathematical modeling to assess whether parasite importation could sustain malaria transmission.

Seventeen studies worldwide have investigated the efficiency of MS to determine the genetic diversity and transmission dynamics of malaria parasites. Among these studies, seven were conducted in South America: four in Peru and three in Colombia. The primary focus of studies conducted in Peru was the Amazon region. The earliest study in this region was conducted in 2015 by Baldeviano et al. They genotyped seven microsatellite loci in 54 P. falciparum isolates to identify the source of the outbreak occurred in Tumbes, on northern coast of Peru. The genotype of the parasites from Tumbes was identical to that of all 11 P. falciparum isolates collected from Loreto. The findings suggested that the outbreak in Tumbes may have originated in Loreto (25). In 2016, Delgado-Ratto et al. genotyped 292 P. vivax isolates from Iquitos and 25 peri-urban and rural villages with 14 microsatellite markers. They identified Iquitos as a reservoir of parasite, spreading genetic diversity to the surrounding study areas (26). Similarly, Ventocilla et al. analyzed 145 P. vivax isolates from the Peruvian North Coast (PNC) using five microsatellite markers. They found low genetic diversity in the PNC but high diversity in the neighboring Ecuadorian Amazon Basin, indicating limited population flow between these regions separated by the Andes Mountains (27). Lastly, Manrique et al. conducted molecular surveillance in the Peruvian Amazon using 16 microsatellite markers on 777 mono-infection isolates. Their findings indicated gene flow of P. vivax, supporting the hypothesis that human mobility can potentially connect even geographically distant areas (12).

Among the studies conducted in Colombia, the earliest was by Osorio et al. in 2007. They used five microsatellite to classify 679 P. falciparum isolates into imported and indigenous. They found that microsatellite markers had low power to distinguish between these types due to a predominance of a single genotype and low genetic diversity in both imported and indigenous cases (19). Chenet et al. investigated the genetic composition of 257 P. falciparum isolates collected between 2002 and 2009. The findings suggested that fluctuations in the number of cases during this period were likely driven due to local rather than imported cases (28). Expanding the scope to inter-regional human migration, Pacheco et al. (29) evaluated the genetic differentiation of 624 P. vivax isolates across four geographically separated areas using eight microsatellite markers. Contrary to expectations of clear geographic structure, the study revealed only moderate genetic differentiation. This suggested a corridor between the northwest and the south Pacific Coast of Colombia, where migration of infected individuals enabled the spread of the parasite (29).

The primary focus of studies published in Africa was on Central Africa. Patel et al. (30) investigated whether microsatellite markers could identify the origin of malaria outbreaks among United Nations soldiers returning from the Democratic Republic of the Congo (DRC) to Guatemala. Using eight microsatellite markers, they found that the parasites in the soldiers were genetically similar to those from the DRC, suggesting the outbreak was likely due to P. falciparum imported cases from the DRC (30). In 2019, Roh et al. focused on distinguishing between imported and locally acquired P. falciparum infections in Eswatini. They genotyped 582 isolates using 26 microsatellite markers and found nearly identical levels of genetic diversity in both imported and local infections, making it challenging to differentiate between the two types of infections (31). In comparison, Tessema et al. integrated travel history, mobile phone data, and parasite genetics to study malaria connectivity within Namibia and across the Angolan and Zambian borders. By genotyping 4,643 P. falciparum isolates using 26 markers, they provided strong evidence that most cases in northeastern Namibia were attributed to local transmission. This finding was consistent with estimates derived from mobile phone and travel history data (24). Atuh et al. (32) examined the genetic connectivity of P. falciparum in Northwestern and Southwestern Cameroon using nine microsatellite markers. Despite high human migration between the regions, the analysis of 232 isolates revealed only a small degree of genetic similarity, indicating limited gene flow between these areas (32). In contrast to previous studies, Lo et al. used 14 microsatellites to assess gene flow patterns of both P. falciparum (n = 226) and P. vivax (n = 205) isolates in Ethiopia. The findings revealed that human migrations may promote parasites gene flow between the northern and eastern regions of the country (33).

We identified three studies conducted in Asia, all carried out in China. Liu Yaobao et al. (34) analyzed 128 P. vivax isolates from Central China using seven microsatellite markers. Despite low endemicity, they found high genetic diversity, indicating gene flow among central provinces. However, the markers were not effective in distinguishing local from imported infections (34). Similarly, Lo et al. (15) examined the genetic diversity of P. vivax across 895 samples from the Myanmar-China border using 11 microsatellite markers. They also observed high genetic diversity, suggesting that human migration facilitated parasite gene flow both locally and across the border (15). In contrast, Liu Yaobao et al. (35) focused on identifying the geographic origin of imported P. vivax infections. They analyzed 602 cases traveled from 26 sub-Saharan African countries to China using 26 microsatellite markers. Genetically related infections were found in people who traveled to the same country at about the same time. The findings demonstrated the potential of these markers to trace the origin of infections on a global scale (35).

