Introduction
Sex biases in disease prevalence are increasingly recognised in studies of natural populations (Poulin, Reference Poulin1996; Schalk and Forbes, Reference Schalk and Forbes1997; Moore and Wilson, Reference Moore and Wilson2002; Foo et al., Reference Foo, Nakagawa, Rhodes and Simmons2017). Much of the scientific work on natural populations has been focused on proximate physiological reasons for differences in infection rates and susceptibility between the sexes, such as the interaction between sex hormones and the immune system (Folstad and Karter, Reference Folstad and Karter1992; Poulin, Reference Poulin1996; Zuk and McKean, Reference Zuk and McKean1996; Klein, Reference Klein2004). Many of these intrinsic factors have been shown to skew infection biases towards males, mostly due to an antagonistic relationship between testosterone and immune function (Foo et al., Reference Foo, Nakagawa, Rhodes and Simmons2017).
However, ecological factors may have a direct proximate role in causing sex-biased infection rates due to behavioural differences between the sexes that, in turn, leads to differential exposure to parasites (Tinsley, Reference Tinsley1989; Krasnov et al., Reference Krasnov, Morand, Hawlena, Khokhlova and Shenbrot2005; Zuk and Stoehr, Reference Zuk, Stoehr, Klein and Roberts2010; Brown and Symondson, Reference Brown and Symondson2014). In a multispecies comparison of sex-specific parental roles in birds, the different breeding behaviours of males and females were suggested as a potential explanation for higher parasite prevalence in females (McCurdy et al., Reference McCurdy, Shutler, Mullie and Forbes1998). For example, in species where only 1 of the 2 sexes incubate the eggs, these stationary individuals were suggested to provide easier targets for the biting insects that transmit avian malaria than individuals of the other sex (McCurdy et al., Reference McCurdy, Shutler, Mullie and Forbes1998). Other studies have argued that this may be especially true for hole-nesting species, where incubating females were suggested to increase the number of ectoparasites and pathogen vectors in the nest by acting as ‘beacons’ of vector attracting compounds, such as CO2 and volatile organic compounds (Caillouët et al., Reference Caillouët, Riggan, Bulluck, Carlson and Sabo2013; Lutz et al., Reference Lutz, Hochachka, Engel, Bell, Tkach, Bates, Hackett and Weckstein2015; Castaño-Vázquez et al., Reference Castaño-Vázquez, Merino, Cuezva and Sánchez-Moral2020). By contrast, in open-cup nesting species male biased infection rates are more common (van Oers et al., Reference van Oers, Richardson, Sæther and Komdeur2010; Lutz et al., Reference Lutz, Hochachka, Engel, Bell, Tkach, Bates, Hackett and Weckstein2015; Calero-Riestra and García, Reference Calero-Riestra and García2016). However, direct links between sex-biased behaviour and the risk of infection are rarely revealed because the timing of infection often remains unknown in studies of natural populations. Here we circumvent this problem by using detailed life-history information, from a long-term study of individually marked collared flycatchers (Ficedula albicollis); to establish the timing of infection and to test expectations consistent with either physiological or behavioural differences being the main cause of a sex-biased infection rate.
Collared flycatchers are migratory, sexually dimorphic, passerine birds that breed across Europe and overwinter in sub-Saharan Africa (Cramp et al., Reference Cramp, Perrins, Brooks and Dunn1993). Males and females have different behaviours at the nest, with females taking sole responsibility for incubation, while males find the nesting location and occasionally visit the nest to supplement the feeding of incubating females (Lifjeld and Slagsvold, Reference Lifjeld and Slagsvold1986). Both sexes contribute to feeding nestlings (Pärt et al., Reference Pärt, Gustafsson and Moreno1992).
