Fitness characteristics of the malaria vector Anopheles funestus during an attempted laboratory colonization

Background The malaria vector Anopheles funestus is increasingly recognized as a dominant vector of residual transmission in many African settings. Efforts to better understand its biology and control are significantly impeded by the difficulties of colonizing it under laboratory conditions. To identify key bottlenecks in colonization, this study compared the development and fitness characteristics of wild An. funestus from Tanzania (FUTAZ) and their F1 offspring during colonization attempts. The demography and reproductive success of wild FUTAZ offspring were compared to that of individuals from one of the only An. funestus strains that has been successfully colonized (FUMOZ, from Mozambique) under similar laboratory conditions. Methods Wild An. funestus (FUTAZ) were collected from three Tanzanian villages and maintained inside an insectary at 70–85% RH, 25–27 °C and 12 h:12 h photoperiod. Eggs from these females were used to establish three replicate F1 laboratory generations. Larval development, survival, fecundity, mating success, percentage pupation and wing length were measured in the F1 -FUTAZ offspring and compared with wild FUTAZ and FUMOZ mosquitoes. Results Wild FUTAZ laid fewer eggs (64.1; 95% CI [63.2, 65.0]) than FUMOZ females (76.1; 95% CI [73.3, 79.1]). Survival of F1-FUTAZ larvae under laboratory conditions was low, with an egg-to-pupae conversion rate of only 5.9% compared to 27.4% in FUMOZ. The median lifespan of F1-FUTAZ females (32 days) and males (33 days) was lower than FUMOZ (52 and 49 for females and males respectively). The proportion of female F1-FUTAZ inseminated under laboratory conditions (9%) was considerably lower than either FUMOZ (72%) or wild-caught FUTAZ females (92%). This resulted in nearly zero viable F2-FUTAZ eggs produced. Wild FUTAZ wings appear to be larger compared to the lab reared F1-FUTAZ and FUMOZ. Conclusions This study indicates that poor larval survival, mating success, low fecundity and shorter survival under laboratory conditions all contribute to difficulties in colonizing of An. funestus. Future studies should focus on enhancing these aspects of An. funestus fitness in the laboratory, with the biggest barrier likely to be poor mating. Supplementary Information The online version contains supplementary material available at 10.1186/s12936-021-03677-3.

Colonization of An. funestus s.s. has however been problematic, and only two strains have been successfully colonized from wild populations despite several attempts. Both strains were colonized at the Vector Control Reference Laboratory (VCRL) in the National Institute for Communicable Diseases, South Africa, from populations in Angola (FANG) and Mozambique (FUMOZ) [32,33]. The FUMOZ strain also maintained at other laboratories worldwide, including in Cameroon and in the UK [12], as well as in Tanzania. Several attempts have been made to colonize new An. funestus strains [34] from wild populations, but methods used to establish FUMOZ and FANG have not been successful elsewhere [35], including when attempted in the same wild populations where FUMOZ was originally derived (Coetzee, pers. commun.). This inability to repeatedly colonize and establish An. funestus in laboratories is responsible for the more limited understanding of the biology of this species compared to other vector species.
Several factors may account for the difficulty of colonizing An. funestus. Chief amongst these is eurygamy, the inability of an insect to mate in flight [36,37]. Eurygamic species are difficult to colonize because they do not exhibit natural mating behaviours, such as swarming [38], under insectary conditions [36,39,40]. Many Anopheles mosquitoes mate naturally in aerial swarms [41][42][43] or, to a smaller extent, indoors [44]. Whilst An. gambiae will mate readily in the laboratory [4], wild and F 1 progeny of An. funestus rarely swarm inside cages. Although mating is hypothesized to be the main barrier to An. funestus colonization, other factors cannot be ruled out due to incomplete or absence of reporting on other aspects of their life history and fitness during attempted colonization. Therefore, it is crucial to comprehensively evaluate how all aspects of An. funestus life history, development and demography respond to standard methodologies for colonization in order to identify where modifications should be focused.
Malaria eradication in most African settings will require a more detailed understanding of the basic biology and ecology of vectors of residual transmission, and which additional strategies will most effectively target them. Historically, most research on African malaria vectors has also been concentrated on the Anopheles gambiae s.l. species group. Anopheles funestus has also recognized as an important vector in many African settings [26,27,45]; and is now the dominant source of transmission in many areas following the decline of An. gambiae [27,28,46]. However in contrast to An. gambiae s.l., much less is known about the ecology and fundamental biology of An. funestus s.l. This knowledge gap is due to a range of factors including the more cryptic nature of it larval habitats and most notably the difficulties with colonizing it. Given the growing prominence of this species in mediating residual transmission across Africa, there is a clear need to overcome these obstacles as required to guide the development and implementation of more effective control strategies.
To address these knowledge gaps, the fitness and behavior of wild An. funestus from Tanzania and their offspring (defined as "FUTAZ", i.e. An. funestus from Tanzania), were quantified during repeated colonization attempts under standard laboratory conditions. From previous studies, it is known that colonization of this species using standard approaches is difficult. The first step in optimizing this process is to understand which aspects of An. funestus life-history and fitness are most impaired during colonization, and thus target modifications appropriately. To assess this, a detailed measurement of the fitness and life-history of wild and F 1 An. funestus were conducted during repeated laboratory colonization attempts. Fitness measures of individuals in this nascent colony were compared to those of a stable An. funestus colony (FUMOZ) to identify the key barriers that hinder successful colonization of this species. The term "fitness trait" refers to measures of mating success (insemination status), fecundity (number of eggs produced), adult body size and survival (larval and adult). Insights gained will guide future research to overcome barriers to colonizing An. funestus, and also increase knowledge on this important vector and its control. Ngowo et al. Malar J (2021) 20:148

