Databases and search strategy

Two investigators (M.H.Z. and H.V.W.) searched ‘Pubmed’ and ‘Proquest’ for the following search terms: Aedes aegypti, Ae. aegypti, A. aegypti, biting rate, biting rates, biting frequency, biting habits, daily bites, blood feeding frequency, blood meals, frequency of feeding, feeding pattern, landing patterns, human bait, and diel biting (see S1 Text for all search terms with Boolean operators).

Study selection

Studies included in this systematic review met the following criteria: (1) studied the biting rate or biting frequency of Ae. aegypti, (2) published in peer-reviewed journals, (3) published as a research article, not as a review article, (4) study conducted near or within a household or in a laboratory, (5) written in English or Spanish, and (6) published between January 1, 1970, and December 31, 2022. Initially, investigators reviewed the title and abstract only to assess whether articles met the inclusion criteria. Subsequently, investigators read the whole text of the manuscripts meeting our inclusion criteria.

A total of 21 studies were selected for detailed analysis (Fig 1). These 21 studies can be broken down into human landing catch studies (16) and multiple feeding studies (5) (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flow chart depicting the process for exclusion and inclusion of articles.

https://doi.org/10.1371/journal.pntd.0010831.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Categorization of studies included in this review.

https://doi.org/10.1371/journal.pntd.0010831.t001

Data extraction

For our analysis, we extracted biting data from five studies [21,23,29–31] that only provided data graphically using WebPlotDigitizer (https://apps.automeris.io/wpd/), a tool designed to extract data points from figures. One of the goals of this study is to investigate how climatic factors, such as rainfall and temperature impact biting of Ae. aegypti. However, only one [31] of the six field studies [21–23,29–31] that reported biting data also provided rainfall (graphically) and temperature data. For that specific study, we extracted the rainfall data using WebPlotDigitizer. For four [21,23,29,30] of the remaining five studies, we extracted temperature and rainfall data from the NASA Power Data Access Viewer (https://power.larc.nasa.gov/data-access-viewer/) over the timeframe of the study period. For temperature, we used the temperature at two meters (T2M) variable, the standard measurement for air temperature [32]. To extract these data, we obtained GPS coordinates for each study location from https://www.latlong.net/, provided in Table 2 for reproducibility. We could not extract temperature and rainfall data for the other study [23] because we did not find the study location in our GPS coordinate source (https://www.latlong.net/).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. GPS coordinates used to extract temperature and rainfall data.

https://doi.org/10.1371/journal.pntd.0010831.t002

Data analysis

One [21] of our reviewed studies reported hourly biting data averaged over three nearby locations. So, for [21], we estimate the mean temperature and rainfall values across these three locations (S1 Table). We also estimate seasonal biting for this study [21] averaging reported location-specific seasonal biting values (S2 Table). We use reported seasonal information and biting values to estimate the seasonal biting values for studies [30,31] (S4 Table). To maintain consistency in the mean biting values that we report, for one study [23], we converted their reported biting data from per night to per person-hour by dividing the per night biting rate by the number of hours and number of collectors (S3 Table). For another study [31], we converted the reported rainfall values from monthly to daily. S4 Table contains all biting data used in this manuscript.

Hourly biting

We found two human landing catch (HLC) studies that reported Aedes aegypti indoor hourly biting (mosquitoes per human per unit time). Chompoosri et al. in Central Thailand [30] observed two biting activity peaks–one in the morning and the other in the afternoon during the rainy season and summer (Fig 2A). During the winter months, they observed a delayed morning peak with no clear afternoon peak. In Northern Ghana, Captain-Esoah et al. [21] also observed two peak biting periods during the day, both in the rainy and dry seasons. However, the mean hourly biting values reported in their manuscript were averaged across both rainy and dry seasons (Fig 2B). Two other publications monitored the hourly biting activity–one [20] reported a bimodal pattern estimating combined hourly biting activities of Ae. aegypti and Ae. albopictus (80.2% and 19.8%, respectively) on Samui Island, Thailand, while the other [22] did not observe any significant difference across the hours of the day while observing biting in rooms and yards in a city in Southern Mexico. Our review found four additional human landing catch (HLC) studies [33–36] that reported bimodal biting patterns without providing hourly data.