Avian malaria is a disease commonly caused by two, well studied haemosporidian parasite genera: Haemoproteus and Plasmodium. Haemosporidians (herein avian malaria) are blood-borne parasites that require both a vertebrate and an insect host to complete their life cycle and its transmission occurs globally (Valkiūnas, Reference Valkiūnas2005). Haemoproteus is largely transmitted by Culicoides biting midges, many species of which directly target and prefer to feed in nest holes (Votýpka et al., Reference Votýpka, Synek and Svobodová2009; Žiegytė et al., Reference Žiegytė, Platonova, Kinderis, Mukhin, Palinauskas and Bernotienė2021), while Plasmodium is more frequent in mosquitoes (Valkiūnas, Reference Valkiūnas2005), such as members of the genus Culex, which prefer to seek for hosts in the open (Ryan et al., Reference Ryan, Lippi, Boersch-Supan, Heydari, Silva, Adrian, Noblecilla, Ayala, Encalada, Larsen, Krisher, Krisher, Fregosi and Stewart-Ibarra2017). Some avian malaria lineages are specific to just one host species, while others have been found in a wide range of possible hosts (Bensch et al., Reference Bensch, Hellgren and Pérez-Tris2009; Clark et al., Reference Clark, Clegg and Lima2014; Ellis et al., Reference Ellis, Huang, Westerdahl, Jönsson, Hasselquist, Neto, Nilsson, Nilsson, Hegemann, Hellgren and Bensch2020). This potential for high specificity, coupled with increasingly rich sampling efforts around the world provides the opportunity to ascertain the potential transmission zones for many avian malaria lineages. To date, at least 4500 unique lineages have been detected in 2100 avian host species from all continents, but Antarctica (Bensch et al., Reference Bensch, Hellgren and Pérez-Tris2009; Ellis et al., Reference Ellis, Sari, Rubenstein, Dickerson, Bensch and Ricklefs2019). Collared flycatchers are commonly infected with avian malaria parasites, with prevalence being as high as 40% in some populations (Kulma et al., Reference Kulma, Low, Bensch and Qvarnström2013; Szöllősi et al., Reference Szöllősi, Garamszegi, Hegyi, Laczi, Rosivall and Török2016; Jones et al., Reference Jones, Kulma, Bensch, Cichoń, Kerimov, Krist, Laaksonen, Moreno, Munclinger, Slater, Szöllősi, Visser and Qvarnström2018). Additionally, an apparent transmission location has been identified for many of these lineages, with collared flycatchers gaining infections in both their breeding and non-breeding ranges (Jones et al., Reference Jones, Kulma, Bensch, Cichoń, Kerimov, Krist, Laaksonen, Moreno, Munclinger, Slater, Szöllősi, Visser and Qvarnström2018). However, few studies to date, have been comprehensive enough to simultaneously investigate the patterns of avian malaria prevalence and diversity across host sex and host age categories to establish when in life and where (i.e. at the breeding grounds or at the wintering grounds) infection occur, nor have consistent patterns of sex-specific infection been detected.
In this study, we used a long-term dataset of breeding collared flycatchers and their avian malaria parasites to test whether sex-differences in parasite infection exist, how parasite prevalence changes across age categories and whether malaria lineage communities differ between the sexes. If hole-nesting behaviour imposes an increased risk of infection, we expect to find female biased infection rates solely among malaria lineages transmitted at the breeding sites in Europe and that this bias builds up across age-classes following repeated breeding events.
