Behavioral patterns of humans and Anopheles arabiensis reveal outdoor malaria transmission risk at Pemba Island, Zanzibar Authors: Ali O Ali, Bakar O Khatib, Khadija F Ali, Fatma A Massoud, Talib S Khatib, Moza K Rashid, Steven Gowelo, Shija J Shija, Kali A Omar, Makame M Kombo, Joseph Joachim Joseph, Nicodem J Govella, Neil F Lobo Published: 2026 ABSTRACT AbstractBackgroundZanzibar has achieved and sustained very low malaria prevalence through high coverage of indoor residual spraying (IRS) and insecticide-treated nets (ITNs). However, residual transmission persists, particularly on Pemba Island. Increasing evidence suggests that the predominance of exophagic vectors and outdoor human activity patterns may reduce the effectiveness of indoor-focused interventions. This study aimed to quantify malaria transmission risk arising from the interaction between human behavior and mosquito biting activity indoors and outdoors on Pemba Island.MethodsFrom February 2024 to January 2025, adult malaria vectors were collected using human landing catches (HLCs) indoors and outdoors at sentinel sites across four districts of Pemba Island. Concurrent human behavior observations (HBOs) were conducted hourly to record location, activity, age, sex, and ITNs use within households. Human biting rates (HBRs) and behavior-adjusted biting rates were calculated to estimate human behavior-adjusted exposure and identify protection gaps by age and sex.ResultsA total of 981 Anopheles mosquitoes were collected, of which 98% of An. gambiae s.l. were An. arabiensis. Mosquito collection was significantly higher outdoor than indoor (Relative rate = 4.37, 95% CI 3.61–5.29, p < 0.0001). Most bites occurred outdoors (≈82%), with peak biting in the early evening (19:00–20:00 h) for both indoor and outdoor. Across the population, approximately 85% of unprotected exposure occurred outdoors while individuals were awake. Children under five experienced relatively more indoor exposure, while school-aged children and adults were predominantly exposed outdoors. Men experienced higher outdoor exposure than women, whereas women had greater indoor exposure while awake. ITNs provided substantial protection for young children but prevented a declining proportion of exposure with increasing age, leaving substantial residual risk across all groups.ConclusionsResidual malaria transmission on Pemba Island is driven primarily by outdoor exposure linked to human behaviour patterns and exophagic An. arabiensis. While ITNs continue to protect young children indoors, they do not address most exposure among older children and adults. Malaria elimination efforts in Zanzibar will require complementary interventions that target outdoor transmission and behavioral risk, alongside existing indoor-focused strategies.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12936-026-06017-5.Keywords: Human behaviour observation, Anopheles arabiensis, Outdoor malaria transmission, Insecticide-treated nets (ITNs), Pemba Island, Zanzibar Background Pemba Island, together with its sister island Unguja, forms Zanzibar and has achieved a dramatic reduction in malaria prevalence to below 1%, maintaining this low level for more than a decade. This success has been driven by sustained implementation of core vector control interventions, including continuous mass distribution of insecticide-treated nets (ITNs) and indoor residual spraying (IRS) [1–3]. Progress toward near-zero malaria transmission has been more rapid and stable on Pemba Island compared with Unguja Island, which has experienced intermittent and seasonal resurgences (National Malaria Strategic Plan (V) 2024–2029, Zanzibar Malaria Elimination Program, unpublished). Despite these gains, residual malaria transmission persists on Pemba Island. Although it is widely acknowledged that identifying the factors sustaining residual transmission is essential, the core drivers of this persistence remain incompletely understood. Growing evidence indicates that the effectiveness of ITNs and IRS has been compromised by changes in both species composition of An. arabiensis and human behavior, that is, the persistence of transmission reflects the intrinsic exophagic and exophilic tendencies of this species, which are less effectively targeted by indoor interventions such as ITNs and IRS, rather than a decline in the efficacy of these core tools. The current transmission pattern is largely influenced by a shift in vector species composition, particularly the dominance of An. arabiensis, alongside human activity patterns that increase outdoor exposure [4–6]. Although vector biting behavior has changed substantially in many settings following the widespread deployment of indoor-focused interventions, the pattern observed in Pemba appears to reflect a shift in vector species composition rather than a survival-driven behavioural adaptation. Substantial evidence indicates that ITNs and IRS remain highly effective against indoor-adapted vectors such as An. gambiae s.s., while the persistence of transmission is increasingly driven by An. arabiensis, a species whose intrinsically exophagic and exophilic ecology makes it less effectively targeted by these interventions [7, 8]. A recent study conducted in Zanzibar reported that An. arabiensis is now the predominant malaria vector across all