Background

Malaria remains a threat to global public health, with 249 million cases and 608,000 deaths in 2022 [1]. The WHO African region is disproportionately affected, and approximately 78% of malaria-related deaths in the region were in children aged < 5 years [1,2,3,4]. More than half of all Ethiopians, mainly those in rural areas, are at risk of contracting malaria [5,6,7]. Unlike most African countries, clinical malaria of public health importance in Ethiopia is caused by both Plasmodium falciparum (P. falciparum) and P. vivax, which co-occur in all malarious areas, with the prevalence attributable to each parasite dependent on ecological settings and seasons [7, 8]. The transmission of malaria is highly variable due to the diverse eco-topography and climate conditions and, in general, transmission is bimodal, mainly occurring following the rainy seasons, “Kiremt” andBelg,” which are associated with major (long) and minor (short) transmission periods, respectively [8]. Since the 2000s, comprehensive preventative and case management interventions, including improved coverage of long-lasting insecticide-treated bed-nets and indoor residual spraying, the rollout and scale-up of artemisinin-based combination therapy, the deployment of a more sensitive and specific rapid diagnostic test (histidine-rich protein-2/3 [HRP2/3]-based test) and treatment at the grassroots level through health extension programs have achieved successive reductions in malaria burden [7, 9,10,11]. As a consequence, the aim of Ethiopia is to achieve zero indigenous malaria cases by 2030 [8, 12]. However, the country has experienced a nationwide resurgence in recent years and an unprecedented increase in case burden [1, 13]. Possible contributing factors include insecticide resistance in the primary malaria vector, the COVID-19 pandemic, the emergence of the HRP2/3 deletion, deterioration of the healthcare system, internal conflicts and invasion by the exotic malaria vector, Anopheles stephensi [14,15,16,17].

Until recently, about 46 Anopheles species and subspecies were recorded in Ethiopia [18, 19]. However, only a few Anopheles species, including An. arabiensis, An. pharoensis, An. funestus and An. nili, were incriminated as vectors of malaria [20]. Anopheles arabiensis is the primary malaria vector in most malaria-endemic areas [21,22,23]. However, its abundance, host preference and Plasmodium sporozoite rate vary across ecological gradients and epidemiological settings [22,23,24,25]. Anopheles pharoensis is of secondary importance [20, 23], with An. funestus and An. nili playing lesser roles in malaria transmission [22, 23].

Anopheles stephensi is an efficient urban malaria vector in southeast Asia and the Gulf Region [26] but is currently expanding its geographical range in Africa, where there have been reports of its presence in Djibouti [27], Ethiopia [17], Sudan [28], Somalia [29] and, more recently, Nigeria, Eritrea, Ghana and Kenya [30]. Anopheles stephensi is known to readily invade urban environments, and immature stages thrive in artificial aquatic habitats, with the consequent potential to increase malaria incidence in cities [27, 31] or reintroduce the disease into regions where it has been successfully eliminated [32,33,34,35].

In Ethiopia, since the first detection of An. stephensi in Kebri Dehar, Somali region [17], surveillance has confirmed its presence in the central, northeast, northwest and southwest parts of the country [36, 37]. Recent studies have shown that An. stephensi is a permissive host to P. falciparum and P. vivax infection [16, 37]. The results of a study from Dire Dawa City, eastern Ethiopia, suggest that An. stephensi was responsible, at least in part, for a malaria outbreak [38]. Similarly, in Djibouti, an upsurge in malaria incidence was observed following the detection of An. stephensi, providing further evidence for the potential for increased risk [39, 40]. In line with the WHO’s call for strengthened entomological surveillance of An. stephensi [41], this study aims to update current data on the distribution of An. stephensi across Ethiopia and to increase understanding of the patterns of household exposure to An. stephensi across the country.

Methods

Study area

Twenty-six urban centers were selected for this study (Fig. 1). A range of variables, including ecological setting, presence of dry ports and major transportation corridors, An. stephensi habitat suitability modeling and malaria endemicity, were considered when selecting the sites [36, 42]. The study sites (Additional file 1: Table S1) are located at altitudes between 339 and 2355 m a.s.l. and range from hot-desert lowland to humid highland environments. The mean annual temperature ranges from 30.9 °C in Afambo, the northeastern tip of the country, to 15.6 °C in Akaki, central Ethiopia, and the mean annual rainfall ranges from 224 to 1883 mm2 [43].