Materials and methods
Sampling and screening
Since 2002, over 2000 nest boxes have been systematically monitored for breeding collared flycatchers on the Swedish island of Öland (56°44′N 16°40′E) (Qvarnström et al., Reference Qvarnström, Rice and Ellegren2010). Between 2002 and 2016, during each breeding season (May–June), male and female flycatchers were caught at the nest and, if necessary, ring-marked and roughly 30ɥl of blood was collected from each bird and stored in ethanol. Females were mostly caught in the middle of the incubation period and males were caught approximately 10 days later while feeding nestlings. There is some evidence that prevalence can show an apparent decrease during the breeding season, even within the space of a few weeks, in collared flycatchers (Szöllősi et al., Reference Szöllősi, Garamszegi, Hegyi, Laczi, Rosivall and Török2016). Therefore, sampling day (May 1st = 1 – July 4th = 65) was noted and included in analyses to account for any potential sampling-date bias. Smaller numbers of both sexes were caught earlier in the breeding season during the courtship period. Unringed flycatchers were aged as either 1 year old or older based on plumage features. Males, by brown, rather than black wing feathers (Svensson, Reference Svensson1992) and females by the shape and wear of their primary coverts (worn and pointed in first-year females, fresh and rounded in older females) (Pärt et al., Reference Pärt, Gustafsson and Moreno1992; Evans et al., Reference Evans, Gustafsson and Sheldon2011). For the purposes of this study, individuals were only screened once during their lifetime. In total, 728 individuals were included from previous studies in this system (Kulma et al., Reference Kulma, Low, Bensch and Qvarnström2013, Reference Kulma, Low, Bensch and Qvarnström2014; Jones et al., Reference Jones, Kulma, Bensch, Cichoń, Kerimov, Krist, Laaksonen, Moreno, Munclinger, Slater, Szöllősi, Visser and Qvarnström2018) and 506 were newly sampled. Nestlings were not screened for avian malaria parasites, as infections are not typically detectable for several weeks after initiated (Cosgrove et al., Reference Cosgrove, Knowles, Day and Sheldon2006). However a previous study on avian malaria prevalence in fledged collared flycatchers detected some infections, suggesting that transmission is indeed occurring before they migrate to Africa (Fletcher et al., Reference Fletcher, Träff and Gustafsson2019). For a detailed list of sample sizes for each sex and age category, see Table 1. Ethical permissions were provided by the Linköping Animal Ethics Board (5-2-18—7556/14).
Distribution of avian malaria lineages in collared flycatchers from Öland, Sweden with assigned transmission locations
DNA was extracted via the high salt technique (Aljanabi and Martinez, Reference Aljanabi and Martinez1997). Briefly, blood was digested overnight in a solution of proteinase K, SDS and Tris and EDTA buffers. DNA was precipitated out using 6 M NaCl and 99% ethanol. Extracted DNA was stored in TE buffer. DNA concentration was quantified using a NanoDrop2000 (Thermo Scientific) and diluted to a concentration of approximately 25 ng μL−1. To ascertain infection status, DNA extracts were screened for Plasmodium and Haemoproteus presence using an established nested PCR technique, targeting a fragment of the cytochrome b mitochondrial gene, using 2 sets of primer pairs (Waldenström et al., Reference Waldenström, Bensch, Hasselquist and Östman2004). Firstly, primers HAEMNF and HAEMNR2, which amplify a 580 base pair DNA fragment, followed by a second round with the primers HAEMF and HAEMR2, which amplify a final 478 base pair fragment. Negative (ddH2O) and positive controls were included to control for possible contamination and amplification failure during PCRs. PCR products were then stained with GelGreen and visually inspected for malaria presence or absence on a 1.5% agarose gel. Positive samples were Sanger sequenced to ascertain lineage identity. Sequences were aligned using Mega7 software and compared with previously published sequences in the publicly available MalAvi database, which collates lineage-specific infection records from around the world (Bensch et al., Reference Bensch, Hellgren and Pérez-Tris2009). Confirmed multiple infections in a single host were rare (2/1234). Both of these individuals were infected with 2 Haemoproteus lineages (hCOLL3 and hPHSIB1), as both of these lineages are also transmitted in Europe, they were treated as a single infection in our analyses. Furthermore, Individuals for which lineage could not be ascertained (100/1234) could, in some cases, refer to sequencing failure due to multiple infections.
