Plasmodium falciparum infection in humans and mosquitoes influence natural Anopheline biting behavior and transmission

Plasmodium falciparum transmission depends on both the frequency of mosquito–human contact and the efficiency of parasite transmission during those contacts. Nonrandom vector–host contact and heterogeneity in human infectiousness can strongly shape transmission dynamics and influence the effectiveness of interventions. Prior laboratory, semi‑field, and field studies demonstrate high inter‑individual variability in attractiveness to mosquitoes and suggest that Plasmodium infection can alter mosquito biting behavior. However, most prior work is experimental or short‑term and does not quantify variation in mosquito–human contact over an entire transmission season within a defined population or consider multispecies bloodmeals. This study aims to determine who gets bitten, who transmits to mosquitoes, and which mosquito characteristics shape biting behavior in a natural high‑transmission setting, thereby informing accurate estimates of transmission patterns and targeting of interventions.

Previous studies have shown substantial heterogeneity in mosquito bite exposure and in human infectiousness to mosquitoes, with laboratory and semi‑field experiments indicating that Plasmodium infection may modify mosquito host‑seeking and biting behavior. Field studies have documented nonrandom human host selection by Anopheles vectors and variable human contributions to the infectious reservoir, but most were limited in duration, scope, or did not account for multisource bloodmeals. There is a need for comprehensive, longitudinal, community‑based assessments that integrate both mosquito and human infection data to quantify biting heterogeneity and its implications for transmission.

Design and setting: Community‑based longitudinal cohort in western Kenya with 15 months of observation (July 2020–September 2021) across 75 households selected by radial sampling in five villages of varying malaria intensity. Participants: All residents aged >1 year. Data collection: Monthly dried blood spots (DBS) from all participants (active surveillance) and additional DBS at symptomatic visits with RDT testing (passive surveillance), including self‑reported bed net use on the night prior to DBS collection. Entomology: Weekly indoor‑resting mosquito collections (one morning per week, three weeks per month) using Prokopack aspiration. Female Anopheles were morphologically identified and abdominal status graded (freshly fed, half‑gravid, gravid, unfed). Freshly fed abdomens were separated for bloodmeal analysis; parts (head, thorax, abdomen) were preserved for molecular assays. A subset of collected mosquitoes (n=622) was reared; survival to day ≥6 was recorded and abdomens tested for P. falciparum to estimate infection following the most recent bloodmeal. Molecular detection: P. falciparum detection by TaqMan qPCR on human DBS, mosquito head‑thoraces (for sporozoites), and reared mosquito abdomens; parasite density estimated via standard curves. An. gambiae s.l. sibling species resolved by PCR targeting the IGS region. Human genotyping: STR profiling (Promega PowerPlex/“Centriq 10” assay) of DBS from participants and mosquito bloodmeals; the bistro R package matched single‑ and multi‑source mosquito bloodmeals to individuals using likelihood ratios, allowing incomplete STR profiles and estimating number of contributors. Matching infections: Infected mosquitoes were linked to potentially infectious humans within the same family compound on the collection day; haplotype sequencing of csp amplicons was attempted on infected human and mosquito samples to explore parasite genetic matches. Outcomes and analyses: - Biting heterogeneity quantified via mean nightly bites per person‑night at risk and Gini indices with simulation under random biting for comparison. - Risk factor analysis using negative binomial mixed models (glmmTMB) with person‑level random effects to estimate biting rate ratios (BRRs) for gender, age categories, infection status, bed net use, season, household structure, and mosquito species. - Discrete choice (mixed logit) models assessed relative risk ratios (RRRs) for host selection as a function of mosquito characteristics (species, multi‑source bloodmeal, sporozoite positivity and density) and human factors. - Maximum human‑to‑mosquito transmission efficiency estimated from the proportion of reared abdomens infected, adjusted by the proportion of immediately processed mosquitoes that had recently fed, and the probability a mosquito bit ≥1 infected person (derived from STR matches), using bootstrap to derive 95% CIs. Assumptions included that fed mosquitoes rested indoors where collected and that matched bloodmeals corresponded to bites in the preceding 24 hours. Data management and visualization were conducted in R/RStudio with standard packages.