districts of Pemba Island, with approximately 81% of biting occurring outdoors [9]. Similar trends have been documented across sub-Saharan Africa, where changes in vector species composition and behavior have increasingly favored An. arabiensis, a species characterized by flexible resting and feeding behaviors, including outdoor and early evening biting [4, 10, 11]. These ecological and behavioral characteristics enable sustained malaria transmission even in settings with high ITNs coverage and IRS implementation [5, 9]. Consequently, residual malaria transmission on Pemba Island is likely driven in part by outdoor exposure to mosquito bites, which is not effectively addressed by current indoor-focused interventions. Human behavior also plays a critical role in determining the protective effectiveness of ITNs. Although ITNs remain a cornerstone of malaria prevention, their protective effect depends strongly on human sleeping patterns, which in many African settings typically begin after 21:00 h. This creates an early evening window during which individuals remain unprotected [5, 6]. Exposure may also continue later in the night among individuals who are awake or asleep without bed nets [12]. Patterns of ITNs use within communities further influence intervention effectiveness. Evidence from Zanzibar indicates that while ITNs can prevent approximately 66% of potential mosquito bites, overall population-level protection is substantially lower, estimated at only 42%, due to incomplete and inconsistent use [13]. High levels of outdoor activity during peak mosquito biting periods, combined with variable ITNs utilization, create exposure gaps that challenge the effectiveness of standard indoor-focused interventions [14]. Even among individuals who report consistent ITNs use during sleeping hours, substantial residual exposure persists in the early evening and early morning when nets are not in use [11]. While entomological data highlight a high density of outdoor-biting malaria vectors and the associated potential for outdoor transmission [9], there remains limited empirical evidence linking these patterns to detailed human behavior across age groups, sex, and timing of ITNs use. A comprehensive understanding of human–vector behavioral interactions is essential, as malaria exposure depends on where people are, when they are active, and whether they are protected (e.g., by ITNs) [15]. Such information is critical for identifying key drivers of residual transmission and informing targeted interventions, including optimization of existing vector control tools, behaviorally informed interventions, and integrated approaches that combine environmental management, personal protection, and vector control strategies [16, 17]. Therefore, this study aimed to assess malaria transmission risk arising from the interaction between human behavior and mosquito biting activity indoors and outdoors on Pemba Island by: (1) characterizing human–vector interaction patterns, (2) evaluating the impact of human behavior on malaria transmission risk, and (3) assessing the effectiveness of current interventions to inform policy and guide the design of robust, context-specific complementary strategies to support malaria elimination on Pemba Island. Methods This study was designed as an operational descriptive assessment intended to characterize exposure patterns under routine programmatic conditions rather than to formally test comparative epidemiological hypotheses. Study area and sentinel sites The study was conducted in Pemba Island, one of the two main islands of Zanzibar, located off the northern coast of Tanzania. Geographically, it is divided into two regions (Kaskazini Pemba, and Kusini Pemba) with four districts (two districts per each region) (Fig. 1). According to the latest census 2022, the population was recorded as 543,41 individuals [18]. The island has a tropical monsoon climate with two rainy seasons: March to May and October to December. In 2024, Pemba recorded warmer and wetter conditions, with temperature anomalies of + 1 °C to + 2 °C and intense rainfall events, including 140.1 mm in 24 h [19]. Map of the Pemba Island showing the distribution of sentinel sites at the district level Entomological surveillance was conducted in four sentinel sites (one site per district) all of which are located at the shehia level (Fig. 1). These sentinel sites were selected based on defined criteria including, a diversity of malaria endemicity, the risk of malaria transmission from environmental determinants such as agricultural practices like rice irrigation, topography, and population density (urban versus rural). In Pemba, high ITN coverage has been sustained through repeated mass distribution campaigns. The 2022 Tanzania DHS/MIS [20] reported that 88.0% of households in Pemba Island owned at least one ITN. More recently, a mass campaign conducted in 2024 distributed 340,895 (99.86%) ITNs across four districts, covering 626,784 people and achieving the target ratio of one net for every two persons in registered households (ITNs mass distribution campaign report—Unpublished). In addition to these periodic mass campaigns (implemented approximately every three years), ITNs are also distributed through routine channels, including antenatal care (ANC) services and provision to malaria