Study area
Wild An. funestus s.l adults were collected from three villages (Tulizamoyo, Ikwambi and Sululu) in Kilombero (8.1539ºS, 36.6870ºE) and Ulanga (8.3124ºS, 36.6879ºE) districts in Tanzania (Fig. 1). The villages were selected because of their high abundance of An. funestus s.l., of which > 93% are known to be An. funestus s.s. [47]. Wildcaught females were transported to the Ifakara Health Institute and used in experiments at the vector biology & control laboratory, the VectorSphere at Ifakara (Fig. 1).

Mosquito sampling
Five houses were selected for mosquito collection in each village. Mosquito collections were conducted for one week in each village in 2019 (Tulizamoyo: 17-23rd June; Ikwambi: 8-16th July and Sululu: 1-10th September). Due to collection logistics and space limitation in the VectorSphere, only one set of experiments (i.e. with mosquitoes from just one village) was done at a time. Trapping was done using CDC light traps [48,49] that were set from 6 pm to 6 am for five consecutive nights per village (yielding ~ 200-300 female An. funestus s.l. per week). Light traps were fitted with larger catch bags to help keep mosquitoes alive without desiccation until morning. Every morning, live female An. funestus s.l. were aspirated from the catch bags into netted cages (30 × 30 cm), provided with 10% glucose solution and brought to the VectorSphere for blood-feeding and further rearing. Inside the VectorSphere mosquitoes were kept under standard conditions of 70-85% RH, 25-27 °C and a 12 h:12 h photoperiod.