Fig. 1
figure 1

Map of study urban centers, with the colored dots indicating Anopheles stephensi-positive (red) and -negative sites (green)

Study design

Spatiotemporal distribution of An. stephensi

Four rounds of adult and immature stage collections were conducted at approximately 3-month intervals from November 2021 to January 2023. At each of the 26 study sites, a preliminary survey was first conducted to locate potential aquatic habitats of Anopheles mosquitoes outside compound/property limit of human dwellings/households. Based on the availability of aquatic habitats, we delineated a 1- × 1-km quadrant for entomological sampling (Fig. 2). In each quadrant, 20 households were selected for adult mosquito collection, of which four households were purposively selected based on their proximity to a major aquatic habitat and four were randomly selected in different directions from each of the four purposively chosen dwellings. Immature-stage mosquitoes were collected from all potential aquatic habitats within the compounds/property limits of the households selected for adult collection as well as from the purposively identified aquatic habitats within the quadrant beyond the compounds/property limits of the selected dwellings. Catches from the immature-stage collections were pooled by habitat type (either artificial or natural) and reared to adults for morphological species identification as described below. Written informed consent was obtained from the head of each household prior to mosquito collection. To increase the probability of detecting An. stephensi, adaptive sampling was employed [44, 45]. Thus, 50% of the households for adult collection were replaced randomly in subsequent collection rounds.

Fig. 2
figure 2

Schematic showing the approach used to select habitats and households for Anopheles mosquito collection, with the aim to study the spatiotemporal distribution of An. stephensi in Ethiopia

Household exposure to An. stephensi

Four urban centers where An. stephensi was detected were selected for more detailed study: Awash Sebat Kilo (Afar Regional State), Danan (Somali Regional State), Metehara (Oromia Regional State) and Jiga (Amhara Regional State). At each selected site, adult mosquitoes were collected from 50 randomly selected households and their surroundings in December 2022 and February 2023. Using a map of urban centers, a household located toward the center of the town was selected first, and then four additional households were selected (at approximately a distance of 100 m, with each of these four dwellings located in a different direction from the first household). This approach was repeated 9 times, to select the remaining 45 households systematically (Fig. 3). Adult Anopheles collections were conducted both indoors and outdoors from the selected households using Centers for Disease Control and Prevention light traps (CDC LTs) and from any structures serving as potential resting places within the household’s property compound using a Prokopack aspirator (John W. Hock Company, Gainesville, FL, USA).

Fig. 3
figure 3

Diagrammatic representation of household selection for studying exposure to An. stephensi. Large shaded circles indicate the selected households together with additional structures within a 50-m radius for entomological collection. Shaded circle designated with A indicates the first selected household (in red) located toward the center of the town. Shaded circles designated with BE indicate 4 additional households that were selected by moving 100 m (blue dotted line) in four directions from the first selected household. The remaining 45 households were selected by repeating this approach 9 times. The numbers in the blue circle indicate different types of structures within a 50-m radius of the selected household for adult or immature-stage mosquito collection

Adult mosquito collections

Trained field workers collected adult mosquitoes from 06:00 to 08:00 h and from 17:30 to 19:00 h with Prokopack aspirators. Indoor resting mosquito collections were conducted along the walls and ceilings of houses and under or behind household furniture. Outdoor collections were made from vegetation, tree trunks or walls of water containers located outside of the selected structures. The collections were made for 15–30 min in each structure, with the time scaled to the number and size of the structures/areas. The head of the household was requested to avoid activities that could repel mosquitoes on the night before collection, such as smoking inside the house, using repellents or spraying aerosol insecticides [46, 47].

Host-seeking adult mosquitoes were collected both indoors and outdoors using CDC LTs. Indoor traps were set near a sleeping space at the foot edge next to an existing bed-net, and corresponding outdoor traps were set within 5–10 m of each household selected for the indoor collection. Traps were set approximately 1.5 m from the ground, both indoors and outdoors. At approximately 18:00 h, the battery was connected to run the trap, and the following morning, at 06:30 h, the trap was removed. The collection cup was tied off securely prior to the trap being switched off and then labeled with the associated household code, whether an indoor/outdoor collection and with the date of collection before being removed from the trap and the mosquitoes retrieved.