patients from households without nets to ensure the coverage is sustained. Data collection Malaria vector sampling From February 2024 to January 2025, adult mosquitoes were collected indoors and outdoors at four district-level sentinel sites using human landing catches (HLCs) conducted between 18:00 and 06:00 h. Human landing catches were performed using adult male volunteers, a method currently permitted under routine programmatic surveillance activities in Zanzibar. All volunteers provided written informed consent and were screened for malaria infection prior to each monthly collection round, which consisted of two consecutive sampling nights. Volunteers who tested positive for malaria were treated according to national malaria treatment guidelines before resuming participation. Using the HLC method, trained volunteers sat on chairs with their lower legs exposed and collected mosquitoes attempting to bite using mouth aspirators [21, 22]. At each sentinel site, two households located approximately 100 m apart were randomly selected from a list of eligible households. Priority was given to households representative of the local setting in terms of house construction (e.g., wall materials and roof type), ventilation characteristics (e.g., open or closed eaves), and surrounding environmental features such as nearby vegetation, water bodies, and animal shelters. Four volunteers were assigned to each household, with two conducting collections indoors and two outdoors. Collections were conducted in two alternating 6-h shifts (18:00–00:00 and 00:00–06:00) to reduce fatigue and improve sampling consistency. To minimize potential bias associated with individual attractiveness to mosquitoes, volunteers rotated monthly between indoor and outdoor positions as well as between households [23]. Mosquito collections were conducted for two nights per month at each household, resulting in a total of 192 sampling nights across the eight study households over the 12-month study period [(2 nights × 8 houses) × 12 months = 192 nights]. Adult mosquitoes were collected for 45 min of each hour, followed by a 15-min break period during which volunteers rested and conducted human behaviour observations. Collected mosquitoes were stored in pre-labeled paper cups according to collection hour and location (indoor or outdoor), placed in insulated cool boxes to preserve specimen integrity in the field, and transported to the laboratory each morning for sorting and morphological identification. Relative rates (rate ratios) were calculated as the ratio of mosquito counts between outdoor and indoor collections, assuming equal sampling effort (person-hours) across both settings. Processing of the mosquito sample and molecular analysis Mosquito samples were morphologically identified using a morphological key [21] under a dissecting microscope. All Anopheles specimens were individually stored in 1.5 ml tubes containing desiccated silica gel under cotton wool for PCR-based species identification. Samples were also tested for the presence of sporozoites using the diagnostic enzyme-linked immunosorbent assay (ELISA) [22]. To eliminate false positives, all circumsporozoite protein (CSP) ELISA-positive lysates were heated to 100 °C for 10 min [23]. All laboratory analyses were conducted at Zanzibar Malaria Elimination Program laboratory. Human behaviour observations (HBO) Human behaviour observations (HBOs) were conducted concurrently with HLC collections to characterize patterns of human exposure to mosquito bites [24]. Observations were carried out during the final 15-min break of each hourly HLC collection period by the same trained field staff conducting mosquito collections. Indoor and outdoor observations were conducted at the same study households selected for HLC collections following standardized observation protocols. Outdoor observations were restricted to a defined area within approximately 10 m of the central study household. All observers received standardized training prior to field deployment to ensure consistency in recording human behavioural patterns across all study sites. In addition, routine supervision and periodic spot checks were conducted by field supervisors to ensure protocol adherence and data quality throughout the study period. During each hourly observation period, all individuals present within and around the study household were systematically categorized according to sex (male or female), location (indoors or outdoors) and behavioural status, including awake, asleep under an ITN, or asleep without an ITN. Age groups were classified using household information and categorized into three groups: < 5 years, 5–15 years, and > 15 years, consistent with the study design. All observations were recorded hourly using structured data collection forms with predefined categories to ensure consistency across sites. The synchronized timing of HBO and mosquito collection data allowed integration of human behavioural and entomological information for estimation of human behaviour-adjusted malaria exposure [13, 24, 25]. Measured indicators and data analysis The human biting rate (HBR) was defined as the total number of mosquitoes collected by HLCs divided by the number of collectors and the number of collection nights, typically expressed as bites per person per night [26]. The HBR was calculated hourly, however, only An. arabiensis was included in HBR calculation as it was the predominant vector species with adequate sample size for both indoor and outdoor collections to do robust analysis compared to other collected species. The human behavior-adjusted biting rate was estimated based on Finda et al. and Monroe et al. [6, 13], by combining HLC and household HBOs data to account for actual exposure in the community. This measure also incorporated ITNs usage (current proportion of vector bites prevented by ITNs users by age and sex) and identified protection gaps, including periods when individuals were indoors awake or asleep with or without a net, and outdoors where net use was not feasible. The estimated proportion of exposure was calculated for the general population and further stratified by age and sex. These exposure estimates were limited to individuals who were not using ITNs at the time of observation. For ITN users, the analysis instead estimated the proportion of vector bites potentially prevented by ITN use, rather than direct exposure. Results Anopheles species caught using HLC and their biting behaviour A total of 981 malaria vectors were collected using HLCs over 192 collections nights. Of these 981 mosquitoes, 948 (96.6%) were An. gambiae sensu lato (s.l), 14 (1.4%) were An. coustani, 10 (1.0%) were An. rufipes, 6 (0.6%) were An. funestus complex, and 3 (0.3%) were An. maculipalpis. Of the 948 An. gambiae s.l. collected, 176 (18.6%) were caught indoors and 772 (81.4%) outdoors. Other species showed a predominantly outdoor distribution, including An. coustani (2 [14.3%] indoors versus 12 [85.7%] outdoors) and An. rufipes (1 [10%] indoors versus 9 [90%] outdoors), while both An. funestus s.l. and An. maculipalpis were exclusively collected outdoors (100%). Among the An. gambiae s.l., 932 (98%) were identified as An. arabiensis, while 16 (2%) were An. merus. Only An. arabiensis samples were included in detailed subsequent HBR analysis, since the hourly catches and their distribution between indoors and outdoors for other species were insufficient for a more analysis that would provide a meaningful interpretation. Of the An. arabiensis collected, 171 (18.3%) were captured indoors and 761(81.7%) were outdoors. More mosquito samples (n = 661) were collected during the first six hours of night (18:00 to 00:00 h), with outdoor collection (outdoor, n = 538 vs indoor, n = 123) approximately four times higher than indoors, indicating that collection was significantly higher outdoor than indoor (Relative rate = 4.37, 95% CI 3.61–5.29, p < 0.0001). The mean outdoor HBR was 0.04 bites per person per night compared with 0.01 indoors. The peak biting hour for both indoor and outdoor occurred between 19:00 and 20:00 h representing 0.02 and 0.08 per person per hour respectively. Human behaviour adjusted data and pattern of exposure Across all age groups, an estimated 84.7% of vector bites occurred outdoors, while 15.3% occurred indoors based on combined HLC and HBO data. Moreover, outdoor exposure was highest during the early evening (18:00–22:00) and early morning hours (04:00–05:00), when individuals were active outdoors and indoor when asleep or awake without using ITN. Among ITN users, ITN use was estimated to prevent 18.4% of potential vector bites. However, substantial residual exposure still occurred during periods when individuals were not under ITN protection. However, this pattern showed substantial heterogeneity by age and sex in both indoor and outdoor vector bites, as well as in the degree of protection offered by ITNs. Proportion of human exposure to malaria vectors by age group and sex Observed exposure patterns varied by age group. Children under 5 years experienced greater observed indoors exposure, with 53.4% of bites occurring indoors. In contrast, school-aged children (5–15 years) and individuals aged (> 15 years) were primarily exposed outdoors, accounting for 80.6% and 94.9% of human behavior adjusted bites, respectively. Sex-disaggregated analyses demonstrated that men were observed to have descriptively higher outdoor exposure than women (89.1% vs 76.6%), whereas, women had greater indoor exposure (23.4% vs. 10.9%) (Fig. 2A–F). Hourly human-An. arabiensis interactions by age and sex. Bars show the proportion of population: indoors awake not under bed net (dark-blue, indoors asleep without bed net (grey), indoors asleep under bed net (squared/crossed hatched), outdoors not under bed net (green). Line graphs indicate directly measured biting rates: indoors (blue) and outdoors (green). Panels represent: A general population, B males, C females, D children < 5 years, E ages 5–15 years, and F individuals aged > 15 years Human behavior adjusted biting rates (community exposure) Overall, of the 15% of observed exposure occurring indoors, 10% took place while individuals were awake and not using a net, while 5% occurred among individuals sleeping without ITNs. Observed exposure varied markedly by age—individuals aged (> 15 years) experienced the highest proportion (95%) of outdoor exposure, while children under 5 years of age exhibited greater indoor vulnerability, with 29% of observed exposure occurring indoors while awake and 18% occurring indoors while asleep but not under ITNs. Sex-specific differences