These findings highlight potential entomological drivers contributing to residual malaria transmission in Zanzibar and the testing of complementary novel techniques that can be used to target outdoor biting and resting mosquitoes. Key findings include: a high proportion of outdoor biting and resting behaviour, high levels of zoophilic behaviour, and varying levels of pyrethroid resistance across sites. All these factors could contribute to ongoing local malaria transmission in Zanzibar.
After eight rounds of indoor and outdoor mosquito collection in 135 households across six Shehias in four districts only a total of 704 malaria vectors were collected. In most trap nights, the average number of malaria vectors caught was zero especially in dry season. This indicates that the mosquito density is very low in Zanzibar especially in the dry season. ZAMEP has also reported the very low numbers of malaria vectors collection from 2012 to 2018 (ZAMEP reports, unpublished data). Malaria vector reduction is likely to be attributed to the continuous use of LLINs as well as implementation of IRS in areas with high density of malaria vectors[27]. In addition, a majority of malaria vectors were An. Arabiensis (98.4%), mostly caught outdoors (85%), which is an opportunistic species [28–30] and it is not surprising that out of 173 blood-fed mosquitoes-39.30% obtained blood from human and 60.70% from non-human hosts (i.e., 34.68%, 21.97%, 3.47%,and 0.58% from Goat, Bovine, Dog and Chicken respectively). In general, a large significant number of mosquitoes with blood from non-human hosts were sampled from outdoor dwellings. In contrast, mosquitoes fed equally outdoor (50%) and indoor (50%) on human blood– exhibiting An. arabiensis feeding behavior [31, 32]. Similar findings on outdoor and indoor biting preference and malaria vector species composition were also obtained by ZAMEP as presented in their entomological report from 2012 to 2018 - indicating high number of An. arabiensis compared to An.gambiae s.s.[33]. This significant reduction of An. gambiae s.s in the island might be due to wide use of indoor interventions (i.e., (LLINs) and (IRS)) [27]. An. arabiensis is becoming a major malaria vector in Zanzibar and its increasing role in malaria transmission has also been reported in various studies in Tanzania [34] and in Kenya [35]. Sporozoite ELISA analysis indicates that all mosquitoes tested were negative for the malaria parasite. Nevertheless, significant outdoor biting behaviour can potentially indicate the high risk of outdoor malaria transmission in Zanzibar. In addition, it was also observed that the artificially created outdoor resting sites collected a large numbers of Anopheline mosquitoes as compared to the number collected resting indoors- and this has also been demonstrated in ZAMEP’s entomological reports (2013- 2017). This might also highlight the mosquito behavior change from resting in indoor dwellings e.g. ceiling, walls to outdoor dwelling like the dug hole and other areas with favorable resting conditions e.g. darkness. The implementation of indoor core interventions i.e. LLINs and/or IRS as previously demonstrated in some studies as one of the driving forces of mosquitoes behavioral change [31, 36, 37]).
In understanding that insecticide resistance may impact the effectiveness LLINs and IRS which are widely used in Zanzibar –insecticide resistance tests based on WHO procedures were carried out [20]. The tests were carried on commonly used insecticide based on samples from only four sites with the highest mosquito densities. The findings indicate that testing outcomes for pyrethroid class insecticides across the four sites were not uniform. The susceptibility tests indicated that mosquitoes were susceptible to some insecticides, displayed possible resistance to others depending on the insecticide tested. Insecticide intensity outcomes also varied across the study sites. Generally, the intensity of resistance across the four sites ranged from low intensity, moderate to high intensity. The susceptibility and intensity variation are also reported in ZAMEP reports for Unguja site but it is uniform in Pemba study sites where mosquitoes have displayed high level of resistance to pyrethroid class insecticide (ZAMEP report 2018). These variations might be explained by several facts including tendency of net use in the particular site, level of urbanization as well as agricultural activities. The use of different insecticides in agricultural activities might trigger high selection pressure of insecticide resistance in one site compared to another site [38, 39]. In contrast, mosquitoes were susceptible to all representative insecticides from the other three classes (i.e., organochlorine, organophosphate and carbamate) across all the four sites. The insecticide resistance carried out by ZAMEP in 2018 in three different sites in Unguja are consistent with our findings (ZAMEP report 2018, unpublished data).Insecticide resistance in African malaria vectors have also been reported in other malaria endemic settings [40–44]. The resistance levels and mechanisms to African malaria vectors have been documented all over in malaria endemic settings[44–47]. The wide spread of insecticide resistance affects both malaria control and transmission[48–51].This on-going mosquito resistance to pyrethroids highlights the need to consider other options for the insecticide used for LLINs and IRS programs.