Immature-stage mosquito collection and rearing

Aquatic habitats within a 50-m radius of the selected households were surveyed for immature stages, with the collections made using dippers and pipettes [48]. The collected immature-stage mosquitoes were subsequently transported to temporary field insectaries in plastic jars where they were transferred into enamel trays containing water from their respective aquatic habitat and larval food; the trays were check each day for pupation. Pupae were collected into beakers containing water from the respective aquatic habitat and placed in rearing cages with cotton balls soaked in a 10% sugar solution at the top of the cages. When all the pupae emerged, the beakers were removed from the rearing cages.

Morphological identification and preservation of mosquito samples

All wild-caught adults and those reared from immature stages were sorted by genus and sex after anesthetization with 70% alcohol. Only female Anopheles and Aedes mosquitoes were counted and recorded. Female mosquitoes belonging to the genus Anopheles were identified morphologically to the species level. Anopheles mosquitoes with speckled legs, with maxillary palpus with two apical pale bands very broad, with speckling on palpus segment three and with second main dark area on wing vein one with two pale interruptions were identified as An. stephensi [49]. Wild-caught female adult Anopheles mosquitoes were sorted based on their abdominal status as freshly fed or unfed, then preserved individually in a 1.5-ml Eppendorf tube with holes at the top and sealed in Ziplock bags containing silica gel. The immature-stage (larvae and pupae) collections were pooled by the household and preserved in absolute ethanol for molecular identification of An. stephensi.

Molecular procedures

The head and thorax of mosquitoes were separated from the abdomen for molecular analyses carried out at the Armauer Hansen Research Institute (AHRI) and Jimma University (JU) Tropical Infectious Diseases Research Center (TIDRC) laboratories. The bisected parts or pooled immature-stage materials were homogenized using a BioSpec BeadBeater apparatus (Bead Homogenizer 96 Microplate; Biospec Products, Bartlesville, USA) in 150 µl of molecular-grade water with 0.2 mg of beads (diameter: 1.0 mm; composition: zirconia/silica; BioSpec Products) at 3800 rpm for 20 s. DNA was extracted from 50- and 100-µl samples of homogenate according to the manufacturer's instructions (Qiagen GmbH, Hilden, Germany) and used for the detection of Plasmodium infection (head-thorax), blood meal analysis (abdomen), molecular species identification (immature stages) and confirmation of morphological identification (adult stage).

A PCR endpoint assay targeting the internal transcribed spacer 2 region (ITS2) was performed for species identification of An. stephensi as described elsewhere [50]. Briefly, the PCR was performed using the ST-F (5’CGTATCTTTCCTCGCATCCA3’), U5.8S-F (5’ATCACTCGGCTCATGGATCG3’) and UD2-R (5’GCACTATCAAGCAACACGACT3’) universal primers in a total reaction mixture of 25 µl containing 1 µl of DNA template, 0.25 µM each of ST-F and U5.8S-F primers, 0.37 µM of the UD2-R primer, 0.25 µM of dNTPs, 1.5 µM of Mg and 0.5 µM of Taq polymerase). PCR cycling conditions were: 95 °C for 30 s; 30 cycles of 95 °C for 30 s, 55 °C for 30 s and 68 °C for 7 min; with a final extension step at 68 °C for 7 min. The identification of specimens as An. stephensi was based on the visualization of a 438-bp band following gel electrophoresis of the PCR products.

Detection of Plasmodium infection

Plasmodium parasites in wild-caught Anopheles mosquitoes were identified either by PCR (AHRI laboratory) or enzyme-linked immunosorbent assay (ELISA) (Jimma University TIDRC laboratory). The PCR used to detect Plasmodium parasites targeted the cytochrome c oxidase subunit 1 (COXI) mitochondrial gene, and the presence of the parasites was based on the visualization of a 540-bp region following gel electrophoresis of the PCR products, as described elsewhere [51, 52]. Briefly, the PCR was performed in a total reaction volume of 25 µl containing 3 µl of DNA template from the head and thorax, primers (0.25 µM), dNTPs (0.2 µM), Mg (1.5 µM) and 1 unit of Taq polymerase. For those samples which tested PCR-positive for COXI, a nested PCR was run targeting the small 18S subunit of P. falciparum and P. vivax [53]. The first amplification reaction (nested-1) was performed in a final reaction volume of 25 µl containing 3 µl of DNA template, and the second reaction was performed by using a 5 µl amplicon of nested-1. The presence of P. falciparum and P. vivax was confirmed upon visualization of 205- and 120-bp bands, respectively, following electrophoresis of the PCR products on 1% agarose gel.