Assigning transmission location
Records of infection from all lineages detected in the individuals from this study were extracted from the MalAvi database. In total, 293 host species were found to share malaria lineages with collared flycatchers. Infection information from captive birds or from experimental infections were removed from the analysis, as were records flagged as likely contamination errors (Bensch et al., Reference Bensch, Inumaru, Sato, Lee Cruz, Cunningham, Goodman, Levin, Parker, Casanueva, Hernández, Moreno-Rueda and Rojo2021). The ranges for each host species were categorised by ecozone, as defined by Schultz (Reference Schultz1995), using range maps from the ‘Birds of the World’ website (Billerman et al., Reference Billerman, Keeney, Rodewald and Schulenberg2021). Each species was determined to be either migratory (individuals moved between ecozones) or resident (any movements remained within 1 ecozone). Lineages with likely transmission on the breeding grounds were determined if the lineage had been detected in resident species in the Palearctic ecozone. Lineages that were not detected in resident, Palearctic host species were determined to be transmitted during the non-breeding season (Jones et al., Reference Jones, Kulma, Bensch, Cichoń, Kerimov, Krist, Laaksonen, Moreno, Munclinger, Slater, Szöllősi, Visser and Qvarnström2018). A full table of lineages and their apparent transmission zones can be found in the supplementary materials (Table S2).
Individuals from which sequences were unable to be resolved (100 individuals) were included in analyses of overall infection prevalence but excluded from transmission specific analyses. A complete list of infection records can be found in Table 1.
Statistical analyses
All analyses were conducted using R version 4.0.5 (R Development Core Team, 2021). To investigate the patterns of malaria prevalence in collared flycatchers, we constructed 8 generalised linear mixed-effects models, exploring: a, overall infection prevalence; b, infection prevalence of European-transmitted lineages; c, infection prevalence of African-transmitted lineages; d, infection prevalence of lineages with unknown-transmission; e, infection prevalence of all Haemoproteus lineages; f, infection prevalence of hPHSIB1 (the most abundant lineage); g, infection prevalence of all other Haemoproteus lineages and h, infection prevalence of Plasmodium lineages using the package ‘lme4’ (Bates et al., Reference Bates, Mächler, Bolker and Walker2015) with binomial error structures and a logit links. Models had the same error structure with binomial fixed effects of sex and age (young = 1st breeding year and old ⩾2nd breeding year) and the interaction between the 2. Furthermore, sampling day was included as a fixed effect, as models including it as a random effect returned a singular fit. Finally, year was included as a random variable. Models exploring African-transmitted and Plasmodium lineages failed to converge, therefore generalised linear models, without random effects, were employed instead. Estimates of effect sizes (odds ratios) were calculated for infection probabilities between age and sex classes to confirm the strength of association between infection and age and infection and sex.
To compare lineage diversity between the sexes, we calculated the Shannon diversity indices and compared them using a Hutcheson t-test, using the package ‘vegan’ (Hutcheson, Reference Hutcheson1970; Dixon, Reference Dixon2003). Hutcheson t-tests are a modified version of the classic t-test that provides a method to compare 2 samples by incorporating the variance of the Shannon diversity index measures. To test whether lineage communities in young or old and male or female flycatchers was structured or not, we used analyses of similarity (ANOSIMs) in package ‘vegan’. ANOSIM detects differences between 2 or more sampling units by comparing between group dissimilarity with the mean of within group dissimilarity. All tests were done using a ‘Bray–Curtis’ dissimilarity matrix with 99 999 iterations. As well as a P value, ANOSIM provides an R statistic that indicates the extent to which groups are separated. R values near to 0 suggest community similarity while R values close to 1 suggest complete community dissimilarity. In addition, indicator lineages, i.e. lineages that occur more frequently in 1 group than expected were detected using the ‘mulitplatt’ function in the package ‘indicspecies’ (De Cáceres et al., Reference De Cáceres, Jansen and De Caceres2016). Finally, to visualise the relatedness of the different lineages and their relative proportions in each sex, a minimum-spanning network (Bandelt et al., Reference Bandelt, Forster and Röhl1999) of detected lineages was created by using the 478 base-pair cytochrome b DNA fragment in the program PopART (Leigh and Bryant, Reference Leigh and Bryant2015).