Laboratory maintenance and fitness measurements for FUTAZ mosquitoes
Once in the VectorSphere, wild female of An. funestus s.l. were given an initial blood meal from a chicken for a maximum of 30 min (from 6:30 pm) inside cages covered with dark cloth. After this first blood meal, mosquitoes were left in the cage until the next morning when their feeding success was recorded by visual observation. Those with a distended, red abdomen were considered to be fully engorged and transferred into individual oviposition cups for egg laying (Additional file 1: Fig. S1). Cotton pads soaked in 10% glucose solution were placed onto the top netting over the cups for additional nutrition. After three days, a small amount of water (~ 5 ml) was put in each cup to stimulate oviposition. Cups were then inspected daily to record if and when eggs were laid, and dead mosquitoes were removed. Mosquitoes that did not lay eggs after 12 days were killed by freezing for 10 min and later dissected to assess insemination. The terminalia and last abdominal segment (segment IX) were cut-open in distilled water to expose the spermathecae. Slide mounts of spermathecae were inspected using a microscope at 400× magnification for presence of sperm (Fig. 2a&b). F 1 eggs from wild-caught An. funestus s.l. were identified to species-level based on their morphological characteristics [50]. Eggs were observed under a Stereo Microscope, and sub-samples of emergent adults verified by PCR [51]. All eggs morphologically confirmed as belonging to An. funestus s.s., were retained for use in subsequent colonization and life history experiments and defined as F 1 -FUTAZ (Fig. 2c). Here, all F 1 -FUTAZ eggs were pooled and redistributed into a series of 5L round plastic basins (30 cm diameter, filled to 3.3 cm with tap water, replaced every 2 days) at approximate densities of 400-600 eggs per basin and left to hatch. There were a From here onwards, the term An. funestus refers specifically to An. funestus s.s.
Larvae were fed daily with a pinch (approx. 0.36 g) of a mixture of finely crushed dog biscuits and brewer's yeast at a ratio of 3:1 [32]. Basins were checked daily to record egg hatching and larval survival. Pupae that emerged over three consecutive days were recorded and retained for use in F 1 adult mosquito fitness experiments. Female: male ratio was also recorded at this stage by looking at the genital lobe shape (i.e. at the end of the pupae abdominal segments just below the paddles, also males tends to be smaller than females [52,53]. Pupae were placed in a single cage (30 × 30 cm) and monitored until emergence (1-2 days). A total of three cages were set up for each of the three independent colonization experiments. F 1 FUTAZ females were offered their first human blood meal five days post-emergence, on the same day and time as females from the FUMOZ colony their fitness was compared with (see below). Females were provided with additional blood meals every five days. The rationale for blood feeding every 5 days was to estimate the survival under conditions where they had access to blood meals at a frequency similar to that expected in the wild. Studies indicate that the gonotrophic cycle length of An. funestus ranges between 2 and 5 days [54,55]; with 5 being selected here due to practical considerations. From the first blood meal onwards, cages were inspected daily to record and remove all dead mosquitoes. A total of three cages were used as replicates in each of the 3 independent colonization experiments. All the dead females were dissected to assess insemination status. Random subsamples of F 1 females were selected after the second blood meal and moved into individual oviposition cups to measure their egg production.

Laboratory maintenance and fitness measurements for FUMOZ mosquitoes
In July 2018, eggs from the FUMOZ An. funestus colony were obtained from the VCRL laboratory in South Africa and used to establish a colony within the VectorSphere, at Ifakara Health Institute in Tanzania. The founder FUMOZ colony at VCRL has been maintained since 2000 [32]. At IHI, the FUMOZ colony was maintained for four generations before starting these experiments. This colony was kept under the same insectary conditions (70-85% RH, 25-27 °C and 12 h:12 h photoperiod) in the VectorSphere and same feeding regime as described above for F 1 -FUTAZ. In this study, the following fitness variables were measured in FUMOZ for comparison with FUTAZ: number of eggs laid per mosquito (fecundity), proportion of eggs hatched, wing lengths, proportion of adult female inseminated, proportion of larvae survived, larval development period, and proportion of pupae emerged as adult, female: male sex ratio at pupae stage and number of days survived by adult females. The definition of all fitness traits measured, and the colonies in which they were made were given in (Additional file 1: Table S1).

Mosquito wing size measurements
The wing lengths of all female An. funestus (Wild FUTAZ, F 1 -FUTAZ and FUMOZ) were measured and used as a proxy for their body size. One wing was removed from each mosquito and placed onto a drop of water on a microscope slide. The wing lengths were measured using the micrometer ruler under a microscope (50 mm micrometer scale in 0.1 mm divisions, 70 mm × 20 mm × 3 mm) [56]. Measurements were taken from the apical notch to the auxiliary margin, excluding the wing fringe (Fig. 2d).

