By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Please be advised that we experienced an unexpected issue that occurred on Saturday and Sunday January 20th and 21st that caused the site to be down for an extended period of time and affected the ability of users to access content on Wiley Online Library. This issue has now been fully resolved. We apologize for any inconvenience this may have caused and are working to ensure that we can alert you immediately of any unplanned periods of downtime or disruption in the future.
Issa N. Lyimo, Environmental and Biomedical Thematic Group, Ifakara Health Institute, PO Box 53, Off Mlabani, Ifakara, Tanzania. Tel.: +255 686 997 013; fax: +255 232 625 312; e-mail: email@example.com
We investigated the fitness consequences of specialization in an organism whose host choice has an immense impact on human health: the African malaria vector Anopheles gambiae s.s. We tested whether this mosquito’s specialism on humans can be attributed to the relative fitness benefits of specialist vs. generalist feeding strategies by contrasting their fecundity and survival on human-only and mixed host diets consisting of blood meals from humans and animals. When given only one blood meal, An. gambiae s.s. survived significantly longer on human and bovine blood, than on canine or avian blood. However, when blood fed repeatedly, there was no evidence that the fitness of An. gambiae s.s. fed a human-only diet was greater than those fed generalist diets. This suggests that the adoption of generalist host feeding strategies in An. gambiae s.s. is not constrained by intraspecific variation in the resource quality of blood from other available host species.
Evolutionary theory predicts that resource specialization evolves only when there is a greater fitness advantage from concentrating feeding on one dietary resource rather than a mixture (Levins, 1962; MacArthur & Pianka, 1966; Pyke et al., 1977). In contrast, generalism is predicted to evolve when there are only moderate fitness differences resulting from feeding on different resources such that there is no net advantage to being selective (Egas et al., 2004; Abrams, 2006a,b). Numerous studies have attempted to test for the existence of dietary fitness trade-offs associated with specialist and generalist feeding, with the majority focusing on the experimental study of phytophagous insects (Price et al., 1980; Futuyma & Moreno, 1988; Jaenike, 1990; Via, 1990). These studies provide some support for the prediction that generalists maximize their fitness by feeding on a variety of resources rather than selecting only one (Bernays et al., 1994; Allard & Yeargan, 2005; Michaud & Jyoti, 2008), and conversely that specialists experience a reduction in fitness when they switch to an atypical host and/or mix their diet between host species (Thomas et al., 2010).
Although phytophagous insects serve as excellent models for the study of host specialization, their broader applicability to other organisms that rely on living hosts such as ectoparasites and insect disease vectors of humans and animals is largely untested. Failure to test theoretical predictions more widely on these and other parasitic organisms limits their application to identify and diminish the factors that drive selection for insect vector specialization on disease-susceptible hosts. Like phytophagous insects, the host range of insect disease vectors varies extensively between extreme specialism and generalism (Lyimo & Ferguson, 2009). Amongst many of the most important insect vectors of human disease, specialization on humans (anthrophily) is common (Lyimo & Ferguson, 2009). Furthermore, variation in the degree of anthrophily within and between vector populations is a strong predictor of spatial variation in disease transmission intensity (Kiszewski et al., 2004; Kilpatrick et al., 2006, 2007). Consequently, identifying the factors that generate selection for extreme anthrophily is a vital first step for investigation of if and how this behaviour could be manipulated to reduce disease transmission.
Mosquitoes in the genus Anopheles are responsible for malaria transmission to humans (Kelly-Hope et al., 2009). There is substantial variation in the host range of Anopheline species ranging from those that feed on a wide range of mammals and birds, to those that feed on just one species (Lyimo & Ferguson, 2009). The most extreme specialist within this genera is probably the African malaria vector An. gambiae s.s, which feeds almost exclusively on humans throughout its range (Kiszewski et al., 2004). According to evolutionary theory, such specialism should only arise if the fitness that these mosquitoes derive from exclusive anthrophagy is substantially higher than from mixing between other available host species. Similar to most haematophagous insects, the key resource that An. gambiae s.s. require from their hosts is protein (required for reproduction), which they acquire from haemoglobin in vertebrate red blood cells (Hurd et al., 1995). Specialization could arise if there is underlying variation in the protein content or nutritional value of host blood, and/or associated behavioural, physiological or ecological differences between hosts that influence the energetic value that mosquitoes derive from them (Lyimo & Ferguson, 2009). Here we explicitly test the impact of the only potential determinant of selection on host species choice, the resource quality of vertebrate blood, and investigate whether its influence on An. gambiae s.s. fitness is sufficient to explain its extreme specialism on humans.