Results
From 1234 (628 female, 606 male) adult collared flycatchers screened for infections with Plasmodium and Haemoproteus we found an overall parasite prevalence of 36.6% by 25 distinct lineages, amounting to 20 Plasmodium and 5 Haemoproteus lineages. Infection rate in first year individuals was not significantly different between males and females but older females had a significant rise in malaria prevalence. This was not observed among male age classes (Fig. 1, Table 2a). Odds ratio calculations suggested that older females were 30% more likely to be infected than first year females (odds ratio: 2.041, 95% CI: 1.428–2.935, p < 0.001).
Overall avian malaria prevalence across age classes in male (blue) and female (red) collared flycatchers with 95% confidence intervals. First year individuals experience similar infection rates, however older females experience a higher risk of infection.
Generalised linear mixed-effects models (GLMM) and generalised linear effects models (GLM) evaluating the role of age and sex in explaining malaria infection prevalence including; (a) all infections (b) only European-transmitted infections (c) only African-transmitted infections (d) only infections with unknown-transmission location (e) only Haemoproteus infections (f) only hPHSIB1 infections (g) only non-hPHSIB1 Haemoproteus infections, and (h) only Plasmodium infections in collared flycatchers. Significant values highlighted thusly (*>0.05; **>0.01; ***>0.001)
When infections were classified by apparent transmission location, we found that, the female biased age-dependent rise in malaria prevalence was driven by infections with European lineages (Fig. 2A, Table 2b). Both sexes were more likely to be infected with African lineages with increasing age but there was no significant effect of sex or the interaction between age and sex on the likelihood of infection with African lineages (Fig. 2B, Table 2c), and the same was true for Plasmodium infections (Table 2d). Our analyses of infection prevalence of all Haemoproteus lineages and of hPHSIB1 infections specifically both revealed the same difference between the sexes with females being significantly more likely to be infected due to an increased infection rate with age that was absent in males. There were no significant effects of age or sex in the model investigating infection patterns of Haemoproteus lineages when infection with hPHSIB1 were excluded, implying that it is the most common lineage, i.e. hPHSIB1, that is the main driver of the overall female biased infection rate. There were also no significant differences in Plasmodium infection prevalence between the 2 sexes (Table 2e–h). Sampling day did not have any significant effect on infection prevalence in any of the models.
Prevalence of European-transmitted (A), African-transmitted (B), lineages of unknown transmission (C), Haemoproteus (D), hPHSIB1 (E), non hPHSIB1 Haemoproteus infections, and Plasmodium (G) lineages in male (blue) and female (red) collared flycatchers with 95% confidence intervals. Older females experience a significant increase in infection risk with European, overall Haemoproteus and hPHSIB1 lineages. Both sexes experience a similar rate of increase in African, unknown-transmission, non hPHSIB1 and Plasmodium lineages over time.
We found fewer overall lineages in males than in females (16 vs 21), however this did not translate to a significant difference in the Shannon diversity of malarial lineages between males (± 95% CI) (H = 1.354 ± 0.259) or females (H = 1.416 ± 0.215), (Hutcheson t-test: t = 0.367; df = 300; p = 0.714). Lineage diversity tended to be higher in older flycatchers (22 lineages; H = 1.475 ± 0.199) than younger birds (12 lineages; H = 1.208 ± 0.297), however this was not statistically significant (Hutcheson t-test: t = 1.497; df = 181; p = 0.136). Finally, malarial lineage communities were not structured in collared flycatchers either by age (R = −0.5, p = 1.000) or sex (R = 1, p = 0.333), with all of the most common lineages were shared between age and sex groups (Fig. 3, Table S3).
Minimum-spanning network of mistochondrial haplotypes of avian malaria lineages, based on a 478 base pair cytochrome b fragment. The size of each haplotype represents the number of individual collared flycatchers carrying that particular lineage. The colours denote the sex and age of the individuals: first-year male (light blue), adult male (dark blue) and young fesmale (pink) and adult female (red).