Statistical analysis
Data analyses were conducted using R statistical software version 3.5.0 [57]. Mean values were estimated for An. funestus fitness traits (Additional file 1: Table S1 in SI), and how these variables differs between FUTAZ and FUMOZ strains. Where possible, fitness traits (i.e. wing length and proportion inseminated) were also compared among the wild FUTAZ, F 1 -FUTAZ and FUMOZ. Additional analysis was conducted to assess the relationship between female body size and fecundity (number of eggs produced) in wild FUTAZ and FUMOZ.
Generalized linear mixed models (GLMM) implemented in lme4 package [58] were used to estimate mean values of fitness traits in wild and F 1 -FUTAZ and FUMOZ strains. Fecundity, the number of eggs laid per mosquito, was modeled as a Poisson variate and wing length was included in the model as fixed effects. For proportion data (here, emergence, insemination, larval survival and sex ratio, (Additional file 1: Table S1) were modeled as a binomial variate with strain used as fixed effects while replicates as random effects. Wing length and strain were also used as fixed effect when assessing insemination. Wing length was modeled as a Gamma variate with an inverse link function, incorporating strain as fixed effect. Tukey's post-hoc tests were used to assess the means differences for different fitness measurements.
Survival analyses were done using a Cox proportional hazard model using the survival package [59] to assess the odds of mortality for males and females for each strain and for females of the two strains of An. funestus (F 1 -FUTAZ vs. FUMOZ). Here, the response variable was the death occurring on each day of observation, while strains and sex were included as explanatory variables. In the analysis of F 1 -FUTAZ, site of collection was included as a random effect by fitting a frailty function [60,61] using a Gamma distribution. Separate analyses were performed for each strain except when the differences between strains were investigated. Log likelihood ratio tests (LRT) were used to test the significance of each variable of interest in all models. All figures were produced using ggplot2 [62] and survminer [63] R packages.

Results
A total of 1,130 adult females of the wild-FUTAZ strain were collected from the three different villages, Tulizamoyo (n = 332); Ikwambi (n = 425); Sululu (n = 373). More than two-thirds (n = 804) of these successfully fed when offered a blood-meal in the insectary, of which 39% (n = 316) laid eggs in the insectary.