In 1999, Chadee and Martinez [24] observed a trimodal pattern comprising peaks one hour after sunrise, one hour before noon, and one hour before sunset. In this study, 10% of bites took place at night within urban areas, whereas no nighttime biting activity was reported within rural areas. On the other hand, an earlier study by Chadee (1988) [33] found that in a rural setting 5% of bites occurred at night.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2.

(a) Hourly biting activities across seasons reported by Chompoosri et al. [30], (b) Mean hourly biting reported by Captain-Esoah et al. [21] for 2015 and 2016.

https://doi.org/10.1371/journal.pntd.0010831.g002

Seasonal biting

Based on available data, here, we reviewed three studies [21,30,31] to understand the biting activity and its variations across seasons. Some of these studies described three seasons (rainy, summer, and winter), while others reported two (rainy and dry or wet and dry). Two of these studies [21,31] provided monthly biting data along with seasonal information, while the other [30] provided hourly biting data across seasons. We used these data to estimate seasonal biting rates for these three studies (S2 Table) and identified that seasonal variations differed by study. One of these studies [21] showed that the mean biting values for the rainy season were 3.5 to 5-fold higher than during the dry season (S5 Table). Another study [31] found that biting rates increased during the wet season by 30% to 100%, depending on the geographic location along the coasts of Moorea Island, French Polynesia (S5 Table). The third study [30] observed that the biting activities during the summer were slightly higher than in the rainy season and the lowest during the winter. All estimated seasonal biting rates are tabulated in the S5 Table.

Indoor–outdoor biting

Our review identified six studies that reported indoor and outdoor Ae. aegypti biting activity. Four of them [21,23,24,33] reported higher biting activity outdoors compared to indoors (Fig 3 and S5 Table). This result was consistent for two [21,23] of these studies, regardless of seasonality. The other two studies [24,33] did not report monthly biting activities, so we could not look at these results stratified by season. This variation may be explained by higher mosquito densities in outdoor settings. In contrast to these four studies, Casas-Martínez et al. [22] observed higher indoor biting activities compared to outdoors, 4.10 female mosquitoes/human-day indoors compared to 1.88 female mosquitoes/human-day outdoors in Tapachula, Mexico. Another study [20] observed indoor-outdoor biting activities for Ae. aegypti and Ae. albopictus together and reported Ae. aegypti as primarily endophagic due to their rare presence outdoors.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Indoor and outdoor biting activities between 1700–2100 hours in Northern Ghana (Karch et al. [23]). Here, the unit is converted from nightly to an hourly rate.

https://doi.org/10.1371/journal.pntd.0010831.g003

Impact of rainfall, temperature, and humidity

The rainfall and temperature data we used for this analysis was either reported in the reviewed manuscript or we collected from the NASA website (see data extraction subsection). Of the six studies [21–23,29–31] in this section, two [22,30] provided too few time points to analyze the relationship between weather patterns and biting activity. Our analysis of two of the remaining four studies provide mixed results. For example, we found using biting data from one study [23] that higher rainfall decreases indoor and outdoor biting activities (S1 Fig), however, using reported aggregated biting data across indoors and outdoors from another study [21], we found that higher rainfall increases biting activities (S2 Fig). With regards to temperature, we found in one study [23] that higher temperature decreases outdoor biting activities (S1B Fig), while in another study [21] we found that temperature had a non-linear relationship with biting activity (S2 Fig). Interestingly, the third study demonstrated a joint effect of temperature and rainfall. Salas-Luévano and Reyes-Villanueva [29] showed that biting frequency is higher either when the temperature is low (18° – 21°C) and the rainfall is moderate or when the temperature is high (25° – 28°C) and the rainfall is low (S3 Fig). Finally, Moorea Island in French Polynesia [31] observed the lowest biting activities on the west coast where rainfall was higher compared to the east coast. The temperature was similar on both coasts. In general, these studies suggest that there is no clear pattern between rainfall and temperature, and biting of Ae. aegypti.