Discussion
Few previous studies have assessed age-dependant malaria prevalence. Two such examples are a study of common house martins (Delichon urbicum) showing a general increase (Marzal et al., Reference Marzal, Balbontín, Reviriego, García-Longoria, Relinque, Hermosell, Magallanes, López-Calderón, de Lope and Møller2016) and a study on Seychelles warblers showing a general decline (van Oers et al., Reference van Oers, Richardson, Sæther and Komdeur2010; Hammers et al., Reference Hammers, Komdeur, Kingma, Hutchings, Fairfield, Gilroy and Richardson2016) in parasite prevalence with age across both sexes. Parasite prevalence is often expected to decrease across age groups, as the most susceptible individuals die young and the surviving older individuals develop immunity to the parasite (van Oers et al., Reference van Oers, Richardson, Sæther and Komdeur2010; De Nys et al., Reference De Nys, Calvignac-Spencer, Thiesen, Boesch, Wittig, Mundry and Leendertz2013; Lynsdale et al., Reference Lynsdale, Mumby, Hayward, Mar and Lummaa2017). By contrast, temporarily or even consistent increases, as observed in the study on house martins, can be expected when parasite taxa that are more benign predominate, and when infections are chronic leading to a build-up of infections as the total time of possible exposure increases (Wood et al., Reference Wood, Childs, Davies, Hellgren, Cornwallis, Perrins and Sheldon2013). In our study, we find that malaria infection rates increase with age, but only in females and only when Haemoproteus infections acquired in Europe are considered. Importantly, we find no sex-difference in infection rate among the first-year breeders. Our analyses furthermore reveal that it is the most abundant Haemoproteus lineage, hPHSIB1, that drives the observed sex-different infection pattern, with females being significantly more likely to be infected later in life than males (Table 2f). The crucial question then becomes whether the observed female-biased built up of malaria prevalence proximately is caused by physiological or behavioural differences between the 2 sexes. While untested sex-specific physiological factors may have some influence on these patterns (Hasselquist et al., Reference Hasselquist, Lindström, Jenni-Eiermann, Koolhaas and Piersma2007), we argue that several lines of evidence allow us to reject the proximate physiological explanation as the main driver of sex-specific haemosporidian infection patterns in collared flycatchers. There is a time-lag between the event when the bird gets the infection due to a vector bite and the advanced stage of infection needed for detection of an ongoing infection based on the methods that we have used (Valkiūnas, Reference Valkiūnas2005; Cosgrove et al., Reference Cosgrove, Knowles, Day and Sheldon2006). This means that the vast majority of first year breeding birds with verified positive infection status with European lineages were infected during the preceding short time-window between hatching and migrating to the non-breeding grounds (Fletcher et al., Reference Fletcher, Träff and Gustafsson2019). Insect vectors, particularly biting midges from the genus Culicoides are known to target nests and nesting birds, particularly those of hole-nesting species (Martínez-De La Puente et al., Reference Martínez-De La Puente, Merino, Tomás, Moreno, Morales, Lobato, Talavera and Sarto I Monteys2009; Votýpka et al., Reference Votýpka, Synek and Svobodová2009; Caillouët et al., Reference Caillouët, Riggan, Rider and Bulluck2012; Žiegytė et al., Reference Žiegytė, Platonova, Kinderis, Mukhin, Palinauskas and Bernotienė2021). As a result, male and female nestlings and, to a lesser extent, fledglings likely present a similar target to biting insects since there are no major sex differences in host-behaviour during this period (Cozzarolo et al., Reference Cozzarolo, Sironi, Glaizot, Pigeault and Christe2019). The lack of a difference in infection rate between first year males and females therefore suggests that any intrinsic physiological differences in attracting infections between male and female collared flycatcher nestlings are small or even slightly male biased (as we find a slight non-significant male biased infection among first-year breeding birds) (Burkett-Cadena et al., Reference Burkett-Cadena, Ligon, Liu, Hassan, Hill, Eubanks and Unnasch2010).
