High levels of resistance to pyrethroids were found in An. coluzzii, particularly in São Tomé island. Mortalities below 90% were obtained in exposures to 10\(\:\times\:\) the diagnostic dose. While this may be considered a predictor of operational failure, the real impact of the insecticide resistant phenotype in vector control remains uncertain [26, 31, 32]. The high levels of pyrethroid resistance found in STP may be a consequence of the intensive usage of this insecticide class for malaria control over the past 20 years in the country [3, 33, 34]. Permethrin has been used in STP since the 1990s in periodical campaigns of bed nets impregnation. Following the introduction of LLIN in 2004 two additional pyrethroids, deltamethrin and α-cypermethrin, were introduced and widely used in the country [35]. In addition, α-cypermethrin was the insecticide of choice for IRS in the National Strategic Plan to Control Malaria implemented in 2004, being used for two consecutive years [33].
When comparing the two islands, pyrethroid resistance was consistently lower in Príncipe island. São Tomé island, and Água Grande district in particular, concentrates over 90% of the ca. 220.000 inhabitants of the country. Furthermore, malaria prevalence has been consistently lower in Príncipe and this region has implemented since 2023 a regime of reactive indoor spraying as opposed to the large-scale IRS employed in São Tomé [6, 7, 36]. These differences may be explained by a lower insecticide selective pressure in Príncipe when compared to São Tomé island.
The presence of the L1014F kdr mutation coupled with partial reversal of the resistant phenotype when mosquitoes were pre-exposed to PBO suggest that pyrethroid resistance in An. coluzzii from STP is mediated by both target-site and metabolic mechanisms. The inhibition of cytochrome P450 oxidase activity by PBO and the consequent increase in mortality after exposure to permethrin or α-cypermethrin suggests a role of this detoxifying enzyme superfamily in the resistance phenotype. This type of metabolic resistance has been reported in continental populations of An. coluzzii, including in nearby west African countries, many times also with the concomitant presence of kdr mutations [37–39]. Among the cytochrome P450 oxidase genes mostly associated with pyrethroid resistance in this vector species are CYP6M2, CYP6Z2 and CYP9K1 [23, 38]. Further studies involving gene expression analysis will be required to ascertain the specific genes involved in the resistance phenotype of An. coluzzii in STP.
The pyrethroid resistance-associated L1014F kdr mutation was found in both islands albeit at higher frequency in São Tomé island, where positive associations between kdr-w and the resistant phenotype were found. This result agrees with other studies that show increasing frequencies of resistance-associated alleles and the development of resistance after deployment of insecticide-based vector control [40, 41]. In STP, the kdr-w mutation was not detected in collections performed in 2003 and 2004 (present study) suggesting a recent emergence of this resistance-associated allele. The first record of the kdr-w allele was in 2010 in São Tomé island, six years after the implementation of IRS with α-cypermethrin, whereas in Príncipe this mutation was reported for the first time in 2014 [33]. However, there is no information about genotyping of kdr mutations between 2004 and 2010 and the first report showed that the kdr-w mutation was already present in 4 districts of São Tomé island in 2010 [33]. The emergence of the kdr-w allele in STP may be a result of mutational events or introgression with kdr-carrying mosquitoes that may have accidently reached the islands. Both origins, including interspecific introgression, have been previously described throughout the African continent [16–20]. Further studies involving haplotype analysis of genomic regions surrounding the kdr locus would help clarify the origin of the kdr-w allele in STP.
Haplotype diversity analysis has also shown evidence for strong positive selection of the kdr-w allele in West Africa, which may explain the rise of this allele from 0.3–54% in An. coluzzii from Ghana [18]. However, such an increase was not so evident in An. coluzzii from STP. The initial report of Chen (2019) [33] describes kdr-w frequencies of 13% in 2010, reaching a peak in 2013 with 78% and decreasing to 40% in 2016. A second report shows kdr-w frequencies between 8% and 44% in three districts of São Tomé island sampled in 2018 [42]. These values contrast with the kdr-w frequencies found in this study (2022–2024) that did not exceed 16%. A decrease in kdr-w frequency may be explained by a relaxation of the selective pressure due to the interruption of pyrethroid-based IRS campaign in 2021 when a rotation scheme with non-pyrethroid insecticides started to be employed for IRS. It may also reflect a possible increase of the outdoor biting and resting preferences of this mosquito vector on the islands leading to a lower selective pressure [10, 11, 33].
Another important aspect was the differences found in susceptibility levels between seasons, with a significant trend of increased resistance in the rainy season. Dry season sampling bias due to lower number of breeding sites might contribute to these differences but this would be a more plausible explanation if resistance was found to be higher in the dry season, suggesting that more siblings harboring kdr alleles (or other resistance mechanisms) had been used in bioassays. However, the differences found between seasons on kdr genotype distribution were only marginally significant in Água Grande and do not seem to explain the observed phenotypic patterns. A more plausible explanation may be attributed to seasonal differences in the physiological/nutritional status of An. coluzzii. We argue that during the rainy season the greater availability of larval habitats suitable for mosquito breeding reduces inter- and intraspecific competition leading to a higher nutritional intake and resulting in adults with an average higher percentage of body fat tissue, known to be rich in detoxifying enzymes [43, 44]. This hypothesis needs confirmation as it contrasts with previous observations that dry season mosquitoes are more resistant than rainy season ones [45, 46].
The higher resistance levels found in the dry season are generally explained by the selective pressure of pollutants in larval habitats that lead to increased expression of detoxifying enzymes such as P450 oxidases [46]. This is more evident in rural areas with marked seasonality where agriculture insecticide usage occurs mainly during the dry season. This may not be the case for São Tomé, in which the short duration of the dry season may not be sufficient to promote a high concentration of xenobiotics in larval habitats.
Regarding non-pyrethroid insecticides, resistance to DDT was also observed in São Tomé (but not in Príncipe). This agrees with the historical use of DDT during the malaria eradication program in the 1980s and the emergence of resistance to this insecticide has been considered a major reason for disruption of this program [1, 24]. This may also reflect cross-resistance, as DDT shares with pyrethroids the same active site, the voltage-gated sodium channel, and the L1014F mutation confers resistance to both insecticide classes [47, 48].
Anopheles coluzzii populations were fully susceptible to pirimiphos-methyl. This result may be explained by the fact that pirimiphos-methyl was used only in one campaign in the country (2019), and subsequently implemented in the rotation scheme, which appears to work positively to prevent resistance from arising [34]. This result contrasts to that observed for pyrethroids when widely used for consecutive years. While mechanisms of organophosphate resistance are likely to be absent from these island populations, this may also reflect lower selective pressure associated with the outdoors behavior of this mosquito in STP.