Several haematological properties including biochemical composition and red cell density vary between vertebrate species (Wintrobe, 1933; Harrington et al., 2001) and could influence the nutritive value and subsequent fitness of mosquitoes that imbibe it (Harrington et al., 2001; Lyimo & Ferguson, 2009). Here we investigated the fitness consequences of consuming a specialist or generalist host diet in An. gambiae s.s. with the aim of testing the prediction that specialists pay a fitness cost for diversifying their host species diet. The first hypothesis tested was whether the fitness of An. gambiae s.s. after feeding on the blood of their preferred human hosts is greater than after consuming blood from other host types commonly available in the same environment (avian, bovine and canine). Secondly, we tested whether over the course of the average reproductive lifespan of An. gambiae s.s. (in which a median of two blood meals are taken, Gillies & Wilkes, 1965) their fitness is higher after feeding exclusively on human blood than after consuming a generalist diet consisting of a mixture of blood meals from humans and common animal alternatives. Lastly, we tested whether the relative fitness of mosquitoes was similar across a range of generalist host diets (human plus one other host species) or depends primarily on the specific composition of host types consumed. Our aim was to evaluate whether variation in the fitness value that An. gambiae s.s. obtains from the blood of vertebrate species commonly available to it is sufficient to account for its extreme host specialization on humans. The experimental system used allowed us to control for all sources of variation in mosquito–host interactions other than vertebrate blood resource quality, thus enabling the first comprehensive test of its role in generating selection on the host species range of this important vector.
Material and methods
The An. gambiae s.s. Keele strain (Hurd et al., 2005) was used in this study. This is an outbred line created from the balanced interbreeding of individuals from four laboratory colonies derived different African populations (Kenya, north Tanzania, south Tanzania and the Gambia, Hurd et al., 2005). This colony was established at the University of Glasgow 4 years prior to these experiments, where it is maintained on human blood that is provided to mosquitoes by membrane feeding thrice weekly. Mosquitoes are maintained under standard insectary conditions of 26 ± 1 °C, 80% relative humidity and a 12:12 hour light/dark cycle. Adult mosquitoes are provided with a solution of 5% glucose/0.05% para-amino-benzoic acid (PABA).
Mosquito blood feeding
Two days before experiments, previously unfed adult females of a similar age (3–5 days post emergence) were randomly selected and transferred into cardboard holding cartons (9.5 cm diameter × 9 cm depth) sealed with mesh at the top. Approximately 100–200 females were placed in each carton and held for 2 days under insectary conditions before being offered their first blood meal. On the day of the feed, mosquitoes were offered a blood meal from a membrane feeder as described previously (Carter et al., 1993). All the vertebrate bloods (human and animal) used in these experiments were collected, stored and processed in a similar manner prior to experimentation. Whole human and animal blood was collected aseptically from healthy adult individuals. All blood was stored in sterile vacutainers containing sodium citrate preservative and held in a fridge at 4 °C prior to use. Blood was stored for no more than 7 days after being withdrawn from a host before being fed to mosquitoes.