were also observed. Women were more likely than men to experience indoor exposure while awake (19% vs. 8%) and not under ITNs (Fig. 3A–F). Human behavioral-adjusted biting exposure to An. arabiensis by age and sex. Behavioral-adjusted exposure is shown for bites occurring indoors awake (blue-dark) and asleep not under bed net (grey), outdoors without bed net (orange), alongside bites prevented indoor though ITNs use (hatched area). Panels represent: A general population, B males, C females, D children < 5 years, E ages 5–15 years, and F individuals aged > 15 years Gap in protection and protection offered by ITNs The gaps in protection analysis reinforced patterns of exposure that are not adequately addressed by current intervention (ITNs). In the general population, most behaviour-adjusted exposure occurred outdoors while individuals were awake (69%), whereas smaller proportions occurred indoors without ITNs protection, including 8% indoors while awake and 5% indoors while asleep. Thus, protection gap analysis indicates that ITNs provided an estimated 18% reduction in potential vector bites, highlighting that a substantial proportion of mosquito bites occurs outside the effective range of ITNs. This proportion of vector bites prevented varied across population groups, with relatively higher among children under five with 38% and progressively lower among older children and adults, 23% and 18% respectively. Similarly, the proportion of bites prevented was differences between males (17%) and females (28%) (Fig. 4, Panel A–F). Gaps in protection and protection offered by ITN. Human behavior adjusted exposure rate to An. arabiensis indoor awake (blue) and asleep not under bed net (grey), outdoor without bed net (orange) is shown alongside with proportion of bites prevented by ITNs use (hatched area). Panels represent: A general population, B males, C females, D children < 5 years, E ages 5–15 years, and F adults > 15 years Pattern of human exposure to mosquito bites (both direct and remaining exposure) and ITNs utilization by age and sex groups Clear differences were observed across age groups, the proportion of vector bites prevented by ITNs declined markedly with increasing age. Among children under 5 years of age, indoor exposure represented a large share of bites 53.4%, and ITNs prevented 38.4% of indoor mosquito bites, reducing overall exposure by nearly half, while the remaining exposure among ITNs users when nets were not in use was 44.7%. Among children aged 5–15 years, indoor exposure accounted for 19.4% of bites, while ITNs prevented 23.6% of potential indoor exposure leaving 37.5% of exposure remaining. Individuals aged less than 15 years were observed to experience predominantly outdoor exposure (95.1%), with ITNs preventing only 17.9% among adults, leaving 34.4% of remaining exposure. Sex-specific differences were also observed. Observed indoor exposure was higher among women (23.0%) than men (10.9%). Correspondingly, ITNs were observed to prevent a higher proportion of indoor bites among women compared with men (28.1% vs. 17.1%). Despite this, women experienced a greater proportion of remaining observed exposure when compared with men (40.6% vs. 29.6%) (Fig. 5). The gaps in protection analysis further reinforced these patterns (Fig. 4, Panel A–F). Patterns of human exposure, represented by bars, to mosquito bites (both direct and remaining exposure) and ITNs utilization stratified by age and sex groups Discussion This study identified An. arabiensis as the predominant malaria vector on Pemba Island, exhibiting substantially higher outdoor biting activity compared with indoor biting. This finding is consistent with previous studies conducted in Zanzibar and elsewhere in Africa [9, 10]. The dominance of An. arabiensis and its predominantly outdoor biting behavior indicate a shift in malaria transmission risk from indoor to outdoor settings. This species is known for its behavioral plasticity and generally considered a less efficient malaria vector compared with highly anthropophilic and indoor-adapted vectors such as An. gambiae s.s. and An. funestus, particularly because of its more opportunistic feeding behavior, greater tendency toward outdoor biting and resting, and capacity to utilize alternative blood-meal hosts. Consequently, An. arabiensis often exhibits lower sporozoite rates and lower vectorial capacity under many transmission settings [27–30]. However, in the context of sustained high coverage of indoor-targeted interventions such as ITNs and IRS, the relative epidemiological importance of An. arabiensis increases substantially because it’s behavioral ecology allows it to maintain residual transmission outside the primary protective range of these interventions. While it may be a comparatively weaker vector intrinsically, its persistence and predominance under current intervention pressure make it highly relevant for understanding and addressing residual malaria transmission [31–34]. These adaptations facilitate residual malaria transmission by increasing human exposure during early evening and early morning hours, when individuals are less likely to be protected by bed nets, particularly in the presence of low-density or asymptomatic parasite reservoirs within the community [25]. Marked heterogeneity in exposure patterns and