A single study by Canyon et al. [37] investigated the impact of humidity in a lab setting and reported a moderate difference in biting frequencies in low (34% RH) versus high (84% RH) humidity conditions (0.077 and 0.083 per mosquito per hour at low and high humidity respectively; note the authors reported in units of per 6 hours resulting in the values 0.46 and 0.5). In their study, the temperature was constant and human blood (not human baits) was the only available diet. However, an earlier study [38] found a greater impact of humidity on biting activities due to the reduction in host-seeking activity at low humidity. Canyon et al. [37] suggested that this discrepancy was due to the difference in host availability, i.e., humidity may have a greater impact when hosts are less available. All biting activities discussed in this section were measured as female mosquitoes per person-hour.

Impact of host availability

One study [39], conducted in a lab setting, examined how human host availability impacts the biting frequencies of Ae. aegypti. In their study, they made humans available to mosquitoes every 6, 12, and 24 hours and estimated the per female mosquito biting frequencies as 0.47 per 6 hours, 0.54 per 12 hours, and 0.7 per 24 hours, respectively (or equivalently, 0.08, 0.045, and 0.03 per hour). Based on these results, the authors concluded that Ae. aegypti are opportunistic feeders, i.e., increasing host availability results in increased biting.

Impact of diet availability

Another lab study [40] investigated the impact of diet availability for four diets (blood + 10% sugar, blood + 3% sugar, blood + water, and blood only) on human-biting. The study observed that the availability of sucrose reduced host biting frequencies by up to 50%, based on available sucrose (median probes/bites per mosquito per hour: 0.037 for the ‘blood + 10% sugar’ diet, 0.047 for the ‘blood + 3% sugar’ diet, and 0.083 for the blood-only diet which were originally reported as median probes/bites per mosquito per 6 hours: 0.22 for the ‘blood + 10% sugar’ diet, 0.28 for the ‘blood + 3% sugar’ diet, and 0.50 for the blood-only diet). To a lesser extent, the availability of water also decreased the biting frequencies (median probes per mosquito per hour: 0.075 for the ‘blood + water’ diet compared to 0.083 for the blood-only diet which were originally reported as per mosquito per 6 hours: 0.45 for the ‘blood + water’ diet compared to 0.50 for the blood-only diet). These investigations were carried out at high relative humidity conditions (84% RH).

Impact of mosquito body size

Two of our reviewed studies examined the relationship between mosquitoes’ body size and their biting frequency on humans. Scott et al. [12] found an inverse correlation in Thailand, where temperatures were higher, whereas almost no relation in Puerto Rico where temperatures were lower overall. The other study by Farjana and Tuno [41] found no correlation in their lab study.

Multiple feeding

We identified five studies that reported multiple feeding of Ae. aegypti in a single gonotrophic cycle. Two of them are mark-release-recapture (MRR) studies. McClelland and Conway [9] released 418 singly marked female mosquitoes, of which 48 were recaptured once and 5 were twice, among which 7 double-fed within 24 hours. Trpis and Hausermann [13] released 563 uniquely marked female Ae. aegypti of which 272 (48.3%) were recaptured. Of these 272 recaptured, 47.8% were recaptured once, one was recaptured 10 times, and the remaining were recaptured anywhere between two and seven times. Based on these values and the measured developmental stages of their ovaries, existing blood in their stomachs, and estimated survival rate, they concluded that most Ae. aegypti bite twice, and some three times within a gonotrophic cycle.