Discussion
Limited understanding of mosquito biology and ecology poses a challenge for the development of effective vector control approaches. Laboratory colonization of target species provides an opportunity to address these knowledge gaps by facilitating detailed investigation of vector biology under controlled conditions where experimental manipulation is possible. Here the fitness traits of An. funestus during colonization attempts from a wild population in Tanzania were characterized to identify the bottlenecks that make this species so difficult to colonize. This is the first documentation of fitness constraints during attempted colonization of this species and the first report of attempted colonization of An. funestus from Tanzania. Consistent with most previous attempts, colonization of this wild An. funestus population proved unsuccessful with no offspring being produced from the F 1 generation. Several life history processes and demographic traits were identified as being impaired when FUTAZ were brought into the laboratory. First, there number of eggs laid by wild FUTAZ when brought in the insectary was lower than in the well-establish FUMOZ line, as were hatching rates, larval survival, mating success and adult female/male survival. F 1 FUTAZ body sizes were slightly smaller compared to their maternal generation in the wild, but did not differ with the FUMOZ strain indicating this trait is unlikely to predict colonization success. Of all these fitness traits, the primary hurdles to colonization are likely to be the extremely low mating success and larval survival of F 1 An. funestus in the laboratory. Until these fitness traits can be improved under laboratory conditions, the colonization of An. funestus is unlikely to be successful and repeatable.
Eggs laid by wild-FUTAZ An. funestus had low proportion of hatching compared to those of FUMOZ, though they were both lower than 50%, indicating many were unviable eggs laid by non-inseminated females. Previous studies investigating the impact of different water sources used for larval rearing in an An. funestus colony (FUMOZ) indicated that their egg hatching rate can exceed 70% [64,65]; confirming hatch rates in this study were low. It is known that females of other Anopheles species can produce unviable eggs without successful mating, or after mating with sperm-less males [66].  Therefore, poor hatching observed in the nascent strain (FUTAZ) here is likely due to first, the low mating success of An. funestus in captivity as has been previously documented in other Anopheles species [35], second since the number of eggs laid was very low, the absence of any hatching could be just stochastic effect, lastly, low hatching rate could also be associated with suboptimal temperature in the laboratory relative to the one in the field to which FUMOZ strain might have been well adapted to it but not FUTAZ [67]. The larval development period of F 1 -FUTAZ (11-14 days) was similar to that reported for An. funestus in other laboratory settings [64,65,68], but faster than FUMOZ development period (21-23 days) observed in this study. The duration of larval development in An. funestus (FUTAZ and FUMOZ) observed here were considerably longer than described for An. gambiae complex in the laboratory [69]. For example, life table analyses of An. gambiae indicate larval development period from eclosion to adult emergence of about 11 days at 27 °C [55,[69][70][71][72]. This long larval development period for F 1 -FUTAZ results in a long estimated generation time of 30-33 days from eggs to first oviposition; which is higher than estimated for other African Anopheles species [64,73,74]. Other life table analyses performed on An. funestus colonies estimated a generation time of approximately 33 days in insecticide-resistant (FUMOZ) and susceptible strains (FANG), [65]. As a consequence of this extended period of larval development, the egg to pupa survival was very low; approximately 6% for F 1 -FUTAZ and 27% for FUMOZ. Due to this long larval development and associated high larval mortality, very large numbers of eggs would be required generate modest numbers of adults in the laboratory. Therefore, the fitness and reproductive success of these resulting adults would have to be very high to yield a further generation.
Analysis of wild FUTAZ adults and their F 1 offspring indicate their fitness is reduced compared to that of a stable An. funestus colony (FUMOZ). Wild-FUTAZ An. funestus brought into the laboratory laid 16% fewer eggs than the FUMOZ colony, and the F 1 generation of FUTAZ produced no viable eggs at all. A previous study measuring the fecundity of F 1 An. funestus using Madagascan population reported that this species can lay and average of 56 to 108 eggs per mosquito in captivity [35], which corroborates with 65 and 76 eggs from wild-FUTAZ and FUMOZ, respectively, from the current study. The number of eggs here is consistent to that reported in resistant and susceptible An. funestus strains in the laboratory, [65].
Mosquito body size is often interpreted as a proxy of their fitness [75][76][77]. Here, wild FUTAZ were somewhat larger in body size than F 1 -FUTAZ and FUMOZ.
Consistent with the hypothesis of body size being an indicator of fitness, wing length was positively correlated with fecundity in the wild population of An. funestus (FUTAZ). However, the opposite was seen in the stable FUMOZ strain where wing size was negatively associated with fecundity. Thus, at least in this one stable laboratory colony, large body size in An. funestus was not a good indicator of reproductive success. Caution is required in extrapolating fitness differences based on An. funestus body size, particularly between field and laboratory strains. Although body size fell between wild-FUTAZ and F 1 -FUTAZ, these mosquitoes were still bigger than the FUMOZ which had the highest fecundity.
The mating success of An. funestus from these populations was extremely low in the laboratory, supporting hypothesis that mating is the key bottleneck for the colonization of this species. Compared to wild-FUTAZ, insemination rates in F 1 -FUTAZ were extremely low (9.2% vs. 72%) and insufficient to establish a further generation F 2 -FUTAZ. This poor mating success is likely due eurygamy, the inability of some Anopheles species including An. funestus to initiate natural swarming behaviour in flight [78,79]. These findings match those of other studies documenting mating as the major obstacle for successful colonization of An. funestus [34,35,80]. To overcome this problem, techniques such as forced mating and exposing mosquitoes during sunset to induce swarming have been applied [81,82]. Other studies have experimented with the use of large cages to stimulate natural mating for Anopheles, and simulate sunset which may be crucial cue for mating [83,84]. However so far these methods have had little or no success over multiple attempts [32,34]. In the current study, no F 2 -FUTAZ offspring were generated because none of the F 1 -FUTAZ laid viable eggs. Further research on how to induce mating behavior in An. funestus, particularly using more realistic semi-field systems, would be of great value. Such studies must focus on both females and males, to determine if males are unwilling to initiate swarming behaviour or not fit enough to do so.
Analysis of adult mosquito survival indicated that the nascent Tanzania colony (F 1 -FUTAZ) had a reduced lifespan compared to stable An. funestus colony (FUMOZ). However adult survival in both cases was relatively high (32 median days for FUTAZ and 52 days for FUMOZ); with both strains living well beyond the minimum period required to produce eggs and transmit malaria. Another laboratory study conducted on FUMOZ where adult life span ranged from 39 to 64 days [65]; again much higher than F 1 -FUTAZ here. The shorter life span of FUTAZ relative to FUMOZ may be a result of the stress from the change of environment, or lack of adaption to laboratory conditions. Nevertheless, this F 1 -FUTAZ survived much longer compared to another competent vectors of malaria transmission, Anopheles arabiensis and An. gambiae s.s. in the laboratory conditions [21]. Previous experiments on parity shows that the median survival of An. funestus in the wild is much shorter, ranging from 7 to 10 days in the wild population [85]. Thus, poor adult survival relative to the wild cannot explain the failure of colonization here.
A potential limitation of our study is that the unfed An. funestus females were used to seed laboratory colonies, requiring us to blood feed them artificially (on chicken blood) in the laboratory to acquire eggs for the next generation. Thus, the mating status and age of the wild females used for colonization were uncertain and likely variable. An alternative would have been to collect only visibly blood fed females during field collections (these individuals would likely had fed on humans if caught inside houses), and used their eggs to generate the F 1 generation. This was considered, but given the much lower abundance of blood fed An. funestus inside houses compared to the numbers of unfed females that can be obtained in CDC light traps; the latter approach was chosen to ensure sufficiently large samples were obtained for colonization experiments. These wild mosquitoes could not be provided with a human blood meal given their malaria infection status was unknown and they were not adapted to membrane feeding, thus chicken blood was provided. This variation in host blood source could have generated some differences in fitness between strains. However, this is unlikely given that previous studies indicate that human and chicken blood meals generate similar egg production in other African malaria vector species under laboratory conditions (An. arabiensis and An. gambiae [21,86]). The F 1 population, upon which the main fitness indicators were assessed, was fed on human blood. Further investigation is required to confirm whether An. funestus fitness is impacted by the type of host blood meals provided and whether there is an optimal diet for laboratory maintenance.