levels of protection was observed across age groups and sexes. Various early evening and morning activities (not directly measured in this study) contributed substantially to exposure to malaria vectors. Similar associations between outdoor exposure and social or cultural activities have been reported in other settings [35]. In the cultural context of Pemba Island, many individuals are awake in the early morning hours preparing for morning prayers (Swalatul-Fajir), further increasing exposure during periods when mosquito activity remains high. These findings align with previous studies documenting the overlap between human activity patterns and mosquito biting behavior [4, 6, 14, 36] and highlight how such interactions sustain residual malaria transmission even in areas with high coverage of indoor vector control interventions. Our findings are also consistent with a systematic review synthesizing evidence on the role of human behavior in residual malaria transmission across sub-Saharan Africa [5]. Although ITNs are highly effective and widely used interventions [37, 38], their protective impact is strongly influenced by human behavior, including who uses nets and when they are used. Consistent with findings from malaria indicator surveys across sub-Saharan Africa, the proportion of vector bites prevented by ITNs in our study was highest among children under five years of age compared with older age groups [39]. Reduced use among older children and adults may therefore facilitate continued malaria transmission by vectors such as An. arabiensis, which exhibit substantial outdoor and early evening biting behaviour, as previously highlighted by Musiba et al. [10]. These behavioral patterns correspond closely with the protection gaps identified in this study, where a substantial proportion of mosquito exposure occurred outdoors or during periods when individuals were awake and therefore not protected by ITNs. Behavior-adjusted biting rates varied across the population, reflecting differences in both human activity patterns and change in vector composition. While most exposure occurred outdoors among individuals who were awake, children under five years of age experienced proportionally more indoor exposure, both while awake and asleep without a bed net. Similar age-specific patterns have been reported in other studies, where younger children experienced greater indoor exposure due to earlier bedtimes and limited outdoor activity [40]. Sex-specific differences in exposure were also evident and could be reflected to various activities Women were more likely to experience indoor exposure while awake, largely and might be due to natural homebased activities, thereby increasing vulnerability to indoor biting [12]. In contrast, men experienced a higher proportion of outdoor exposure. These findings reinforce evidence from other studies showing that even among individuals reporting consistent ITNs use during sleeping hours, substantial exposure can occur during early evening and early morning periods when nets are not in use [10, 40]. Similar behavioral–entomological overlap has been documented in Madang Province, Papua New Guinea, where vector biting shifted toward earlier hours following mass ITNs distribution [41]. Together, these patterns highlight how the interaction between human behavior and vector activity can sustain residual malaria transmission. ITNs remain one of the most effective malaria control tools; however alone they do not provide complete protection against malaria transmission. A study from Ghana demonstrated that ITNs alone reduced malaria transmission by only 27%, whereas combined deployment with IRS increased reduction to 58% [42]. Reduced effectiveness of ITNs has been attributed to multiple factors, with utilization patterns linked to human behavior representing a major contributor to protection gaps [43]. The study findings highlight that human strongly influence exposure to malaria vectors. Most bites occurred outdoors while individuals were awake, particularly among adults and males engaged in evening or early morning social and occupational activities [44]. In contrast, children and women experienced relatively more indoor exposure due to earlier bedtimes and domestic routines. These behavioral differences contribute to persistent protection gaps and residual transmission despite widespread ITNs coverage. Similar conclusions were drawn by Cooke et al. [15], who reported that ITNs prevented only approximately half of total exposure, largely due to a substantial proportion of mosquito bites occurring before sleeping hours. Although outdoor exposure represented the largest protection gap, indoor exposure among individuals who are awake or asleep without bed nets also remains substantial. Protection tend to decrease with age due to declining ITNs adherence [43], increasing vulnerability during both indoor and outdoor periods when nets are not used. Comparable patterns have been observed in other malaria endemic settings [12]. Lobo et al. [45] further highlighted that while ITNs are highly effective against late-night, indoor-biting mosquitoes, their overall impact can be constrained by incomplete coverage and behavioral shifts in vectors, including earlier and outdoor biting. Consistent with this evidence, our findings show that most