Sex differences in composition of malaria lineages associated with corresponding variation in virulence may result in higher observed prevalence in the sex carrying the most benign lineages as selective removal would occur in the sex carrying the most virulent lineages. However, we find no strong evidence for differing avian malaria communities in the 2 sexes, with lineage communities in males vs females being entirely unstructured (Fig. 3), with all the most abundant lineages being shared between the sexes. In addition, while 1 study found no sex-specific survival difference for infected collared flycatchers (Kulma et al., Reference Kulma, Low, Bensch and Qvarnström2014), more recent work has found that female collared flycatchers, and not males, may actually suffer a subtle increase in mortality when infected with haemosporidians (Jones, Reference Jones2019). This means that we can rule out selective removal of infected males as a possible explanation to the observed female bias in the built up of prevalence across age classes. We moreover consider sex-biased clearance to be an unlikely explanation to these results. Relatively short-lived bird species such as collared flycatchers and common house martins are not expected to invest as many resources into managing and clearing infections as longer-lived species such as the Seychelles warbler where clearance has been observed (Miller et al., Reference Miller, White and Boots2007). Anecdotally, we have observed rare incidences of apparent infection clearances in both sexes (females 8%, males 6% of repeated birds), although whether this is due to poor detection of dormant infections or whether birds really clear the infections remains to be investigated (Jarvi et al., Reference Jarvi, Schultz and Atkinson2002). This fact thereby rules out more effective clearance of disease by males as a likely explanation to the sex-difference in age-dependent prevalence that we find.
Sex-differences in breeding ecology and associated behaviours may be a major cause of differences in exposure to vectors, which in turn may result in higher frequency of infections observed among females following breeding events. Females spend more time inside the nest, incubating the eggs and keeping newly hatched nestling warm, meaning that they may be more exposed to vectors than males (Martínez-De La Puente et al., Reference Martínez-De La Puente, Merino, Tomás, Moreno, Morales, Lobato, Talavera and Sarto I Monteys2009; Tomás et al., Reference Tomás, Zamora-Muñoz, Martín-Vivaldi, Barón, Ruiz-Castellano and Soler2020). Studies on other species have found that females have elevated metabolic rates during incubation and as an increased metabolic rate increases the quantity of vector attractive compounds such as CO2 and volatile organic compounds, this makes females more of a target (De Heij et al., Reference De Heij, Van Der Graaf, Hafner and Tinbergen2007; Nord et al., Reference Nord, Sandell and Nilsson2010).
We find that male collared flycatchers mainly obtain avian malaria early in life, most likely during their stationary nestling stage while females maintain a high risk of infection throughout their lives following repeated breeding events. Taken together, these findings are most compatible with a difference in the breeding behaviours of males and females playing a major role in explaining the observed overall sex-biased infection rate. This conclusion is further supported by the fact that we only find a female specific age-dependent built up of infections of malaria lineages transmitted in Europe while a similar pattern is not found when considering malaria lineages transmitted in Africa. We also argue that a bias in sampling time of males and females is an unlikely alternative explanation to the observed sex difference in age-dependent infection patterns. This is because we find very little evidence for a change in infection patterns over the course of single breeding season in almost all cases (with the exception of a small, but significant decrease in some of the rarer Haemoproteus infections over time), a pattern also detected in another collared flycatcher population (Szöllősi et al., Reference Szöllősi, Garamszegi, Hegyi, Laczi, Rosivall and Török2016).