• Mosquito collections and infections: 3660 female Anopheles collected; species composition approximately 39% An. gambiae s.s., 46% An. arabiensis, 7% An. funestus s.l. Of 622 reared females, 412 survived to day ≥6 and 34 (3.3%) had P. falciparum in the abdomen. Estimated maximum human‑to‑mosquito transmission efficiency was 0.16 (95% CI: 0.12–0.22). - Human infection: 6414 human DBS collected; 1758 (27%) qPCR‑positive for P. falciparum; individuals were infected at a median of 20% of monthly visits. - Bloodmeal matching: Of 1064 STR‑typed freshly fed abdomens, 654 (61%) were single‑source and 123 (16%) multi‑source; at least one cohort participant was identified in 628 (85%) bloodmeals. Among 726 mosquito–human pairs, 94% were collected in the family compound of the matched person; in multi‑source meals with two matched contributors, 88% were from the same family compound. - Biting heterogeneity: 20% of people received 86% of observed bites. Mean nightly biting rates per person ranged from 0.03 to 15. Inequality exceeded random expectation (Gini index 0.51, 95% CI: 0.48–0.54). - Individual risk factors for being bitten (BRR, multivariable): male sex 1.68 (95% CI: 1.28–2.19); age 5–15 years 1.49 (95% CI: 1.13–1.98) vs >15; current P. falciparum infection 1.25 (95% CI: 1.01–1.56). Lower biting rates observed for An. funestus (BRR 0.57, 95% CI: 0.40–0.65). - Potential onward transmission: Restricting to infected humans, school‑age boys (5–15) received about 50% of bites that could lead to onward transmission. Infectious mosquito bites: participants infected with P. falciparum had higher biting risk from infectious mosquitoes (BRR ≈2.6; 95% CI: 1.40–6.80). Estimated entomological inoculation risk for males 5–15 with infection was 6.4× higher than any other group. - Mosquito characteristics and host selection (discrete choice models): Infectious (sporozoite‑positive) mosquitoes preferentially bit males over females (RRR 2.02, 95% CI: 1.28–3.15), children <5 (RRR 2.69, 95% CI: 1.19–6.09) and 5–15 years (RRR 2.21, 95% CI: 1.37–3.59) over >15, and those not sleeping under a net (RRR 2.22, 95% CI: 1.15–4.60). Critically, infectious mosquitoes were nearly three times more likely to bite P. falciparum‑infected individuals (RRR 2.76, 95% CI: 1.65–4.61). Higher mosquito sporozoite densities were associated with a greater probability of biting infected versus uninfected humans (per 100 sporozoites: BRR 1.92, 95% CI: 1.23–2.89). Multi‑source biting mosquitoes were less likely to bite males (RRR 0.59, 95% CI: 0.33–0.77) and more likely to include a child <5 years (RRR 4.10, 95% CI: 1.80–9.35) compared to >15 years, consistent with shared sleeping spaces.

The study demonstrates pronounced heterogeneity in mosquito biting across a defined community, with school‑age boys disproportionately bitten and contributing to onward transmission. Infectious mosquitoes displayed a marked preference for feeding on individuals currently infected with P. falciparum, and this preference strengthened with increasing sporozoite density, consistent with parasite‑mediated modification of vector host choice. These patterns, observed over an entire transmission season with direct matching of bloodmeals to individuals, clarify who is most at risk of receiving bites and who is most likely to propagate the parasite to mosquitoes. The findings highlight school‑age children—particularly boys—as key drivers of transmission and suggest that targeting interventions (e.g., net use reinforcement, chemoprevention, or transmission‑blocking strategies) to this group could disproportionately reduce transmission. Similar biting preferences across An. gambiae s.s. and An. funestus and evidence of shared parasite haplotypes suggest a co‑recombining reservoir that may facilitate parasite genetic diversity through repeated polygenomic infections.

Using longitudinal, individual‑level bloodmeal matching in a natural setting, the study quantifies strong biases in mosquito biting and links mosquito infectiousness to preferential feeding on infected humans. School‑age boys are disproportionately bitten and contribute the majority of potentially onward‑transmitting bites, with markedly elevated estimated inoculation risk. These insights improve transmission modeling and inform the design of targeted, transmission‑reducing interventions. Future research should refine discrimination of parasite stages in mosquito tissues, expand reared‑mosquito sample sizes to more precisely estimate human‑to‑mosquito transmission efficiency, and evaluate targeted interventions for school‑age children on community‑level transmission and parasite population structure.

• Reared mosquitoes: Initial feeding status could not be determined, and the small number of infected reared abdomens limited inference on differential onward transmission. - Mosquito infection detection: qPCR on head‑thoraces cannot definitively distinguish sporozoites from residual blood‑stage parasites, raising potential misclassification despite sensitivity analyses using varying positivity thresholds. - Matching constraints: Not all mosquito STR profiles matched cohort members, implying some bites were on non‑enrolled individuals; although methods accommodating incomplete and multisource profiles increased usable matches by ~37%, residual misclassification is possible. - Assumptions: Analyses assumed equal transmission efficiency and gametocyte carriage across groups and that matched bloodmeals represented bites within 24 hours and occurred within the household; deviations could bias estimates. - Potential contamination and storage issues may affect parasite detection and stage attribution in mosquito samples. - Limited power for some subgroup analyses, especially with infectious mosquitoes and multi‑source bloodmeals.