To detect Plasmodium via ELISA, the head-thorax of each mosquito was separated from the abdomen and ground in blocking buffer containing IGEPAL CA-630 (Sigma-Aldrich, St. Louis, MO, USA) in a 1.5-ml grinding tube. Antibodies against the circumsporozoite protein (CSP) of P. falciparum, P. vivax-210 (Pv-210) and P. vivax-247 (Pv-247) were detected using a sandwich CSP-ELISA [54, 55]. More specifically, 50 μl of species-specific capture monoclonal antibody (mAb) was added to each well of a micro-ELISA plate. The binding sites were blocked by adding 200 µl of blocking buffer and incubating the plate for 1 h at room temperature. Then, 50 μl of mosquito homogenate or control sample was added to the respective labeled wells. The plates were subsequently covered and incubated for 2 h at room temperature, following which 50 μl of peroxidase-linked mAb were then added to the wells and the plates incubated for 1 h at room temperature. Finally, 100 µl of peroxidase substrate solution (ABTS [2,2’-azino-bis 3-ethylbenzothiazoline-6-sulfonic acid]; Kirkegaard and Perry Laboratories [KPL], Gaithersburg, MD, USA) was added and the plate incubated for 30 min for P. falciparum and 60 min for P. vivax. Between the addition of each reagent or sample and the incubation, the wells were shaken and washed. Absorbance was read at 405–411 nm using an ELISA reader (model ELX800; BioTek, Winooski, VT, USA) after 30 and 60 min of incubation for P. falciparum and P. vivax, respectively. Samples with a value higher than twofold of the mean absorbance value of the negative controls were considered to be positive for Plasmodium parasites.

Detection of blood meal sources

The blood meal source of wild-caught, freshly blood-fed female Anopheles mosquitoes was analyzed by PCR (AHRI laboratory) or ELISA (Jimma University TIDRC laboratory). For the PCR, the extracted DNA from the abdominal region of female Anopheles was analyzed by a multiplex PCR assay as previously described [56]. The PCR assay was performed in a reaction volume of 25 µl containing universal vertebrate-specific and species-specific primers for pigs, humans, goats, dogs and cows [0.2 μM of each primer], 1× GoTAQ (Promega, Madison, WI, USA), 1.5 mM MgCl2, 0.2 mM dNTPs and 13.85 µl of molecular-grade water. The PCR amplification conditions were: 95 °C for 5 min; 40 cycles of 95 °C for 60 s, 57 °C for 60 s and 72 °C for 60 s; with a final extension of 72 °C for 7 min. The PCR products were confirmed by gel electrophoresis.

For the ELISA, the abdomens of freshly fed female Anopheles mosquitoes were homogenized in phosphate-buffered saline (PBS) in a 1.5-ml grinding tube following a standard protocol [55]. More specifically, 100 μl of homogenized sample or control was loaded added into a well of an ELISA plate and the plate incubated for 2 h at room temperature, following which the wells were washed 3 times with 200 µl of PBS-Tween 20 solution. Then, 50 µl of host-specific peroxidase-labeled mAb of human, bovine and goat (Sigma-Aldrich) was added and incubated for 1 h in the dark, followed by 3 washes with 200 µl of PBS–Tween 20, with the plate shaken 5 times with each wash. Finally, 100 µl of ABTS was added to each well as the substrate for the peroxidase enzyme, and the mixture was incubated for 30 min. The plates were observed visually, and their optical density was read at 405–414 nm using an ELISA reader (model ELX800; BioTek). Samples with a value higher than twofold the mean absorbance value of the negative controls were considered to be positive for Plasmodium parasites.

Data management and statistical analysis

The data were collected using tablets with data forms developed in REDCap [57, 58] and uploaded to the AHRI data server on a daily basis. Data were downloaded and cleaned using Microsoft Excel (Microsoft Corp., Redmond, WA, USA), and Microsoft Excel and Stata software release 14 (StataCorp LLC, College Station, TX, USA) were used for analysis. Only Anopheles mosquito catches identified to species level were included in the statistical analyses. Site positivity proportion was determined by dividing the number of sites at which Anopheles species were identified in at least one round of entomological surveys by the number of sites surveyed. To estimate catches per method of collection, method-specific catches of species were divided by the total number of that species caught. Relative abundance of Anopheles species was investigated per site as catches of specific species divided by the total adult-stage collections per site. The mean number of catches per trap was determined by considering the number of traps in all collection rounds. Household exposure was defined as the presence of either immature or adult stages of An. stephensi in a 50-m radius of surveyed households. The level of household-exposure was estimated as the fraction of the surveyed households per number of households detected with An. stephensi. Mixed blood meal sources were included in the nominator to determine the blood meal indices. Sporozoite rate was determined as the fraction of Anopheles species tested per that species detected with Plasmodium parasite by PCR and ELISA.