Host species impacts from one blood meal
In nature, An. gambiae s.s. will blood feed and produce an egg batch once every 2–4 days, with the median number produced before death being two (Gillies & Wilkes, 1965). However, an initial set of experiments was performed in which mosquitoes were provided with only one blood meal to test for intrinsic variation in the resource quality of blood from different host species. Mosquitoes were randomly allocated to one of four host treatment groups: chicken, cow, dog or human blood (Lorne Laboratories, Reading, UK). Within each treatment group, mosquitoes were further randomly subdivided into four feeding cartons (holding 40–50 mosquitoes each), which served as independent replicates of each diet treatment (4 host treatments × 4 replicates/treatment = 16 cartons). Each of these 16 replicates was fed on the blood of a different host individual (four per species) provided from a different membrane feeder. Two hours after the feed, cartons were inspected and all unfed mosquitoes were removed and killed. Blood fed mosquitoes were kept in these original holding cartons for 3 days and provided access to a solution of 5% glucose/0.05% para-amino-benzoic acid. On day 3, survivors were transferred into individual 30-mL universal tubes filled with approximately 1 cm of distilled water to allow them to oviposit. Mosquito oviposition rate (proportion laying eggs) and fecundity (the number of eggs laid) were evaluated the following day by examining each tube and counting any eggs laid with the aid of a dissecting microscope. The survival of mosquitoes was checked on a daily basis from the first day after the blood feed. Mosquitoes were provided with the standard glucose solution on cotton wool pads during this period.
Uniform and mixed host species diets
For their first blood meal, all An. gambiae s.s. were provided with human blood (Patricell UK, Nottingham, UK). Mosquitoes were split into three groups, with each group being fed from the blood of a different human donor. Three days after the feed, all survivors were transferred into individual 30-mL universal tubes filled with water for oviposition as described above. After tubes had been inspected for oviposition the next day, all surviving mosquitoes were pooled in a holding cage and then randomly allocated into one of four different treatments for their second blood meal: human (Patricell UK), cow, dog or chicken blood (Harlan Laboratories Ltd, Belton, UK). With each of these four treatments, mosquitoes were further randomly subdivided into three feeding cartons (holding 10–25 mosquitoes each), which served as independent replicates of each diet treatment (4 host treatments × 3 replicates/treatment = 12 replicates overall). Each of these twelve replicate groups was fed on the blood of a different host individual, provided from a different membrane feeder. The 4-day gap between the first and second blood meal was selected to mimic the blood feeding cycle of An. gambiae s.s. in nature (e.g. 2–4 days between blood feeds, Gillies, 1953). After the second blood feed, all mosquitoes that were observed to have fed (by visual inspection 2 h after the feed) were held in cartons for 3 days and then transferred into individual universal tubes containing 1 cm of distilled water for oviposition. The following day, tubes were inspected to measure oviposition and fecundity as described earlier. All mosquitoes were then transferred into dry universal tubes and maintained there until the end of the experiment (provided with a solution of 5% glucose/0.5% PABA as described earlier). All mosquitoes were checked on a daily basis for further 18 days after their second blood feed (23–25 days from emergence) to monitor survival. This entire experimental procedure was repeated twice to make two blocks.
Statistical analysis was conducted to assess the impact of host diet diversity on three key measures of mosquito fitness: oviposition, fecundity and survival.
First, we investigated whether the proportion of mosquitoes that laid eggs after feeding was influenced by host treatment using generalized linear mixed effect models with binomial errors (glmer) in the R statistical software package (Crawley, 2007). Here ‘host treatment’ was taken as the main effect and the different host individuals within a host species were fit as a random effect. A base statistical model including only the random effect of host individual was constructed, to which the main effect of ‘host treatment’ was added to form the full model. The significance of host treatment was evaluated using likelihood ratio tests (Burnham & Anderson, 2002). Variation in mosquito fecundity (the number of eggs laid) was similarly analysed using generalized linear mixed effect models (lmer) in the R statistical software package (Crawley, 2007).
As the median number of egg batches produced by An. gambiae s.s. before death is two (Gillies & Wilkes, 1965), analysis of mosquito cumulative egg production over their first two blood meals can provide an approximation of the lifetime reproductive success of an average individual. Here we estimated the cumulative number of eggs produced over two feeding cycles to approximate the impact of specialist and generalist blood diets on lifetime reproductive success. In these experiments it was not possible to measure the lifetime egg production of individual mosquitoes, because individuals had to be pooled into groups for blood feeding within which they could not be individually tracked. However, cumulative egg production over two blood meals was estimated by summing the distribution of egg numbers produced after the first and second blood meals on different host individuals. As all mosquitoes initially fed on humans, it was assumed that their fecundity after the first blood meal was similarly distributed. The observed range of eggs produced from the first blood meal (Block 1: 0–103 eggs; Block 2: 0–159 eggs) was split into 11 intervals to obtain a distribution describing fecundity in units of 15 eggs. Distributions of eggs laid after the second blood meal were computed on a similar basis for each of the four host treatments, and an estimate of cumulative egg production obtained by summing distributions from the first and second blood meals. A Kruskal–Wallis test was used to evaluate whether there were statistically significant differences in cumulative egg production between host blood treatments.
