According to the host manipulation hypothesis, parasites modify the hosts' phenotype to maximise their transmission success. Avian malaria parasites and related haemosporidians are vector-borne parasites infecting birds worldwide. Recent studies have reported a greater attraction of mosquitoes to infected birds, supporting the host manipulation hypothesis. Changes in the composition of the uropygial gland secretion of birds associated with infections have been proposed as the potential mechanisms explaining this pattern. Here we critically review the published information on the host manipulation hypothesis in the context of avian malaria infections. We focus this article on the suggested role of the secretions of the uropygial gland and bird odours as mosquito attractants. The role of uropygial gland secretions as attractants of mosquitoes was poorly supported by available literature. In contrast, changes in the odour profile of infected birds or a reduction in the anti-mosquito behaviour of infected individuals may explain the parasite-mediated effects on mosquito attraction and biting rates. Finally, we propose future research approaches to identify the role of parasite infections on the interaction between birds and insect vectors.—Martínez-de la Puente, J., Díez-Fernández, A., Soriguer, R.C., Rambozzi, L., Peano, A., Meneguz, P.G. & Figuerola, J. (2021). Are malaria-infected birds more attractive to mosquito vectors? Ardeola, 68: 251-264.
The host manipulation hypothesis argues that parasites are able to modify host phenotypes to enhance parasite transmission success. This hypothesis has been largely tested using different study models considering a diversity of parasites infecting invertebrate and vertebrate hosts, including birds (reviewed in Heil, 2016).
Avian malaria parasites belonging to the genus Plasmodium are widespread parasites of the phylum Apicomplexa, commonly found infecting birds in every continent except Antarctica (Valkiūnas, 2005). Avian Plasmodium is phylogenetically related to other common parasites also infecting birds such as the malaria-like parasites of the genera Haemoproteus and Leucocytozoon. These three parasite genera are transmitted from an infected bird to a new susceptible host by different dipteran insect vectors. Species of Plasmodium are transmitted by mosquitoes (Family Culicidae) (Santiago-Alarcón et al., 2012). Hippoboscids (Hippoboscidae) and biting midges (Ceratopogonidae) are considered the main vectors of Haemoproteus species while parasites of the genus Leucocytozoon are mainly transmitted by blackflies (Simuliidae) (Valkiūnas, 2005; Santiago-Alarcón et al., 2012). However, some exceptions have been reported to this general pattern: for example, Leucocytozoon caulleryi is transmitted by biting midges (Culicoides spp.) instead of blackflies (Valkiūnas, 2005).
Avian malaria and malaria-like parasites are important selective agents affecting the ecology, evolution and conservation of their bird hosts. These parasites are known to induce deleterious effects on bird health (Marzal et al., 2008), breeding success (Merino et al., 2000) and survival (Marzal et al., 2008; Martínez-de la Puente et al., 2010; Asghar et al., 2015), acting as drivers of some avian population declines (Van Riper III et al., 1986; Niebuhr et al., 2016; Dadam et al., 2019). Thus, factors increasing contact rates between infected and susceptible birds and insect vectors could have important consequences for individual fitness and population dynamics and should be studied. According to the host manipulation hypothesis, vector-borne avian malaria parasites may improve their transmission success by increasing the feeding rate of competent vectors on infected birds. These effects could be driven by an increase in the attractiveness of infected birds to vectors. Birds protect themselves from mosquito bites by moving their legs, wings and heads to drive away mosquitoes (Darbro & Harrington, 2007). An increase in the biting rate of mosquitoes on vertebrate hosts could be also favoured by a reduction in the anti-mosquito behaviour of infected individuals. These possibilities have been broadly studied using multidisciplinary approaches during the last years providing contrasting results. Thus, the ability of these parasites to modify bird phenotypes, and thereby the mechanisms used by parasites to increase their transmission, remain unclear. In this article, we critically review the published information on the bird manipulation hypothesis in the context of avian malaria infections and focus on the suggested role of the secretions of the uropygial gland and body odour of birds as attractants of mosquitoes. In addition, we propose future research possibilities to test the host manipulation hypothesis, considering novel approaches based on the study of the role of bird odours in determining mosquito-bird-parasite interactions.