Only two studies were conducted in Oceania. To understand the migration patterns of P. vivax in Papua New Guinea (PNG), Fola et al. genotyped 219 P. vivax isolates using ten microsatellite markers. The study identified three major genetic populations, with gene flow patterns aligning with human migration routes. Notably, gene flow was higher among Mainland parasite than among Island populations (14). In 2024, Sekine et al. traced the origins of P. vivax among 69 cases from Aneityum Island in Vanuatu. The results of genotyping with five microsatellite markers showed that the parasites responsible for the outbreak in 2002 were imported. However, the exact source(s) could not be determined (36).

We found eight studies that used SNP markers to assess the geographic origin of malaria infections, with two of these studies conducted in Asia. In Sri Lanka, Dewasurendra et al. (37) investigated P. vivax origins using a 40-SNP barcode on samples collected from 2005 to 2011. Nine samples clustered with the South American group, indicating these isolates were most probably imported cases. However, the lack of travel histories made such inferences remain unconfirmed (37). In contrast, Chang et al. (4) integrated data from a 101-SNP barcode with travel survey and mobile phone data to assess malaria transmission countrywide in Bangladesh. They could detect likely imported cases among 1,412 malaria-positive cases (4).

We identified a study conducted in Peru, South America by Kattenberg et al. They used a 33-SNP barcode to predict the origin of P. vivax infections in the Peruvian Amazon. The samples from a border community showed high similarity to Brazilian isolates, likely indicating the introduction of P. vivax from Brazil. Moreover, seven out of 230 samples (3%) were predicted to originate from Vietnam, Afghanistan, and Iran. However, this finding could not be fully confirmed as incorrect predictions might have arisen due to missing data in the SNP barcode (38). In Central America, only one study in Honduras explored the population structure of both P. vivax and P. falciparum isolates. The results showed that two out of three isolates with chloroquine-resistant parasites were imported cases from Africa. In contrast, SNP barcodes could not provide any information about the origin of infection for P. vivax (39).

Lastly, we found four studies conducted on a global scale. The first study, published by Preston et al. in 2014, included 711 P. falciparum isolates from 14 countries and developed a 23-SNP barcode. The SNP barcode determined the geographic origin of parasite isolates with 92% accuracy, as confirmed by available data regarding the origin of each sample (40). The same approach was applied by Ba et al. to compare P. vivax population in Mauritania, West Africa, with parasite populations in other countries. The results of a 38-SNP barcode revealed that the Mauritanian samples (n = 100) formed a distinct geographical group, with slightly closer relatedness to Ethiopia than to the samples from Southeast Asia (41). In comparison, Trimarsanto et al. (42) used a smaller number of SNPs (n = 33) to predict the infection origin of 799 P. vivax isolates collected from 21 countries. The findings confirmed the capacity of this SNP panel to predict infection's country of origin accurately (42). The most recent publication was conducted by Phelan et al. in 2023. They used the largest number of genetic markers (n = 274), incorporating a combination of drug resistance mutations, mitochondrial SNPs, and a set of established SNP markers. Analyzing data from 7,152 P. falciparum and 1,502 P. vivax samples revealed that genetic markers could predict the geographical origin of infections, with accuracy rates of 96.1% at the continental level and 94.6% at the regional level (22).

Khaireh et al. genotyped five microsatellites and five SNPs associated with pyrimethamine resistance in 181 P. falciparum isolates collected over an 11-year period (1998, 1999, 2002, and 2009) from the low malaria transmission setting of Djibouti. The most notable finding was the decline in genetic diversity among parasite populations over the study period. This was possibly due to a decline in malaria transmission from neighboring countries, particularly Ethiopia (23).

We identified eight studies focused on targeted genome sequencing. Cui et al. investigated the genetic diversity of the P. vivax apical membrane antigen-1 (PvAMA-1) gene, an important vaccine candidate for vivax malaria, in 152 imported cases from China and 73 isolates from Myanmar. They found high genetic diversity and positive selection in this gene among imported cases along the China–Myanmar border, implying that traveling may affect the population structure and genetic characteristics of malaria on the Chinese side of the China–Myanmar border (43). Xu et al. (44) studied nucleotide diversity in the polymorphic region of the rif gene which is involved in immune evasion. High genetic differentiation was found between 403 malaria cases imported from Ghana to China and South Asian populations, including, Thailand, Vietnam, Myanmar, and Cambodia. The findings indicated that the P. falciparum populations have a substantial continental genetic structure (44). Ruybal-Pesantez et al. (45) analyzed nine genetic variants in the DBLα region of var genes, which are involved in immune evasion, to assess parasite relatedness. Most of the 58 P. falciparum isolates in Ecuador were closely related, indicating local transmission rather than importation from other regions. However, the genetic similarity of some isolates to those in South America, particularly from Colombia (n = 5) and Peru (n = 1), suggested historical parasite importation (45). To investigate the contribution of parasite importation on local transmission in Kenya, Markwalter et al. (46) analyzed data from 2,521 P. falciparum isolates together with their travel histories. Deep amplicon sequencing of the circumsporozoite protein (csp) and apical membrane antigen 1 (ama1) genes led to the identification of 72 csp haplotypes and 88 ama1 haplotypes. The findings suggest that travelers’ contribution to overall transmission was very limited, making the likelihood of establishing new transmission chains by travelers very unlikely (46).