The impact of host treatment on mosquito survival was analysed using the Cox proportional hazard model (COXPH) in the R statistical software package (Crawley, 2007). Differences in survival between treatment groups were assessed from the day after the first blood meal in single host experiments, and from the second blood meal onwards in mixed host diet experiments. In these analyses, host treatment was considered as a main effect, and a frailty function was used to incorporate the random effect of host individual (within each host species treatment) into the Cox model.
Host species impacts from one blood meal
The oviposition rate of mosquitoes after one blood meal varied between host species (, P < 0.001), with An. gambiae s.s. having a significantly higher probability of laying eggs after feeding on human and cow blood, than chicken or dog (Fig. 1a). However, the number of eggs laid by ovipositing mosquitoes was similar across host species (, P = 0.66, Fig. 1b). Mosquito survival after taking one blood meal varied between host species (, P < 0.001, Fig. 2). The longevity of An. gambiae s.s. was similarly high after taking a human and cow blood meal, but was significantly lower on chicken and dog blood than on humans (Table 1).
Table 1. Survival of Anopheles gambiae s.s. after one blood meal. The estimated odds of mortality of An. gambiae s.s. after taking one blood meal from different host species relative to human blood. Numbers in brackets are the 95% confidence intervals of the odds ratio.
Uniform and mixed host species diets
Over the two experimental blocks, the reproduction of a total of 140 and 200 An. gambiae s.s. were assayed respectively. The oviposition rate of An. gambiae s.s. after feeding on two human blood meals did not differ from those fed on a meal of human followed by cow, chicken or dog blood (Block 1: , P = 0.15, Fig. 3a; Block 2: , P = 0.53, Fig. 3b). Furthermore, the number of eggs laid by ovipositing An. gambiae s.s. following their second blood meal did not vary between mosquitoes in the human-only and mixed host species diets (Block 1: , P = 0.62, Fig. 3c: Block 2: , P = 0.28, Fig. 3d). Finally, the estimated cumulative egg production of mosquitoes over two blood meals did not vary between those fed a diet of only human blood, or a combination of human and animal blood (Block 1: , P = 0.99, Fig. 4a, Block 2: , P = 0.99, Fig. 4b). Post hoc comparison also indicated that there were no significant differences between the three different generalist host blood diets (human–cow, human–chicken and human–dog) in terms of mosquito oviposition (Block 1: , P = 0.11, Fig. 3a, Block 2: , P = 0.52, Fig. 3b), fecundity (Block 1: , P = 0.35, Fig. 3c Block 2: , P = 0.78, Fig. 3d) or cumulative egg production over two blood meals (Block 1: , P = 0.95, Fig. 4a, Block 2: , P = 0.95, Fig. 4b).
The survival of An. gambiae s.s. after feeding on two blood meals did not vary between those fed only human blood, and those given human and cow, chicken or dog blood (Fig. 5a,b, Table 2). Similarly, no significant differences in mortality were observed between mosquitoes in the three different generalist host diet groups (Block 1: , P = 0.63, Block 2: , P = 0.65).
Table 2. Survival of Anopheles gambiae s.s. after two blood meals. The estimated odds of mortality of An. gambiae s.s. after taking two blood meals from different combinations of host species relative to human-only blood meals. The numbers in brackets are the 95% confidence intervals of the odds ratio (OR).