Vector attraction towards malaria-infected birds
According to the host manipulation hypothesis, parasites manipulate the phenotype of the infected individuals to increase parasite transmission (Heil, 2016). In the case of avian malaria, the host manipulation hypothesis proposes that parasites may manipulate infected birds to attract more mosquitoes than are attracted by uninfected ones. Researchers have used different approaches to compare the attraction of mosquitoes towards infected and uninfected birds. For instance, Lalubin et al. (2012) used a dual-choice olfactometer to compare the attraction of Culex pipiens towards wild Great Tits Parus major that were uninfected or naturally infected by Plasmodium parasites. Contrary to the predictions of the host manipulation hypothesis, these authors found that mosquitoes were more attracted to uninfected birds than by infected ones. However, the use of naturally infected birds does not exclude the possibility that vectors were attracted to uninfected individuals because of some other physiological difference that could decrease their initial probability of infection. In a field experiment, a lower number of biting midges were captured in nest-boxes of Blue Tit Cyanistes caeruleus pairs medicated with the antimalarial drug primaquine than in nests of control pairs (Tomás et al., 2008). These results support the host avoidance hypothesis which argues that insect vectors may develop different mechanisms to reduce their contact rates with infected vertebrates, because the parasites may also have deleterious effects on the vectors, especially after these bite birds with high intensities of infection (Anderson et al., 2000; Valkiūnas et al., 2014; Bukauskaitė et al., 2016; Gutiérrez-López et al., 2019a).
Dipteran insect vectors use different cues, such as chemical and visual stimuli, to locate their hosts (Lehane, 2005). It has been proposed that insect vectors are attracted by the secretion of the uropygial or preen gland. This secretion contains different compounds including alcohols, aldehydes and waxes that could be used by mosquitoes to locate their hosts. If parasites modify the composition of the secretion of the uropygial gland, this could be a potential mechanism to increase the attractiveness of infected birds to mosquitoes. In this case, it could be expected that i) parasites modify the composition of the uropygial gland secretion of birds and ii) mosquitoes are more attracted to the secretions of parasite-infected birds than uninfected ones.
Grieves et al. (2018) found that malaria infection (Plasmodium sp. lineage 99% similar to P-SOSP 2) modified the wax ester composition of the secretions of the uropygial gland of Song Sparrows Melospiza melodia. However, it is unclear if the volatile and semivolatile compounds of the uropygial gland secretions of birds derived from these wax esters. By contrast, Díez-Fernández et al. (2020b) did not find differences in the composition of the volatile fraction of uropygial gland secretions of House Sparrows Passer domesticus whether uninfected or infected by Plasmodium parasites. Birds in this latter study were naturally infected by Plasmodium parasites corresponding to four different genetic lineages (SGS1, GRW11, COLL1 and PADOM01). These contrasting results suggest that differences in the effect of parasites on the composition of the secretions of the uropygial gland may vary between bird species, parasite lineages and the fractions of the secretion analysed (volatile fraction vs wax esters).
Different authors have analysed the attraction of mosquitoes towards the secretion of the uropygial gland. Russell and Hunter (2005) found that CDC traps located 5m above ground level and supplemented with uropygial gland secretions captured more Cx. pipiens and Cx. restuans mosquitoes than unbaited traps. However, this attractant effect of the bird secretions was not found when traps were located 1.5m above ground level. The use of dual choice olfactometers found no attraction toward uropygial gland secretions by Cx. pipiens or Aedes caspius mosquitoes (Díez-Fernández et al., 2020a). In addition, while live adult House Sparrows attracted more Cx. pipiens mosquitoes than nestlings, a similar degree of attraction was observed when mosquitoes were exposed to uropygial gland secretions of these bird age classes (Garvin et al., 2018). No differences were also reported by Martínez-de la Puente et al. (2011) when comparing the attraction of biting midges to miniature UV-CDC traps whether or not these were baited with the uropygial gland secretions of pigeons. Moreover, biting midges were absent in unoccupied nest-boxes baited with the uropygial gland secretions of Blue Tits Cyanistes caeruleus (Martínez-de la Puente et al., 2011). Thus, in spite of the findings of Russell and Hunter (2005), the role of uropygial gland secretions as attractants of mosquitoes is poorly supported.
Díez-Fernández et al. (2020b) found differences in the attraction of Cx. pipiens to stimuli from birds uninfected and infected by Plasmodium parasites, with a higher attraction of mosquitoes to the body odours but not the uropygial gland secretions of infected House Sparrows. In this study, the authors used birds naturally infected by avian Plasmodium parasites. Therefore, it was not possible to distinguish if the observed differences were the cause or the consequence of parasite infections. In spite of this limitation, these results support the absence of a direct impact of infection by avian Plasmodium parasites on mosquito attraction to birds by their uropygial gland secretions. While the odour of infected birds attracts more mosquitoes, these results raise new questions on the potential causes of changes in bird odour profiles and the nature of the substances involved in the higher attraction for mosquitoes associated with parasitic infections. Previous studies on other malaria models support a link between host odours, parasitic infections and vector attraction (De Moraes et al., 2014; Schaber et al., 2018). For example, Plasmodium falciparum infections are associated with the presence of terpenes (alpha-pinene and 3-carene) in the breath of children that may increase the attraction of mosquitoes towards infected individuals (Schaber et al., 2018). Some components of bird odours, such as nonanal, have been identified as attractants of mosquitoes (Syed & Leal, 2009). The odour profile of birds may be determined, at least in part, by the surface microbiota on the skin and feathers (Krause et al., 2018). Bird microbiota may be altered by properties of some components of the uropygial gland secretions (Table 1) potentially affecting birds's odours and their interaction with vectors (see Magallanes et al., 2016). In addition, uropygial gland secretions may protect plumage by forming a physical barrier to microbes (Reneerkens et al., 2008). Blood parasite infections may reduce the antimicrobial activity of uropygial gland secretions (Magallanes et al., 2016), which could explain, at least in part, the differential mosquito attraction to the odours of Plasmodium spp. infected birds and of uninfected ones. New studies are necessary to test these possibilities.