Unlike other studies, Hamedi et al. genotyped nine Single Tandem Repeat (STR) markers on the pvmdr1 gene, which is associated with drug resistance, particularly in P. vivax. They could not find any notable genetic differentiation between 23 imported cases from Afghanistan or Pakistan and 73 autochthonous cases in Iran (47). In Brazil, Griffing et al. genotyped blood spots collected on filter paper from the 1980s and 1990s across three regions to investigate internal migration of P. falciparum. They analyzed 190 samples using 28 microsatellite markers on the dhfr and dhps genes, known to be associated with sulphadoxine/pyrimethamine (SP) resistance. Their findings suggested that internal migration of the parasite between the regions might lead to admixture of parasite lineages (48). In one of the most recent studies, Holzschuh et al. sequenced 187 P. falciparum samples from two regions in Ethiopia. Unlike other research, they used multiplex amplicon deep sequencing to analyze 35 loci, including those associated with drug resistance. The findings suggested that individuals with recent travel history might initiate local transmission of the malaria parasite within the community which then led to ongoing local transmission (49). Pierre-Louis et al. examined drug-resistance genotypes among 234 U.S. travel-related P. falciparum infections from 2018 to 2021. They found that Pfcpmp alone was not a reliable marker for determining geographic origin. However, a panel combining 61 SNPs in Pfcpmp with 69 SNPs in Pfs47 improved classification accuracy to about 95% at the continent level, highlighting the complementary benefit of sequencing these markers together (50).

Only seven studies used whole genome sequencing (WGS) data with different approaches: one focused on the mitochondrial genome of P. falciparum (51) and two on P. vivax (21, 52) while the others examined the whole genome of P. falciparum (53) and P. vivax (54–56). In 2019, Schmedes et al. sequenced the Plasmodium mitochondrial genome to trace the origin of imported P. falciparum cases in the U.S. from 2014 to 2017. Of the 265 cases, 13 samples were mapped to specific regions, like Philippines and Ghana. However, most samples could not be traced due to the presence of common mitochondrial haplotypes (51). Earlier, in 2014, Rodrigues et al. obtained similar results by analyzing 348 mitochondrial genomes from P. vivax parasites worldwide and sequences from 69 imported P. vivax infections diagnosed in the United States. Given the genetic similarities between mitochondrial lineages from Africa, South Asia, Central Asia, and the Middle East, mitochondrial haplotypes could not accurately assign the geographic origin of malaria cases (52). The same approach was applied by Li et al. to investigate the geographical origin of P. vivax cases in Hainan, China. The origin of seven out of 14 imported cases could be inferred which was consistent with recorded travel histories. However, the origin of other imported cases could not be traced, raising questions about the suitability of using mitochondrial genomes in this context (21).

In 2020, Buyon et al. explored patterns of recent common ancestry among 59 P. vivax samples from Panama collected between 2007 and 2009 and 2017–2019. The results revealed that four samples with travel history did not show recent ancestry connections with other Panamanian samples; instead, they clustered with samples from a previous study conducted in Colombia. Nevertheless, there was no indication of outcrossing between the potentially imported parasites and the local Panamanian parasite population (54). On a larger scale, Benavente et al. used WGS data to determine the geographical origin of 433 P. vivax isolates from 17 countries worldwide. The findings showed that a 71-SNP barcode had high predictive power, successfully identifying the geographic origin with 91.4% accuracy (55). In comparison, Kattenberg et al. developed two assays based on WGS data to predict the geographic origin of imported P. vivax infections at the country level and to detect small-scale population genetic variations within Vietnam. Using a 72-SNP barcode, 327 out of 402 samples (88.8%) were correctly mapped to expected origin, as determined by travel history or the sample collection site (56).

On a smaller geographic scale, Morgan et al. used the WGS data to investigate the importation of P. falciparum parasites from the high-transmission regions of mainland Tanzania (n = 36) to the low-transmission areas of Zanzibar (n = 21). The findings suggested that human travel to high-risk malaria regions is the most likely source of parasite importation into Zanzibar (53).