Further work still needs to be done to resolve the transmission zones of several of the malarial lineages detected in this study. For some lineages, such as pSGS1, transmission appears to occur globally (Bensch et al., Reference Bensch, Hellgren and Pérez-Tris2009; Marzal et al., Reference Marzal, García-Longoria, Cárdenas Callirgos and Sehgal2015) making it impossible to infer where flycatchers were infected with this lineage. It is possible that some of the lineages currently determined to be transmitted in Africa may also be transmitted in Europe or vice-versa. Furthermore, while fewer studies have collected samples in Africa than in Europe, there have still been several important community-wide studies that have given a clear picture of lineage prevalence and range, at least for the most common lineages (Loiseau et al., Reference Loiseau, Harrigan, Robert, Bowie, Thomassen, Smith and Sehgal2012; Lutz et al., Reference Lutz, Hochachka, Engel, Bell, Tkach, Bates, Hackett and Weckstein2015; Tchoumbou et al., Reference Tchoumbou, Mayi, Malange, Foncha, Kowo, Fru-cho, Tchuinkam, Awah-Ndukum, Dorazio, Nota Anong, Cornel and Sehgal2020). Yet, further work is still needed to ascertain the exact parameters for successful transmission for these most abundant lineages. However, we think it is unlikely that such changes in transmission classification will result in changes to our interferences on sex-biases in infection risk in collared flycatchers, due to the scarcity of many of the lineages. Nevertheless, there remains a large gap in our knowledge regarding the exact transmission pathways of hPHSIB1, the most abundant lineage in collared flycatchers. This lineage appears to be most common at higher latitudes in the Palearctic (Huang et al., Reference Huang, Ellis, Jönsson and Bensch2018; Jones et al., Reference Jones, Kulma, Bensch, Cichoń, Kerimov, Krist, Laaksonen, Moreno, Munclinger, Slater, Szöllősi, Visser and Qvarnström2018), suggesting dependence on more northerly vectors or colder climates for successful transmission, thereby making it unlikely that transmission also occurs in Africa. A recent study found hPHSIB1 sporozoites in the salivary glands of the biting midge Culicoides segnis (Chagas et al., Reference Chagas, Hernández-Lara, Duc, Valavičiūtė-Pocienė and Bernotienė2022). However, C. segnis is a relatively uncommon species in Scandinavia (Ander et al., Reference Ander, Meiswinkel and Chirico2012), and given the abundance of hPHSIB1 in collared flycatchers in our population and its presence in several other Eurasian passerines (Hellgren et al., Reference Hellgren, Waldenström, Peréz-Tris, Szöllősi, Hasselquist, Krizanauskiene, Ottosson and Bensch2007; Palinauskas et al., Reference Palinauskas, Iezhova, Križanauskienė, Markovets, Bensch and Valkiūnas2013; Ellis et al., Reference Ellis, Huang, Westerdahl, Jönsson, Hasselquist, Neto, Nilsson, Nilsson, Hegemann, Hellgren and Bensch2020), it is likely to exploit several other vector species too.
Our results show that there are considerable underlying differences in parasite infection rates across sex and age categories, with female collared flycatchers having an overall greater risk of contracting avian malaria. This sex-difference is almost entirely driven by the most abundant Haemoproteus lineage, hPHSIB1 that is transmitted in Europe. Higher parasite prevalence in females appears to be an unusual trend in birds and highlights how idiosyncrasies in ecology or behaviour may produce contrary patterns. Future studies need to focus on the role of malaria vectors in the transmission process, to fully understand the dynamics and impacts of vector transmitted diseases. We conclude that our results are most compatible with the observed sex-differences in parasite prevalence in collared flycatchers being driven by differences in reproductive behaviours that, in turn, leads to higher female exposure to vectors. Behaviourally driven differences in exposure to vectors are often overlooked but can have strong implications for immunological, conservation or ecological research.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0031182024001021.
Data availability statement
Data is available on request from the authors.
Acknowledgements
We would like to thank numerous field assistants for their help in sample collection; Eryn McFarlane and Jack Shutt for statistical help and discussions and Reija Dufva for assistance in the lab. We would also like to thank 2 anonymous reviewers for their helpful critiques.
Authors’ contributions
WJ and AQ conceived and designed the study. WJ and PNKM did the laboratory work. WJ performed statistical analyses. WJ and AQ wrote the manuscript with input from PNKM.
Financial support
Funding was provided by FORMAS (2018-01563) and Vetenskapsrådet (2016-05138).
Competing interests
None.
Ethical standards
Ethical permissions were provided by the Linköping Animal Ethics Board (5-2-18—7556/14).
Open access