Different compounds of bird uropygial gland secretions potentially affecting bird-mosquito interactions. Some of these components may act as attractants or repellents of insect vectors as well as possibly having deleterious effects on arthropods and other organisms potentially affecting the mosquito-host interactions.
[Diferentes compuestos de las secreciones de la glándula uropigial de las aves que podría afectar las interacciones ave-mosquito. Algunos de estos compuestos podrían actuar como atrayentes o repelentes de insectos vectores así como podrían tener efectos deletéreos sobre los artrópodos y otros organismos, afectando potencialmente las interacciones mosquito-hospedador.]
Mosquito biting rates and avian malaria infections
After reaching a host bird, mosquitoes and other dipteran insect vectors feed on blood by biting unfeathered parts such as tarsi or eye-rings. Different methods have been used to identify the susceptibility of avian malaria-infected and uninfected birds to mosquito attacks by exposing birds individually (Gutiérrez-López et al., 2019b) or in pairs (Cornet et al., 2013a, 2013b; Yan et al., 2018) to insect bites under controlled conditions in the laboratory (Figure 1). In the latter case, molecular techniques were used to assess the individual origin of the blood meals present in the abdomens of vectors. It was found that more Cx. pipiens mosquitoes bit Canaries Serinus canaria that were experimentally infected with P. relictum lineage SGS1 than bit uninfected birds. However, such differences were only significant during the chronic phase of infection. Similarly, Yan et al. (2018) performed two experiments to test whether the infection status or the intensity of infection affect the number of bites that each bird received from mosquitoes. Although the rates at which infected and uninfected individuals were bitten did not differ in this study, they found that mosquitoes fed at a greater rate on birds infected by Plasmodium parasites (controls) than they did on those infected individuals with an experimentally reduced parasite load. However, contrary to the case of Cornet et al. (2013a), birds tested by Yan et al. (2018) were free to respond to mosquitoes attempting to bite them and, consequently, the results were influenced both by the extent of attraction of mosquitoes and by the susceptibility of the birds to mosquito bites. This is especially important because fewer mosquitoes may be able to complete a blood meal on birds that display more active anti-mosquito behaviour, such as preening or scratching. Overall, these studies support the hypothesis that prior infection with avian Plasmodium parasites affects the attraction of Cx. pipiens mosquitoes to bird hosts, with the phase of infection and/or parasite load in the host likely eliciting a stronger response by mosquitoes. Thus, the infection characteristics of the host need to be considered in epidemiological studies of avian malaria parasites because these variables can explain part of the heterogeneity in attacks by vectors (Cornet et al., 2013a; Yan et al., 2018), the success of parasite development in mosquitoes (Pigeault et al., 2015) and the impact of parasite infections on the insect vectors (Bukauskaitė et al., 2016; Gutiérrez-López et al., 2019a).
Interestingly, the infection by Plasmodium parasites may also affect the behaviour of vectors (Rossignol et al., 1986; Choumet et al., 2012). Parasites may increase their transmission in different ways including an increase in the frequency or duration of the contacts between insect vectors and susceptible hosts (Heil, 2016). Aedes aegypti mosquitoes carrying P. gallinaceum sporozoites ingested a lower blood volume during a blood meal and were more likely to probe for a second meal than uninfected mosquitoes (Koella, 2002). Further experiments have revealed that mosquitoes carrying the infective forms of avian malaria parasites need more time to complete a blood meal than uninfected mosquitoes and are more likely to take reduced blood meals (Rossignol et al., 1986). Thus, mosquitoes may bite several times to obtain a complete blood meal. This change in mosquito behaviour could be explained by changes in the regulation/activity of the enzyme apyrase, which is involved in the anticoagulation of mosquito saliva (Rossignol et al., 1984; Thiévent et al., 2019).