Recently, the addition of a synergist, piperonyl butoxide (PBO), to pyrethroid nets has been investigated. Available evidence suggest that pyrethroid-PBO nets are more effective than standard pyrethroid nets in settings with insecticide resistance[52, 53]. While pyrethroid-PBO nets can help to maintain protection in some settings, they should not be considered a tool for insecticide resistance management, rather broader insecticide resistance management strategies are essential [WHO, 2017b].
Most entomological studies will normally calculate and report entomological inoculation rate (EIR) which is the practical indicator of human exposure mosquitoes bites infected with transmissible sporozoite-stage malaria parasites[54].This indicator can be used to assess the impact of vector control tools (VCTs) and it is important for VCTs to sustainably reduce EIRs to levels below 1 in order to interrupt malaria transmission[55]. In this study, EIR was not calculated due to the absence of any sporozoite positive mosquito. The findings from ZAMEP’s mosquito routine collections in Zanzibar indicated that five sporozoite positive Anopheles arabiensis mosquitoes detected by CSP ELISA have been confirmed in the last six years (2012 -2018) in Unguja. However, no positive sporozoite mosquitoes have been detected in Pemba in the same period, (ZAMEP report, unpublished data). This indicates that the transmission is declining to the extent that it is becoming difficult to estimate EIR or it might require using advanced techniques for sporozoites detection in low transmission settings.
A key limitation of the study was the inability to detect sporozoite rates from blood-fed mosquitoes collected. CSP ELISA was used for sporozoite detection and the study found no sporozoite positive mosquitoes. As a result, it was not possible to calculate the current EIR value (given sporozoite infection rate as a factor). This could reflect the characteristically low transmission of malaria in Zanzibar along with Anopheles mosquito density. Though, we anticipate that more advanced sporozoite detection methods (i.e., PCR ELISA) may have allowed for better detection of sporozoite rates. Higher resolution sampling in hot spot areas could be helpful in catching sporozite positive mosquitoes. These findings illustrate that, in areas of persistent low malaria transmission more powerful methods may be needed to detect sporozoite infection rates. In similar settings, entomological studies alone may be inadequate for making a conclusion regarding the presence of local malaria transmission but will require using other techniques such as detection of malaria parasite movement through genetic sequencing.
Another limitation of the study methodology is associated with the type of trap used to collect mosquitoes. Hourly mosquito collections were carried out in the peri-domestic setting using the double net trap (MD-Mini) [16] instead of human landing catch (HLC) which is often considered the gold standard[56]. Therefore, the recorded biting rates may have been impacted by the trapping method used. However, a previous study showed no significant difference between number of Anopheles mosquito caught by HLC and DN-Mini [16].The double net trap was used to increase the safety of volunteers during the collection by ensuring that they are not freely exposed to mosquito bites.
Despite these limitations, this study has contributed to understanding the magnitude and entomological drivers contributing to the residual malaria transmission in Zanzibar and areas with similar transmission settings. The study suggests increased outdoor-biting proportions, shifts in peak biting times to early-evening hours, and the rise of pyrethroid-resistant Anopheles vectors. Therefore, the findings from this study highlight the need to consider using bednets treated with other insecticide (e.g., a LLIN that incorporates a synergist piperonyl butoxide (PBO) [53], different insecticide for IRS programs in Zanzibar[53] and/or rotation of insecticide application mode [44], and proper implementation of larviciding which target mosquitoes at aquatic stages [57–59]. In addition, complementary tools enhancing indoor protection and especially targeting outdoor transmission have the potential to contribute to eliminating residual malaria transmission in Zanzibar.