indoor exposure occurred when individuals were not using ITNs, either while awake or asleep. This observation underscores that the effectiveness of ITNs depends not only on coverage but also on consistent and appropriate use [40]. Studies conducted in Zanzibar have similarly reported that overall population-level protection from ITNs remains limited, with approximately 40% of exposure prevented [32], while other studies have documented net use in only about half of the population [17]. Analyses across 19 sub-Saharan African countries further confirm that ITNs utilization is a key determinant of malaria risk and may be influenced by misconceptions and behavioral factors about net use [46, 47]. Consequently, ITNs alone may be insufficient to eliminate exposure, particularly during peak biting period that occur outdoors or before sleeping hours. Study limitation This study has several limitations that should be considered when interpreting the results. First, mosquito sampling was carried out on only two nights per month per household, which may not fully represent temporal variation in mosquito density and biting behaviour driven by factors such as weather conditions, lunar cycles, or other environmental influences. Second, human behaviour observations were collected manually and may have been affected by observer bias (Hawthorne effect), where individuals alter their behaviour when being observed. Third, the specific activities undertaken during human behavioural observations were not recorded in detail. Finally, observations were restricted to an approximate 10-m radius around the study households, potentially missing activities occurring elsewhere in the community that could contribute to exposure. Despite these limitations, the integration of entomological and behavioural data provides useful insights into how behavioural patterns of human interact with mosquito biting behaviour shaped by species composition, thereby improving understanding of exact point of human-vector interaction and gap in protection to the existing malaria interventions that can result the persistence of residual malaria transmission. Conclusion ITNs remain an important protective measure, particularly for young and vulnerable populations. However, our findings suggest that, ITNs do not fully cover the substantial proportion of exposure occurring outdoors. By integrating entomological data with observed human activity patterns, this study suggests that malaria exposure in Pemba Island is strongly influenced by the interaction between vector species composition and patterns of human behaviour—especially the outdoor biting tendency of An. arabiensis—and human activities occurring during early evening and early morning hours. Progress toward malaria elimination on Pemba Island will therefore require complementary strategies beyond ITNs, including integrated approaches targeting outdoor transmission, such as larval source management, spatial repellents as recommended by the World Health Organization, and community-based interventions aligned with local behavioural patterns. Incorporating age- and sex-specific considerations into control strategies will be important for addressing existing exposure gaps. Continued integration of entomological surveillance with human behavioural monitoring will be essential for identifying changing risk patterns and informing adaptive malaria control efforts in Pemba and similar elimination settings. Supplementary Information Supplementary material 1. Supplementary material 2. Acknowledgements We acknowledge the Ministry of Health Zanzibar (MoHZ) for the approval of using secondary data. We also, extend their gratitude to the management of the Zanzibar Malaria Elimination Program (ZAMEP) for providing access to the valuable data from storage files. Finally, we sincerely appreciate the assistance of Shehas from respective study Shehia for their support and co-operation in allowing the team for the access to the Shehia and head of the households for allowing the team to conduct mosquito collection in their houses. Last but not list, we are grateful to the invaluable contributions and support from various people and institutions. Abbreviations Human behaviour observation Human biting rates Human landing catch Indoor residual spraying Insecticide-treated nets Long-lasting insecticide-treated nets Zanzibar Malaria Elimination Programme Author contributions Conceptualization and implementation of the study: AOA, BOK, KFA, FAM, TSK, MKR, SG, SJS, MMK, KOA and NJG. Data management: AOA, JJJ and KAO. Data analysis: AOA, JJJ, SG, NJG and NFL. Drafting the manuscript: AOA and NFL. Review and editing of manuscript: All authors. All authors read and approved the final version of the manuscript. Funding This study was part of the routine activities conducted by Zanzibar Malaria Elimination Program (ZAMEP) that was financially supported by American people through the U.S President`s Malaria Initiative (PMI) under the terms of USAID/PSI Cooperative Agreement Number 72062122CA0008. Data availability No datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate The study was approved by the Ministry of Health Zanzibar (MoHZ), and permission was granted to use secondary data from the ZAMEP software and other sources. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.