Concluding remarks and future study possibilities
The host manipulation hypothesis has mainly been tested considering a handful of bird-parasite-vector assemblages and virtually no study has analysed the Haemoproteus/biting midge-hippoboscid nor Leucocytozoon/black fly systems. This is especially significant considering the large number of vector species potentially involved in the transmission of these parasites, which may affect the observed patterns of host-parasite-vector associations. In addition, more than 4,000 lineages of avian malaria and malaria-like parasites have been recorded to date (according to Malavi; Bensch et al., 2009). These lineages may present different virulences in their bird hosts (Ilgūnas et al., 2019a; 2019b) that may determine different degrees of vector attraction to infected individuals. However, the relationship between parasite virulence and vector attractivity has never been tested. Interspecific differences in the intensity of anti-mosquito behaviours (Darbro & Harrington, 2007) and bird body size and coloration (Yan et al., 2017) may determine the differential susceptibility of each bird species to vector attacks, with some species being preferred by mosquitoes while others are bitten less often than expected from their densities in the wild (Simpson, 2009; Rizzoli et al., 2015). The different vector species involved in the transmission of avian malaria and malaria-like parasites may use different cues to locate their bird hosts, potentially explaining discrepancies between studies.
It is also important to standardise the methods used in studies considering such aspects as the use of immobilised/free moving exposed birds and the number of insect vectors and/or hosts included in each experimental trial. For example, studies using immobilised birds may provide information on the importance of such cues as odour or temperature in host selection by mosquitoes, while using free-moving birds allows the impact of anti-mosquito behaviour on mosquito feeding success to be considered. In addition, anti-mosquito behaviours may be more frequent/intense in trials using a greater number of mosquitoes (Darbro & Harrington, 2007). The number of birds exposed to mosquitoes may also affect the relative attractiveness of a specific host with respect to those available nearby, finally affecting the feeding patterns of insect vectors. For example, in cavity nesting species, nestlings are exposed to insect attacks in close proximity and consequently the bites received by an individual will be the outcome of the relative attractiveness to mosquitoes of this individual in relation to its nest mates (Christe et al., 1998). However, most studies conducted until now have been performed by exposing birds to vector bites individually or in pairs. In addition, different host-related factors including prior experience of mosquito attacks, haematocrit, body temperature or sex classes could be also important determinants of the susceptibility of individuals to vector attacks. Again, in the case of bird sex, experiments analysing its impact on mosquito feeding preferences have given mixed results, because differences have only been reported in some bird/mosquito species combinations (Gutiérrez-López et al., 2019b; Cozzarolo et al., 2019; Burkett-Cadena et al., 2014). For example, Gutiérrez-López et al. (2019b) exposed House Sparrows Passer domesticus and Jackdaws Corvus monedula to the bites of two mosquito species, Cx. pipiens and Ae. caspius. In this study, Ae. caspius showed a higher biting rate on female Jackdaws than on males. However, there was no significant difference in the biting rates of either mosquito on House Sparrows nor in the case of Jackdaws exposed to Cx. pipiens. Nevertheless, mixed infections by different parasite genera of avian malaria and malaria-like parasites are frequently found in wild birds (Marzal et al., 2008; Martínez et al., 2009; Ciloglu et al., 2019). In addition, infections by other vector-borne blood parasite taxa, such as microfilariae, Trypanosoma spp., Hepatozoon spp. and Lankesterella-like parasites, are common in birds (Merino et al., 1997; Merino et al., 2007). These mixed infections could affect the attractiveness of avian hosts to insect vectors in different ways as these parasites are transmitted by different vector groups. In addition, parasites could induce strong deleterious effects on non-competent insects, potentially favouring an avoidance rather than an attractive effect. For instance, Haemoproteus parasites, which are transmitted by biting midges (Culicoides spp.), increase mortality in mosquitoes (Valkiūnas et al., 2014). Thus, bird-mosquito interactions may be driven by arms-races between mosquitoes and parasites, with the latter potentially increasing the attraction of mosquitoes to infected individuals and selection acting on mosquitoes to reduce contact rates with the more virulent parasites that infect birds.
This study was supported by projects CGL2015-65055-P and PGC 2018-095704-B-I00 from the Spanish Ministry of Economy and Competition and from the European Regional Development Fund (FEDER). In addition, JMP received funding from a 2017 Leonardo Grant for Researchers and Cultural Creators, of the BBVA Foundation. The BBVA Foundation accepts no responsibility for the opinions, statements and contents included in the project and/or the results thereof, which are entirely the responsibility of the authors. ADF was supported by a Severo-Ochoa grant from the Spanish Ministry of Economy and Competitiveness (SVP-2014-068571). The comments of J.A. Masero and two anonymous referees improved a previous version of this manuscript.
Study conception: JMP, ADF, PGM, JF; Writing/manuscript preparation-writing the initial draft: JMP; Writing/manuscript preparation-critical review, commentary or revision: all authors; Funding acquisition: JMP, RCS, JF.