Host diet treatment
Human + Chicken
Human + Cow
Human + Dog
Here we experimentally investigated the consequences of host species diversity for the malaria mosquito An. gambiae s.s. to test the evolutionary prediction that organisms evolved for specialism have reduced fitness on generalist diets. Comparisons of the fitness that mosquitoes derived from single blood meals indicated the existence of some underlying variation in the resource quality of host blood, with An. gambiae s.s. having greater oviposition success and survival after feeding on human and bovine blood, than on dog or chicken. However, when evaluated within the context of the multiple feeding events that An. gambiae s.s. is expected to take during its lifetime (median of 2), neither their reproductive success nor long-term survival was influenced by the diversity of their host species diet (human-only or a mixture of human followed by cow, chicken or dog). Furthermore, the number of eggs laid by An. gambiae s.s. after one or two blood meals was unrelated to host species, and there was no evidence from either single or multiple feeding experiments that human blood gave rise to a distinct advantage over all other host types that could account for its extreme specialization in nature (e.g. human blood was associated with some advantages over avian and canine blood, but not bovine). Thus, we hypothesize that selection for specialization in this mosquito is unlikely to be generated by underlying variation in quality between their preferred human hosts and other alternatives readily available in the same environment.
Haematological properties such as red blood cell size, density and amino acid composition are known to vary between the host species investigated here (Wintrobe, 1933; Nemi, 1986; Hawkey, 1991; Hawkey et al., 1991). Such variation could be responsible for between-host differences in mosquito survival and oviposition rate we observed after one blood meal. However, there was negligible impact of interspecific haematological variation after the second blood meal, possibly because mosquitoes acquired enough nutrients from their first (human) blood meal to offset any modest deficits in the nutrient quality of later meals. As human blood was identified as one of the better resources in the single-feed experiments, it is possible that by providing this first before meals from other host species could have masked the costs of a generalist host diet in which non-human hosts were consumed first or repeatedly. However, putting together the two ‘best’ host types as identified in single-feed experiments (human and cow) did not give rise to greater fitness than a combination of one of the best and the worst (human and chicken). This suggests that the fitness effects of individual blood meals may not be additive and that any deficits arising from interspecific host haematological variation can be offset provided mosquitoes obtain more than one blood meal. Further experiments in which the order and frequency of human and animal blood meals given to An. gambiae s.s. is varied are required to confirm this.
We found no impact of host species on the number of eggs laid by ovipositing mosquitoes in either single or multiple feeding experiments. This result contrasts with some previous studies indicating that mosquito fecundity depends on host species (Woke, 1937a,b; Bennett, 1970; Downe & Archer, 1975; Mather & DeFoliart, 1983). Discrepancies between these and the current study may reflect genuine biological differences in the impact of host species on different mosquito species. Alternatively, the impact of host species may have been underestimated in this study due to the manner in which blood was presented. Here, a fixed volume of blood was presented to mosquitoes in standardized membrane feeding devices, for a fixed time period (longer than the average duration of a feeding event in the wild). This design allowed us to isolate the specific impact of blood on mosquito fitness, while controlling for variation in all other host physiological and behavioural traits that could influence blood intake. However, the lack of time limitation may have allowed mosquitoes to adapt their feeding effort (e.g. by adjusting the volume of blood taken, or the time spent feeding) to maximize their resource intake regardless of the type of blood consumed. If such compensations are less possible when mosquitoes are feeding on live hosts, the impact of inherent variation in blood quality could have a substantially greater impact on mosquito fitness in the wild than estimated here.
In this study, the long-term survival of An. gambiae s.s. was influenced by host species only when blood intake was restricted to one meal, but not when multiple meals were taken. These results are consistent with the majority of previous laboratory studies that indicate mosquito survival is unrelated to host blood meal diversity (reviewed in Lyimo & Ferguson, 2009). However, Harrington et al., 2001 did find that the survival of Aedes aegypti was significantly longer on its naturally preferred human hosts than that of laboratory rodents. This was postulated as being a consequence of variation in the isoleucine content of blood between these species, which was correlated with the acquisition of energetic reserves from blood feeding (Harrington et al., 2001). It is unknown whether variation in blood isoleucine content could account for the enhanced survival of An. gambiae s.s. after taking one blood meal from humans or cows here. However, if these host species do exhibit significant biological variation in blood isoleucine content, it had little impact on the fitness that An. gambiae s.s. acquired from their second blood meal. If the impact of host-specific variation in haematological parameters such as isoleucine concentration is highly context dependent in An. gambiae s.s. (e.g. dependent on the number and type of previous blood meals taken), it may be unlikely to provide a sufficiently reliable signal to drive selection for host choice.