Histological techniques and DNA profiling of human blood have also been used to study the feeding frequency of Ae. aegypti within a gonotrophic cycle. Two histological studies by Scott and his colleagues [11,12] reported multiple feeding of Ae. aegypti. They diagnosed field-collected Ae. aegypti mosquitoes and reported that many mosquitoes fed twice on the same day or by the following day. Harrington et al. [8], using DNA profiling of human blood, found that 10–13% of engorged mosquitoes fed on more than one person in a 24-hour period.

Importance of accounting for multiple feeding

Multiple feeding studies are crucial to estimate per mosquito biting rate because they challenge the classic approach of estimating biting rate as the inverse of the gonotrophic cycle duration (GCD). One multiple feeding study [13] identified that the majority of Ae. aegypti fed more than once within a gonotrophic cycle. Thus, an assumption that at least 50% feed at least twice results in a mean of 1.5 bites per GCD. Several recent studies [42,43] studying Ae. aegypti across the globe reported the duration of the gonotrophic cycle as between 3 to 4 days. Based on these values, the traditional classic approach that ignores multiple feedings estimates the per mosquito biting rate to be between 0.25 and 0.33 (¼ and 1/3) bites/day while accounting for multiple feedings estimates of 0.38 to 0.5 (1.5/4 to 1.5/3) bites/day, which is 50% higher than the traditional estimate.

The mark-release-recapture (MRR) and histological studies are commonly used to track and estimate multiple biting of Ae. aegypti; however, only MRR studies are able to track multiple biting within any given period, whereas the histological studies can only track 48 hours prior to capture. With estimates of the gonotrophic cycle duration and multiple feeding per gonotrophic cycle, we can estimate the mean per mosquito biting rate (MBR) for any mosquito population using the following equation:

Research shows that lower temperatures slow blood-meal digestion and lead to a longer gonotrophic cycle [25–27]. This variability in the GCD and the influence of multiple feeding on the biting rate estimates illustrate the importance of using regionally specific estimates.

Biting rate estimation

Of the 21 manuscripts that we reviewed, only Scott et al. [12] provides the necessary data to estimate per mosquito biting rates. They provide both the number of feedings within the past 2 days or more, and the interval between their feeding time and the day of capture for each captured mosquito. Specifically, Scott et al. [12] collected mosquitoes once a week over a 2-year period in both Thailand and Puerto Rico. Tables #4 and #5 in their manuscript [12] describe the data they collected relevant to estimating the per mosquito biting rate. In Table #4 found in [12], we observe that in Thailand, 595 out of 1,300 mosquitoes fed on the day of capture of which 255, 310, and 30 fed one, two, or three times respectively, whereas in Puerto Rico, 548 out of 1,156 mosquitoes fed on the day of capture of which 293, 244, and 11 fed one, two, or three times respectively. The table #4 also provides data on the number of mosquitoes that had their last blood meal on days other than the day of capture. In Table #5 found in [12], we observe that in Thailand, 65% of mosquitoes that fed twice took both blood meals on the same day, and 78% of mosquitoes that fed three times took their last two blood meals on the same day. In Puerto Rico, 43% of mosquitoes that fed twice took both blood meals on the same day, and 82% of mosquitoes that fed three times took their last two blood meals on the same day. Using these data, they estimated the average number of feeding events at 0.76 and 0.6 per mosquito per day in Thailand and Puerto Rico, respectively.

We use the same dataset to re-estimate the average per mosquito biting rate based on a different interpretation of the data. For our calculations, we assume that biting and feeding rates are equivalent. To estimate a per mosquito biting rate, we relied on the number of bites the mosquitoes took on the day of capture (Tables #4 and #5 in [12]). In the Thailand study, 255 mosquitoes bit only once, and that bite occurred on the day of capture. Also, 310 mosquitoes bit twice where the last bite was on the day of capture (Table #4 in [12]), of which 65% had both their meals on the day of capture (Table #5 in [12]) and the remaining 35% had only the last bite on the day of capture. Finally, they reported 30 mosquitoes that bit three times where the last bite was on the day of capture. Since Scott et al. [12] did not report any mosquitoes having all three meals on the same day, 78% of mosquitoes that had three bites (30 mosquitoes) had their 2nd and 3rd bites on the day of capture and the remaining 22% had only their 3rd bite on the day of capture. In Table #5 in [12], 61% of mosquitoes that took three bites in total are listed as having taken their first two bites on the same day. We assumed that those mosquitoes could not have taken those bites on the day of capture since no mosquitoes were identified as having taken all three bites on the day of capture.