Results

Anopheles mosquito fauna and abundance

Among the study sites, 96.2% (25/26) were positive for Anopheles mosquitoes, including An. stephensi, and 99.2% (5353/5398) of the total catches were identified to species level. Of the total adult Anopheles catches, 86.6% (1612/1862) were collected by CDC LTs, and 13.4% (250/1862) were collected by Prokopack aspirators (Fig. 4), with 79.5% of adult catches identified as An. arabeinsis, 7.7% as An. stephensi and 7.4% as An. pharoensis. Anopheles coustani, An. tenebrosus, An. fuenstus and An. rufipes accounted for only 2.5%, 1.8%, 1.1% and 0.1% of adult catches, respectively. Anopheles stephensi was predominant in four sites of the 26 surveyed (Babile, Kebri Dehar, Danan and Modjo), with 77.3–100% relative abundance. Among the immature-stage collections, An. stephensi was predominant in five sites of the 26 surveyed (Babile, Kebri Dehar, Danan, Dubti, and Modjo), with 75–100% relative abundance. Anopheles arabiensis was the predominant species across most sites (20/26) (Table 1). In two study sites, Anopheles mosquitoes were either not found (Kebri Beyah) or not identified morphologically (Yabelo) due to damage during collections.

Fig. 4
figure 4

Anopheles mosquito catches. A total number of Anopheles mosquitoes collected and morphologically identified to species level. B proportion of adult Anopheles catches according to method of collection. CDC LT, U.S. Centers for Disease Control and Prevention light traps

Table 1 Distribution and abundance of Anopheles mosquitoes across study urban centers, Ethiopia, from surveys performed in 2021–2023

Spatiotemporal distribution of An. stephensi

In four rounds of entomological surveys, 30.8% (8/26) of the sites were positive for An. stephensi, immature and/or adult stages. Both adult and immature stages of An. stephensi were detected at six of these positive sites (Babile, Danan, Dubti, Jiga, Kebri Dehar, and Modjo), while at the remaining two sites either adults (Assosa) or immature stages (Ataye) were detected (Table 2). In four of the An. stephensi-positive sites, adult or immature stages were recorded every round. The mean number of wild-caught adult An. stephensi per trap was 0.15 (CDC LT: 0.04 per trap night; Prokopack aspirator: 0.38 per collection) (Additional file 1: Table S2). Anopheles stephensi larvae were collected from a range of artificial and natural aquatic habitats.

Table 2 Spatiotemporal distribution of Anopheles stephensi across study urban centers, Ethiopia, 2021–2023

Household exposure to An. stephensi

During additional extensive entomological sampling in and around households, An. stephensi was detected in three of the four urban centers, namely Danan, Awash Sebat Kilo and Metehara. Among the 50 surveyed households, 30% (15/50) were positive for An. stephensi in Danan. The household positivity rates were 26% (13/50) and 18% (9/50) in Awash Sebat Kilo and Metehara, respectively. Anopheles stephensi was not detected in any of the sampled households in Jiga during the sampling period (Table 3).

Table 3 Household exposure rates for immature, adult and both stages of Anopheles stephensi across four urban centers, Ethiopia, 2023

Species identification by molecular methods

A total of 80 field-caught adult mosquitoes identified as An. stephensi based on morphological characteristics and 28 pooled immature-stage mosquito samples were screened using molecular methods to confirm species identity. Of the 80 adult samples identified morphologically as An. stephensi, the molecular assays confirmed the identity as An. stephensi in 78 samples; in the remaining two samples, the products were either not amplified or were found to be non-An. stephensi. Among the pooled immature-stage mosquito samples, 82.1% (23/28 pools) were confirmed to contain An. stephensi.

