1. IntroductionIn Brazil, many species of Culicidae (Diptera) are of medical interest [1,2] and much of the knowledge about their bioecology and epidemiological role is the result of research and entomological surveys associated with outbreaks of arboviruses [3,4,5,6,7]. In these studies, the mobile human attraction technique using capture hand-nets was widely used, often in entomological captures in the forest canopy, aimed at increasing the sample of mosquitoes that feed on the blood of birds and nonhuman primates, which are considered important hosts for several arboviruses [5,8,9]. For this reason, in Brazil, the hand-net was adopted as the standard technique by health surveillance services across the country, with canopy capture recommended as a complementary procedure [10].However, in captures with humans beings, the attractants are subject to variations in individual performance and the attractiveness of each collector, which can imply a bias resulting in sample divergences [11]. Moreover, collections performed by individuals raises concerns about occupational risks [12,13], especially when working at heights in tree canopies, according to the specific safety protocols of each country.Thus, there is a need for alternative techniques to replace the standard hand-net technique that can ensure representativeness in mosquito sampling. Service (1993) and Santos et al. (2021) point out that the selection of the appropriate method should take into account the specific characteristics of the natural history, biology, and ecology of the target species of potential vectors [14,15].One of the alternative techniques is the use of automatic suction traps with a luminous attractant, commonly used to collect nocturnal species [13,16]. In Brazil, the association of a carbon dioxide source and other types of odor attractants (such as octenol and lactic acid) have been used to increase the sensitivity of this type of trap for daytime use [17,18,19,20,21]. Other studies were restricted to afternoon and nocturnal periods [22,23,24,25,26], or focused on the urban environment, for capturing Aedes aegypti and Culex quinquefasciatus [27,28,29].Recently, our group began looking for alternative and/or complementary methods for daytime collections in accordance with forest stratification to carry out routine surveys in different biomes. In a previous study, we conducted a comparative evaluation of in-person capture using hand-nets and automatic traps with carbon dioxide (dry ice) and BG-Lure® as attractants on the ground and in the canopy stratum of an Atlantic Forest environmental reserve [21]. In the present study, we apply the same methodology used in the previous study for a comparative evaluation of these methods in the Cerrado biome. 3. ResultsWe collected a total of 1555 Culicidae specimens in 20 taxa, of which 17 species belonged to 8 genera. Only 24 specimens were males, corresponding to the genus Culex and to the species Aedes albopictus and Psorophora albigenu (Table S1). Eighty-five percent of the specimens were captured from the ground.Aedes scapularis was the most abundant species, followed by Haemagogus janthinomys/capricornii, Psorophora albigenu, Sabethes albiprivus, and Ae. albopictus (Table 1). All species, except Limatus durhamii and Sabethes belisarioi, were more abundant at the ground level (n = 1302) than in the canopy (n = 229), including Hg. janthinomys/capricornii and Sa. albiprivus, which were the dominant species in the canopy. Haemagogus leucocelaenus was a rare species at ground level, with only two specimens collected (Table 1). The hand-net capture technique obtained significantly higher yields than the automatic traps, both in the canopy and on the ground, accounting for 78.6% of the total of specimens collected in the canopy and 72.1% on the ground (Table 1 and Table S2). The traps using only CO2 and CO2 + BG-Lure® obtained an intermediate yield and showed no significant difference between them in relation to the same stratum (Table S2). Traps using exclusively the BG-Lure® attractant had the lowest yield in both strata (Table 1, Table S2 and Table S3). Regarding the indicators for richness, diversity, and dominance (Table 2), we observed a qualitative similarity between the canopy and ground strata, as the differences between each method in the canopy had the same pattern on the ground. In addition, the values of the same method in the canopy were close to those on the ground. The differences were small between the values for absolute richness and abundance for the different techniques and attractants in each stratum and, in general, with greater richness on the ground compared to the canopy. The biggest difference was in the trap results associated with BG-Lure® in the canopy, where only one specimen was collected and, thus, presented very different indicators of the same method at ground level. Hand-net capture was the technique with greater richness and diversity and less dominance in both strata.Species accumulation curves show a tendency to anticipate stability for the canopy compared to the ground (Figure 2). This stability in the canopy added to the qualitative and quantitative results, as seen in Table 1.The analysis of the pair-to-pair comparison of different methods (stratum–technique–attractant) showed significance