Conclusions
Laboratory colonies remain fundamental for research on the biology and control of mosquito vectors, by providing a stable and standardized source of mosquitoes for experimental studies. This study provides additional evidence of the intractability of An. funestus to colonization. By quantifying a comprehensive range of fitness traits during unsuccessful attempts, this study generates insights into the most important barriers to colonization. Of the range of traits investigated, the primary barrier to colonization was identified as low mating success, compounded further by the slow development and poor survival of the small numbers of larvae produced. Additionally, both the fecundity and adult survival of An. funestus offspring from wild parents were reduced under laboratory conditions, but these impacts may have been relatively minor compared to the consequences of poor mating success and poor larval survival. This combination of fitness deficits presents a major challenge for successful colonization and mass rearing of An. funestus. To overcome this, future research should focus on enhancing the efficiency these life-cycle processes under insectary conditions. Additionally, the demographic rates estimated from wild and F 1 -FUTAZ will provide useful baseline information for understanding and modeling An. funestus population dynamics in general, and guiding further attempts to its colonization.
Additional file 1: Table S1. Descriptions of terms used and strains compared as used in this study. Table S2. Total number of mosquitoes for each replicates and sex as used during the survival analysis between two strains (of FUTAZ and FUMOZ). Figure S1. Images of: a) Series of wooden cages with oviposition cups, each containing a single fully-engorged Anopheles funestus female. Individual cups were used to measure the number of eggs laid by a single mosquito after full blood meal, b) Technician counting the number of eggs and measuring the wing sizes of individual mosquitoes which have laid eggs, c) Technician aspirating mosquitoes from the rearing cage using mouth aspirator and d) bowl contains eggs of Anopheles funestus. All experiment was done within the VectorSphere at Ifakara Health Institute.