Although our results indicate blood meal diversity does not have a large impact on An. gambiae s.s. survival, it is possible it does have a moderate effects that would be unlikely to detect under standard laboratory conditions. For example, as is standard practice for laboratory studies, all blood fed mosquitoes were given unlimited access to glucose and water throughout their lifetime. Sugar feeding is known to extend mosquito longevity (Gary & Foster, 2001, 2004), and although Anopheles are known to do so in nature (Beier, 1996; Gary & Foster, 2006; Gary et al., 2009), it is unlikely they do so as readily in the wild as in these laboratory conditions. Consequently, this provision of sugar could have masked differences in mosquito survival arising from their blood diet. Previous studies have also shown that sugar provision in addition to blood can reduce (Gary & Foster, 2001; Bracks et al., 2006) or enhance An. gambiae s.s. reproductive success (Manda et al., 2007). Further experiments including contrasts of blood diets in the presence and absence of sugar meals are required to confirm the predicted impacts of host species on mosquito fitness described here.
Although the existence of fitness trade-offs in resource exploitation have been widely predicted for dietary specialists (Levins, 1962; MacArthur & Pianka, 1966; Pyke et al., 1977), their existence has failed to be confirmed in several insect model systems (Agrawal et al., 2002). In these studies, it has been hypothesized that the lack of fitness differences between uniform and mixed host diets is because the plant resources incorporated in both meal types had similar nutritional value (Rapport, 1980; Hauge et al., 1998), which may also be true for the blood sources investigated here. Alternatively, failure to detect trade-offs associated with host dietary diversity in this and other studies may be a consequence of the homogeneous background of insect lines used in laboratory studies. Although these experiments used a relatively outbred line of An. gambiae s.s. (Hurd et al., 2005), the adaptations that favour specialization on humans could have been partially eroded during the colonization process and limited our ability to detect an advantage of human blood.
Although we found no evidence of an overall fitness advantage from feeding only on human blood, it is premature to dismiss the possibility that trade-offs in host exploitation under more natural conditions may explain why An. gambiae s.s. has evolved a specialist feeding strategy. For example, defensive behaviour in response to mosquito biting can vary between host species (Edman & Scott, 1987) and may generate selection for specialization on weakly protective species. Other host factors such as the ease with which their skin can be penetrated, range over which their odour cues can be detected (Gillies & Wilkes, 1972), suitability of microhabitats and relative availability will influence their encounter and exploitation rates by mosquitoes and could exert stronger selection on host choice than variation in blood quality. Further examination of these host-specific factors under natural conditions is needed to evaluate their role in structuring An. gambiae s.s. host range.
The intense specialism of An. gambiae s.s. on humans is largely responsible for its ability to maintain malaria transmission levels within its sub-Saharan African range that are well beyond those achieved in other parts of the globe. In addition to the immediate need to protect humans from their bites using insecticide-treated nets and other measures, it is possible that longer-term reductions in exposure could be achieved through environmental manipulation of the selective forces that promote anthrophily. These results suggest that transition from a human specialist to more generalist strategy need not be impeded by the relative fitness value of uniform and mixed species blood diets, and that any potential costs of generalism are more likely to arise through ecological or behavioural factors.
INL designed and executed experiments, performed data analyses and prepared the manuscript draft. SPK executed single blood meal experiments. LCR advised on methodology, reviewed the manuscript and provided comments. HMF supervised the execution of experiments, data analysis and reviewed the manuscript.
We thank Elizabeth Peat and Dorothy Armstrong for technical assistance during this study. INL was supported by a University of Glasgow Faculty Studentship. Funding was provided by a BBSRC David Philips Fellowship to HMF. Research leading to these results also received funding from the European Union Seventh Framework Programme FP7 (2007–2013) under grant agreement no 228421 Infravec.