values between practically all methods, with the exception of the trap using only the BG-Lure® installed in the canopy (with negative results) and the trap installed on the ground versus the traps with the CO2 + BG-Lure® 1 and 2 in the canopy (Table 3). The highest coefficients occur between the traps using CO2 and CO2 + BG-Lure® at the ground level and between the nets on the ground and in the canopy.The Bray–Curtis similarity analysis for the different methods for collecting mosquito species is used to verify clusters of those methods that are most similar in qualitative and quantitative terms of species. Figure 3a shows the similarity dendrogram based on the analysis of all species found. A cluster that stands out includes all the traps at ground level that used CO2 (exclusively or with BG-Lure®) and the hand-net in the canopy, which suggests that one method could replace the other or serve as a replica. At the other extreme, the ground-trap methods and traps using only BG-Lure® in the canopy have the fewest similarities, being isolated from the others. This, however, can be attributed to different reasons, the first being the method presented superior abundance and richness, and the latter being almost null.The analysis considering each of the four most abundant genera in individualized dendrograms is presented in Figure 3b–e. The genus Psorophora presents a very similar large group comprising several methods; ground hand-net methods and traps with only BG-Lure® in the canopy appear in isolation, following a pattern similar to the dendrogram for all species, which also presents the group of traps with CO2 in the canopy. The genus Aedes also shows some variation of this pattern for the central group, with fewer similarities than Psorophora. In the case of the Haemagogus and Sabethes genera, both presented a group with great similarity composed of nets in the canopy and on the ground, suggesting the human attractant as an aggregator. However, Haemagogus seems to be the genus that expresses more standardized responses to methods, while Sabethes shows more diffuse behavior for technique –attractant–stratum combinations. 4. DiscussionOur study is the most recent survey of diurnal mosquitoes in the savanna biome, where species richness and abundance were studied according to forest stratification. The main objective was to evaluate alternative capture methods to the hand-net, using the attractants BG-Lure® and carbon dioxide in order to increase the effectiveness of automatic traps, especially for species of epidemiological interest that frequent the forest canopy. Thus, we sought to carry out collections in a favorable seasonal period, mainly for the species of the Aedini and Sabethini tribes [38,39,40,41,42,43].Based on the results of the species accumulation curves for both strata, we verified that the sampling in the period was efficient, allowing for a good characterization of the diurnal mosquito fauna in the “Pé-do-Gigante” sector of the park. Thus, although we used a shorter collection period (27 days) than in other studies conducted in savanna areas (40–60 days) [3,44,45], we found similar species richness for the genera Aedes, Haemagogus, Psorophora, and Sabethes, with 14 species found in our study, and 14 to 17 in the others. It is also worth mentioning that these three studies were carried out in central Brazil, between latitudes 15.5 and 17.0°S, while our study, at latitude 21.5°S, represents an important record of this fauna in a region close to the southern limit of this biome in Brazil [46].In addition, due to the methodology of simultaneous collection in the canopy and on the ground, a robust comparison between these two strata was possible. In other systematic studies [39,47] or specific entomological surveys associated with epidemic outbreaks [48,49,50] that also investigate the community of diurnal mosquitoes using the mobile human bait technique with hand-nets, collections were not carried out in the canopy. Thus, all the data on the abundance, richness, and dominance of species in the region were related to collections at the ground level. Our study provides pioneering results on the stratification of mosquito communities in the northwest region of the state of São Paulo.Considering the general results from the two strata, we can observe the dominance of Aedes scapularis and the subdominance of Haemagogus janthinomys/capricornii, Psorophora albigenu, Sabethes albiprivus, and Ae. albopictus. In the comparison between strata, Ae. scapularis was dominant on the ground, while Hg. janthinomys/capricornii dominated in the canopy. Qualitatively, the list of species in the canopy and at ground level was very similar. In quantitative terms, practically all the species found in the canopy had a lower number of specimens compared to collections carried out on the ground, which was also observed in other studies carried out in the savanna [3,44,45].This suggests that microclimatic conditions in the canopy may be less favorable than at ground level. We should emphasize that the collection points were in the “Cerrado ss” physiognomy, where the shading and size of the trees were less than in other physiognomies found in the park, such as the “Cerradão”, gallery forest, and semideciduous seasonal forest [31,51]. In this sense, the canopy