Detection of blood meal sources

Blood meals were analyzed in 784 Anopheles mosquitoes, of which 386 samples were assayed by multiplex PCR (Additional file 1: Fig. S1) and 398 specimens were analyzed using an ELISA. Among Anopheles mosquitoes identified with a blood meal source, 76.9% (50/65) of An. stephensi and 56.2% (181/322) of An. arabiensis were found with single host blood. The blood meal indices for An. stephensi were 69.2% for bovines (BBI), 32.3% for ovines (OBI) and 24.6% for humans (HBI), including the mixed blood meal sources (Table 4); for An. arabiensis, the blood meal indices were 65.4% for bovines, 46.7% for ovines and 35.8% for humans.

Table 4 Determination of blood meal sources of wild-caught adult Anopheles mosquitoes collected across urban centers in Ethiopia using multiplex PCR and enzyme-linked immunosorbent assay, 2021–2023

Detection of Plasmodium infection

Among the 1847 Anopheles mosquitoes examined for Plasmodium infection, 69.4% (1282/1847) and 30.6% (565/1847) were assessed using PCR (Additional file 1: Fig. S2) and ELISA, respectively (Table 5). None of the 197 samples identified as An. stephensi were found to be infected with Plasmodium parasites. Overall, the Plasmodium sporozoite rate of An. arabiensis was 4.3% (62/1434), of which 16.1% (10/62) were P. falciparum and 83.9% (52/62) were P. vivax. In An. pharoensis, the Plasmodium sporozoite rate was 6.7% (8/119), of which 12.5% (1/8) were P. falciparum and 87.5% (7/8) were P. vivax. Two other Anopheles mosquito species were detected with sporozoites of P. vivax: An. coustani (12.8%, 5/39) and An. funestus (11.5%, 3/26).

Table 5 Sporozoite rate in wild-caught adult Anopheles mosquitoes collected across study urban centers, Ethiopia, 2021–2023

Discussion

The results of the present study add to the body of available evidence on the distribution and abundance of invasive An. stephensi in Ethiopia. Entomological surveillance in 26 urban centers between 2021 and 2023 revealed that An. arabiensis was the predominant Anopheles species in the catches, accounting for 79.5% of all collections, followed by An. stephensi, accounting for 7.7% of the total Anopheles catches. The relative abundance of adult An. stephensi was greater than that of An. arabiensis in Babile, Kebri Dehar, Danan and Modjo. Modjo is located along the main ground transportation route or corridor that connects Ethiopia to Djibouti [36]. Generally, adult An. stephensi collections were low (mean: 0.15 catches/trap), and most of the immature-stage collections were from artificial aquatic habitats.

Since the first detection of An. stephensi in eastern Ethiopia in 2016, new positive sites have been identified in subsequent surveys [36, 37, 59]. In line with these findings, we detected An. stephensi in western Ethiopia (Assosa), in an area bordering Sudan, which might indicate the continued spread of this species. We also found An. stephensi at all previously reported sites, as well as at new sites along its purported invasion route. However, the site positivity of An. stephensi was relatively low (8/26 sampled sites) compared to that reported in previous studies. A study in 2020 that covered 10 sites in eastern Ethiopia reported the presence of An. stephensi at all sites [36]. Similarly, sampling at 21 sites between 2018 and 2020 revealed 61.9% positivity for An. stephensi [37]. Another study conducted by the PMI Vector Link Ethiopia project showed the presence of An. stephensi in 16 urban settings, of which nine (56.3%) sites were newly positive for An. stephensi [60]. One explanation for the differences in results might be the selection of study sites; most of the collection points in these previous studies were purposefully chosen to detect An. stephensi, while the current study followed substantial random steps in the selection of study sites and collection points within sites. Our approach has the advantage of providing unbiased distribution estimates, but it does reduce the probability of detection. In addition, some of our study sites were located far from major transportation corridors [61], which are considered as the main invasion routes.

In line with the findings of other studies [16, 36], we also noted that An. stephensi was more readily detectable as immature stages than as adults in most of the positive sites. The highest proportion of An. stephensi collections (85.7%) was obtained as immature stages (larvae and pupae) in aquatic habitats. A range of aquatic habitats were positive; for example, in Dubti, immature stages of An. stephensi were detected in both artificial habitats (water tanks, barrels, buckets, tires) and natural habitats (ponds, streams, swamps and marshes). It has been reported that An. stephensi can breed in various aquatic habitats with differing physicochemical characteristics, such as salinity and turbidity [62]. In Modjo, Danan, Kebri Dehar and Babile, immature stages of An. stephensi were detected only in artificial habitats. This variation highlights how larval source management of An. stephensi, which has been recommended by the WHO [63] and is being implemented by PMI VectorLink and others, will be more complex than simply targeting container habitats.