The methods used to analyze genetic data can be broadly grouped into three main groups: those that assess genetic diversity, examine population structure, and determine relatedness among populations or individuals. The most common methods involved assessing genetic diversity and population structure. In assessing genetic diversity, expected heterozygosity (HE) was the most commonly used method applied in 18 studies, accounting for 44% of the total (Table 2). This metric was primarily used to indirectly measure the impact of imported cases on genetic variation within the local populations, specifically in the context of longitudinal studies. These studies measured HE over a defined period allowing the identification and analysis of trends in HE. The primary assumption in these studies was that, in the absence of imported infections, the HE would remain relatively stable over time. In contrast, the introduction of imported infection was expected to alter the HE values. As a result, these studies did not necessarily aim to test specific hypotheses comparing HE between different time points. Instead, they focused on comparing the summary statistics of HE to observe broader trends in genetic diversity. The Fixation Index (FST) was the second most frequently used method, appearing in 15 studies (37%). FST was used to measure the genetic differentiation among populations, in some cases, extrapolating the FST as measure of gene flow.

The Admixture model was used in 16 studies (39%) to analyze population structure, providing insights into the contribution of different populations to the genetic composition of individuals and revealing historical and ongoing parasite migration across geographical regions. Principal Component Analysis (PCA) was employed in 13 studies (32%) to detect population structure and substructure (Table 2). Both the Admixture model and PCA uncovered patterns and structure in genetic data to some extent. However, as exploratory methods without hypothesis testing, their results were fundamentally more subjective than objective.

We found only five studies (12%) that applied methods capable of directly estimating potential evidence of migration between populations. Delgado-Ratto et al. (26) assessed human migration patterns in and around Iquitos City using a Bayes approach based on the coalescence theory. The findings showed that such a model can estimate the magnitude and direction of parasite importation. Unfortunately, the study did not collect any information on travel patterns, making it almost impossible to confirm the findings (26). To explore gene flow between the Peruvian North Coast and the Ecuadorian Amazon Basin, Ventocilla et al. developed two migration models using a Bayesian approach with MIGRATE-N. The models showed the presence of three distinct subpopulations with some level of migration occurring between them. These findings were consistent with the results obtained from the structure analysis (27). In a similar approach, Lo et al. estimated the intensity and direction of gene flow using a Bayesian approach implemented in BayesAss program. The bidirectional migration was observed between the north and east Ethiopia, indicating that human migrations may facilitate the parasite gene flow (33). In 2017, Lo et al. applied a similar method, using a Bayesian approach to estimate the posterior probability distribution of the proportion of migrants from one population to another. The findings revealed a greater migration rate from Myanmar to China. By contrast, migration from China to Myanmar was relatively small (15). Unlike the other studies, Fola et al. (14) used divMigrate-online software to estimate migration patterns among different geographic regions of Papua New Guinea. Their findings showed more migration within the mainland than within the Island region and relatively limited migration between them (14). In addition, we identified two studies that developed a novel method to determine if an isolate is imported. Chang et al. compared the geographic distance between isolates with geographic distances estimated based on genetic data. They quantified the probability of a geographic distance given an SNP difference by applying Bayes’ rule (4). In contrast, Tessema et al. developed an approach that combined parasite genetic data and travel history data to estimate local and cross-border importation in Namibia. Infections from Namibia were more found to be genetically related to those from southern Angola and Zambia, compared to northern Angola. This indicated parasite mixing within the geographically connected Namibia-Angola-Zambia regional zone (24).

Interestingly, we identified four studies employing machine learning approaches to detect imported malaria. In 2022, Trimarsanto et al. developed a Naive Bayes classifier to predict the origin country of the infection. This likelihood-based classifier demonstrated a high capacity to predict the infection's country of origin with 95% accuracy. However, its application was limited to situations where the genetic reference panel of a given country was available and had an adequate sample size (42). The same approach was employed by Kattenberg et al. in 2022 to predict the country of origin for imported P. vivax cases, achieving a high prediction accuracy of 88.8% (56). In contrast, Pierre-Louis et al. trained this classifier using P. falciparum genomes of known geographic origin to classify travel-related cases at the continental level. The findings demonstrated a high accuracy of 95% (50). Lastly, Benavente et al. applied the Random Forests approach to first identify the SNPs with the highest predictive importance and then to classify the geographical origin of P. vivax malaria. This method achieved a high predictive accuracy of 91.4% in identifying the geographic origin of the infection (55).

Among the 41 studies, only one used mathematical modeling to estimate the impact of travelers on the persistence of malaria transmission. The study modeled the reproductive number (R) and effective reproductive number (Rt) both with and without considering travelers. The findings indicated that the case reproductive numbers were approximately 1 and remained unchanged when travelers were excluded from the transmission networks, suggesting that malaria transmission is primarily driven locally rather than by travel to other regions (46).