was more subject to desiccation, affecting the intensity and maintenance of relative humidity, which is a key environmental factor for the main species found there [39,52,53].The values of the indicators of diversity, uniformity, and dominance were similar for the same method between the canopy and the ground. The same was not observed for the richness and abundance indicators, which were higher in the soil. The Spearman’s correlation results corroborate this, since significant differences were found between the different methods in the same stratum and between strata. Contrarily, these indicators were different between the strata and methods in a similar study we carried out in the Atlantic Forest biome in Cantareira State Park [21].However, in both studies, the hand-net was the technique with greater richness and diversity and lower dominance in both strata, while the indicators of traps using only BG-Lure® demonstrated inefficiency, a result also found for wild diurnal mosquitoes in Brazil by other researchers [19,54]. This shows that the exclusive use of BG-Lure® may be inefficient for wild neotropical species. The use of CO2 as an attractant demonstrated worse performance than its use together with BG-Lure®, a different result from that found in the Atlantic Forest [21] and probably due to the differences in specific environmental conditions in the savanna. It is worth mentioning that there are few attractants available for commercial use, as can be seen in many studies carried out in Brazil to evaluate the performance of mosquito collection with automatic traps [19,21,22,23,24,25,27,28,29,54,55]. This suggests the need to develop in situ studies using olfactometry to discover different substances that present selective attractiveness for each species of interest, according to the biome where it will be used. Among the most abundant species of mosquitoes collected in the “Pé-do-Gigante” sector, all of them showed epidemiological importance, especially for the transmission of yellow fever. Haemagogus janthinomys, Hg. leucocelaenus, and Sa. chloropterus are considered the main vector species in South America [56], the first two with the greatest range in Brazil [57,58]. In our study, we used the nomenclature Haemagogus janthinomys/capricornii because we were unable to determine which of the two species occurred at the site, since they are morphologically differentiated only by the male genitalia, and the investigated site is in the co-occurrence zone [2]. At any rate, whatever the species, it was the most abundant in the canopy compared to the other mosquito species, with a greater number of specimens at ground level than in the canopy. This result differs from several studies in which the species is cited as an acrodendrophile [19,45,52,59,60]. On the other hand, it adds to the results of other studies where variations in this pattern were observed [3,44,54].The other species with epidemiological importance for yellow fever, which presented the greatest number of specimens, were Ae. scapularis, Ae. albopictus, Ps. albigenu, Sa. albiprivus, and Sa. glaucodaemon, and, to a lesser extent, Ae. serratus, Ps. ferox, and Hg. leucocelaenus. The rarity of Hg. leucocelaenus is noteworthy, since it is one of the dominant species in Cantareira State Park [21] and a vector for the transmission of the yellow fever virus in other areas of the Atlantic Forest [58,61,62]. Perhaps this species has no favorable habitat in the savanna, at least in the stricto sensu physiognomy investigated in this study. The assessment of its presence and abundance in enclaves of seasonal semideciduous forests, gallery forests, and in the physiognomy of the “Cerradão” deserves to be studied in order to better understand the details of the ecology of the species in this biome, since previous studies have not exactly focused on this [26,44,45,47,63].An important result of this research is the assessment of the success of different techniques and attractants in the canopy and on the ground for capturing the genera Aedes, Haemagogus, Psorophora, and Sabethes, providing valuable data for arbovirus surveillance services in Brazil. In the similarity cluster analysis, the selectivity of hand-nets for the genera Haemagogus and Sabethes were evident in the dendrograms, while for the genera Aedes and Psorophora, the similarity between methods was more similar to the general dendrogram (Figure 3a all species), where the hand-net on the ground and the trap using only BG-Lure® in the canopy appear distant from the other methods, with the former having the highest relative yield and the latter the lowest relative yield. In investigations of epidemics or epizootic outbreaks, the right place and time are crucial for obtaining good samples [10,64]. Thus, the use of the technique with the highest yield for the target species is very important. However, it is worth considering that the “ideal” collection technique is not always possible for surveillance services. Moreover, the use of automatic traps may be the only way to expand the space–time factor in routine and large-scale surveys [65,66]. Thus, if it is necessary to make use of these alternative techniques, this study, and that of Deus et al. (2022), may represent an important reference for the operational planning of field actions [21].