In the current study, An. arabiensis was the most abundant species at 20/26 sites, which is in line with the findings of other studies showing that this species is still the predominant malaria vector in different eco-epidemiological settings of Ethiopia [20, 25, 64]. Even though An. arabiensis is considered less adapted to urban ecology [65], our findings suggest that it is likely to be the primary malaria vector in urban centers in Ethiopia. The other Anopheles species collected in this study were An. pharoensis, An. coustani, An. funestus, An. tenebrosus and An. rufipes, which together accounted for 12.8% of the total adult Anopheles catches. Of these five species, An. pharoensis and An. funestus have been reported to be secondary or suspected malaria vectors in Ethiopia [66]. We detected An. pharoensis infected with P. falciparum or P. vivax at one and five of our study sites, respectively, while An. coustani and An. funestus were detected with P. vivax sporozoites across four and three of the study sites, respectively.

The level of household exposure to An. stephensi was heterogeneous across the study sites, with household positivity for both (adults and immature) stages ranging from 18% in Metehara to 30% in Danan. The level of household exposure to adult An. stephensi was highest in the region where this species was first reported as an invasive species (Danan) and lower in more central parts of the country (Awash Sebat Kilo and Metehara). A similar trend was observed for household exposure to the immature stages of An. stephensi. These results could be due to well-established populations of An. stephensi in areas where it was first reported since its invasion.

Our findings reveal that An. stephensi prefers non-human vertebrate hosts for their blood meal. The most prevalent blood meal among An. stephensi detected with sources of blood was cattle (69.2%), followed by goats (32.3%). These findings are consistent with results previously reported in Ethiopia [16, 37] and India [67], which showed that most An. stephensi fed on livestock. In the present study, one-third of An. stephensi fed on unidentified blood meal sources, which might be due to a lack of host antibodies or primers for blood meal analysis. It is noteworthy that at some of the study sites, especially in eastern Ethiopia, the most readily available animals were camels and that these sites were where most of the tested An. stephensi were collected. Despite the relatively high non-human vertebrate host blood meal indices, 24.6% of An. stephensi were found with human blood, including mixed blood meal sources. The blood meal source of vectors might be affected by multiple factors, including host availability and proximity, possibly explaining why 76.9% of the An. stephensi and 56.2% of An. arabiensis fed on a single blood meal source of either an animal or a human host.

Of the 197 screened An. stephensi, none were detected with Plasmodium parasites. These findings are similar to those of another study in which none of the tested An. stephensi was positive for Plasmodium [36]. However, a study conducted in 2019 in Awash Sebat Kilo reported an infection rate of 2.8% for P. vivax and 1.4% for P. falciparum, based on analysis of homogenates of whole mosquitoes [16]. The authors of another study in which the heads and thoraxes was used to detect Plasmodium by ELISA reported that the sporozoite rate was 0.5% in Dire Dawa and 0.3% in Kebri Dehar for P. vivax [37]. The most recent study, from Dire Dawa, implicated An. stephensi in an outbreak and detected a P. falciparum sporozoite rate of 1.2% [38]. It should be noted that our study might be limited in its ability to elucidate An. stephensi sporozoite infection, as a large proportion of the adult catches (91/146) were from a single aquatic site using the Prokopack aspirator.

There are a number of limitations to our study. The limited number of An. stephensi adults caught indicates the need for further studies to investigate more efficient trapping methods. We employed both PCR and ELISA for detecting blood meal sources and Plasmodium infection in Anopheles mosquitoes, which might limit the direct comparability of our findings across study sites. The collection rounds did not directly coincide with the malaria transmission seasons in Ethiopia, and some of the sites were excluded due to civil unrest.

Conclusions

Our findings reveal an expansion of An. stephensi into a new western geographical range and its transition to predominant species status in some areas where it was first detected. These results highlight the need for enhanced entomological surveillance with efficient traps to determine the bionomics and relative contribution of An. stephensi for malaria transmission in the region. In the meantime, the plan set forth to limit the spread and contain An. stephensi establishment should be put into action.