Based on these data, we first estimated the total number of bites taken on the day of capture (Table 4 in this manuscript). We then multiplied this number by 0.95 and 0.97 for Thailand and Puerto Rico, respectively, because 95% of engorged Ae. aegypti in Thailand fed on humans (88% fed on human blood only and 7% fed on mixed blood) and 97% of engorged Ae. aegypti in Puerto Rico fed on humans (95% fed human blood only and 2% fed on mixed blood) [26]. Next, we divided by the total number of engorged mosquitoes to calculate the average per engorged mosquito daily biting rate (Table 4 in this manuscript). Finally, we adjusted our estimates because only engorged mosquitoes were captured and analyzed. Scott et al. [26] estimated that 59 and 58% of captured female mosquitoes were engorged in Thailand and Puerto Rico, respectively. The same study also reported data on gravid mosquitoes and gut contents. In this estimation, we assumed that a portion of the non-engorged mosquitoes, those that were non-gravid and had non-empty gut contents, also bit at the same rate as the engorged mosquitoes. We also assumed that the remaining (gravid or empty) did not bite during the observation period.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Per mosquito biting rate estimation for Ae. aegypti using data from Scott et al [12].

The data represent engorged mosquitoes captured on a single day (595 of 1,300 from Thailand and 548 of 1,156 from Puerto Rico). These mosquitoes were grouped into those in which the histology analysis indicated the mosquito bit one, two or three times. The total number of bites that occurred on the day of capture were tallied (Column 2).

https://doi.org/10.1371/journal.pntd.0010831.t004

Based on data presented in [26], we estimated that among the non-engorged mosquitoes, 41% and 42% (100–59% and 100–58%) in Thailand and Puerto Rico, respectively, 44% and 31% (100–56% and 100–69%) respectively were reported as non-gravid. Thus, the percentages of non-engorged mosquitoes that were both not gravid and not empty in Thailand and Puerto Rico, respectively were 10% (0.41*0.44*0.56) and 5% (0.42*0.31*0.39). The biting mosquitoes are consist of engorged mosquitoes and the non-engorged mosquitoes that are neither gravid nor empty. Thus, the overall percentages of mosquitoes that were biting were therefore 69% (59% + 10%) and 63% (58% + 5%) respectively. We used these numbers to adjust the biting rate estimates, originally measured for engorged mosquitoes, to the total mosquito population (Table 4 in this manuscript).

Our approach estimates biting rates of 0.41 and 0.35 bites per mosquito per day in Thailand and Puerto Rico, respectively. In comparison, Scott et al. [12] estimated the biting rates as 0.76 and 0.63 respectively. The difference comes about because for mosquitoes that bite twice, they counted all mosquitoes that had both bites on the same day instead of only those mosquitoes that had both bites on the same day and the last bite on the day of capture i.e., for Thailand, 65% of 42% (541 of all mosquitoes) instead of 65% of 24% (310 of all mosquitoes) while for triple-biters, they counted mosquitoes that bite three times and imbibed any two consecutive meals (1st and 2nd or 2nd and 3rd) on the same day instead of counting only those mosquitoes that had the last bite on the day of capture and had last two meals on the same day i.e., for Thailand, 61% and 78% of 5% (60 of 1,300) mosquitoes instead of only 78% of 2.5% (30 of 1,300) mosquitoes. Moreover, they did not consider those non-engorged mosquitoes that were neither gravid nor empty. To illustrate the difference between these two approaches, we represent our estimation in a similar fashion as of Scott et al. [12]. For Thailand, our estimate is:

For Puerto Rico, our estimate is: